Gauge Field Theories, Second Edition



Gauge Field Theories, Second Edition Quantum field theory forms the present theoretical framework for our understanding of the fundamental interactions of particle physics. This up-dated and expanded text examines gauge theories and their symmetries with an emphasis on their physical and technical aspects. Beginning with a new chapter giving a systematic introduction to classical field theories and a short discussion of their canonical quantization and the discrete symmetries C, P and T , the book provides a brief exposition of perturbation theory, the renormalization programme and the use of the renormalization group equation. It then explores topics of current research interest including chiral symmetry and its breaking, anomalies, and low energy effective lagrangians and some basics of supersymmetry. A chapter on the basics of the electroweak theory is now included. P ROFESSOR P OKORSKI, a distinguished theoretical physicist, has presented here a self-contained text for graduate courses in physics, the only prerequisite is some grounding in quantum field theory. Born in 1942, Professor Stefan Pokorski received his PhD in theoretical physics in 1967 from Warsaw University, where he is now holder of the Chair in Theoretical Particle Physics. He has been a member of the Polish Academy of Science since 1991 and was Director of the Institute for Theoretical Physics, University of Warsaw, from 1984 to 1994 and President of the Polish Physical Society from 1992 to 1994. A visiting scientist at CERN and the Max-Planck Institute for Physics, Munich, for periods of time totalling almost ten years, Professor Pokorski has also made regular visits to universities and laboratories around the world. Professor Pokorski has had many scientific articles published in leading international journals and in recent years has concentrated on physics beyond the standard model and supersymmetry.

This page intentionally left blank

CAMBRIDGE MONOGRAPHS ON MATHEMATICAL PHYSICS General editors: P. V. Landshoff, D. R. Nelson, D. W. Sciama, S. Weinberg
J. Ambjørn, B. Durhuus and T. Jonsson Quantum Geometry: A Statistical Field Theory Approach A. M. Anile Relativistic Fluids and Magneto-Fluids J. A. de Azc´ rraga and J. M. Izquierdo Lie Groups, Lie Algebras, Cohomology and Some a Applications in Physics† J. Bernstein Kinetic Theory in the Early Universe G. F. Bertsch and R. A. Broglia Oscillations in Finite Quantum Systems N. D. Birrell and P. C. W. Davies Quantum Fields in Curved Space† S. Carlip Quantum Gravity in 2 + 1 Dimensions J. C. Collins Renormalization† M. Creutz Quarks, Gluons and Lattices† P. D. D’Eath Supersymmetric Quantum Cosmology F. de Felice and C. J. S. Clarke Relativity on Curved Manifolds† P. G. O. Freund Introduction to Supersymmetry† J. Fuchs Affine Lie Algebras and Quantum Groups† J. Fuchs and C. Schweigert Symmetries, Lie Algebras and Representations: A Graduate Course for Physicists R. Gambini and J. Pullin Loops, Knots, Gauge Theories and Quantum Gravity† M. G¨ ckeler and T. Sch¨ cker Differential Geometry, Gauge Theories and Gravity† o u C. G´ mez, M. Ruiz Altaba and G. Sierra Quantum Groups in Two-dimensional Physics o M. B. Green, J. H. Schwarz and E. Witten Superstring Theory, volume 1: Introduction† M. B. Green, J. H. Schwarz and E. Witten Superstring Theory, volume 2: Loop Amplitudes, Anomalies and Phenomenology† S. W. Hawking and G. F. R. Ellis The Large-Scale Structure of Space-Time† F. Iachello and A. Aruna The Interacting Boson Model F. Iachello and P. van Isacker The Interacting Boson–Fermion Model C. Itzykson and J.-M. Drouffe Statistical Field Theory, volume 1: From Brownian Motion to Renormalization and Lattice Gauge Theory† C. Itzykson and J.-M. Drouffe Statistical Field Theory, volume 2: Strong Coupling, Monte Carlo Methods, Conformal Field Theory, and Random Systems† J. I. Kapusta Finite-Temperature Field Theory† V. E. Korepin, A. G. Izergin and N. M. Boguliubov The Quantum Inverse Scattering Method and Correlation Functions† M. Le Bellac Thermal Field Theory† N. H. March Liquid Metals: Concepts and Theory I. M. Montvay and G. M¨ nster Quantum Fields on a Lattice† u A. Ozorio de Almeida Hamiltonian Systems: Chaos and Quantization† R. Penrose and W. Rindler Spinors and Space-time, volume 1: Two-Spinor Calculus and Relativistic Fields† R. Penrose and W. Rindler Spinors and Space-time, volume 2: Spinor and Twistor Methods in Space-Time Geometry† S. Pokorski Gauge Field Theories, 2nd edition J. Polchinski String Theory, volume 1: An Introduction to the Bosonic String J. Polchinski String Theory, volume 2: Superstring Theory and Beyond V. N. Popov Functional Integrals and Collective Excitations† R. G. Roberts The Structure of the Proton† J. M. Stewart Advanced General Relativity† A. Vilenkin and E. P. S. Shellard Cosmic Strings and Other Topological Defects† R. S. Ward and R. O. Wells Jr Twistor Geometry and Field Theories†
† Issued as a paperback

This page intentionally left blank

Gauge Field Theories Second Edition
Institute for Theoretical Physics, University of Warsaw

PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING) FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE The Pitt Building, Trumpington Street, Cambridge CB2 IRP 40 West 20th Street, New York, NY 10011-4211, USA 477 Williamstown Road, Port Melbourne, VIC 3207, Australia © Cambridge University Press 1987, 2000 This edition © Cambridge University Press (Virtual Publishing) 2003 First published in printed format 1987 Second edition 2000 A catalogue record for the original printed book is available from the British Library and from the Library of Congress Original ISBN 0 521 47245 8 hardback Original ISBN 0 521 47816 2 paperback (Second edition)

ISBN 0 511 01746 4 virtual (netLibrary Edition)

In memory of Osterns – my mother’s family

This page intentionally left blank


Preface to the First Edition Preface to the Second Edition 0 0.1 0.2 Introduction Gauge invariance Reasons for gauge theories of strong and electroweak interactions QCD Electroweak theory Classical fields, symmetries and their breaking The action, equations of motion, symmetries and conservation laws Equations of motion Global symmetries Space-time transformations Examples Classical field equations Scalar field theory and spontaneous breaking of global symmetries Spinor fields Gauge field theories U (1) gauge symmetry Non-abelian gauge symmetry From classical to quantum fields (canonical quantization) Scalar fields The Feynman propagator Spinor fields Symmetry transformations for quantum fields Discrete symmetries Space reflection Time reversal Charge conjugation Summary and the CPT transformation ¯ CP violation in the neutral K0 –K0 -system Problems

page xvii xviii 1 1 3 3 5 11 12 12 13 16 18 20 20 22 29 29 31 35 36 39 40 45 48 48 53 56 62 64 68

1 1.1






x 2 2.1

Contents Path integral formulation of quantum field theory Path integrals in quantum mechanics Transition matrix elements as path integrals Matrix elements of position operators Vacuum-to-vacuum transitions and the imaginary time formalism General discussion Harmonic oscillator Euclidean Green’s functions Path integral formulation of quantum field theory Green’s functions as path integrals Action quadratic in fields Gaussian integration Introduction to perturbation theory Perturbation theory and the generating functional Wick’s theorem An example: four-point Green’s function in λ 4 Momentum space Path integrals for fermions; Grassmann algebra Anticommuting c-numbers Dirac propagator Generating functionals for Green’s functions and proper vertices; effective potential Classification of Green’s functions and generating functionals Effective action Spontaneous symmetry breaking and effective action Effective potential Green’s functions and the scattering operator Problems Feynman rules for Yang–Mills theories The Faddeev–Popov determinant Gauge invariance and the path integral Faddeev–Popov determinant Examples Non-covariant gauges Feynman rules for QCD Calculation of the Faddeev–Popov determinant Feynman rules Unitarity, ghosts, Becchi–Rouet–Stora transformation Unitarity and ghosts BRS and anti-BRS symmetry Problems Introduction to the theory of renormalization Physical sense of renormalization and its arbitrariness Bare and ‘physical’ quantities Counterterms and the renormalization conditions 71 71 71 75 76 76 78 82 83 83 87 88 90 90 92 93 97 100 100 102 105 105 107 109 111 113 120 124 124 124 126 129 132 133 133 135 140 140 143 147 148 148 148 152






2.7 3 3.1



4 4.1

Contents Arbitrariness of renormalization Final remarks Classification of the divergent diagrams Structure of the UV divergences by momentum power counting Classification of divergent diagrams Necessary counterterms λ 4 : low order renormalization Feynman rules including counterterms Calculation of Fig. 4.8(b) Comments on analytic continuation to n = 4 dimensions Lowest order renormalization Effective field theories Problems Quantum electrodynamics Ward–Takahashi identities General derivation by the functional technique Examples Lowest order QED radiative corrections by the dimensional regularization technique General introduction Vacuum polarization Electron self-energy correction Electron self-energy: IR singularities regularized by photon mass On-shell vertex correction Massless QED Dispersion calculation of O(α) virtual corrections in massless QED, in (4 ∓ ε) dimensions Self-energy calculation Vertex calculation Coulomb scattering and the IR problem Corrections of order α IR problem to all orders in α Problems Renormalization group Renormalization group equation (RGE) Derivation of the RGE Solving the RGE Green’s functions for rescaled momenta RGE in QED Calculation of the renormalization group functions β, γ , γm Fixed points; effective coupling constant Fixed points Effective coupling constant Renormalization scheme and gauge dependence of the RGE parameters

xi 153 156 157 157 159 161 164 164 166 168 170 173 175 177 179 179 181 184 184 185 187 190 191 194 196 197 198 200 200 205 208 209 209 209 212 214 215 216 219 219 222 224



4.4 5 5.1


5.3 5.4


6 6.1

6.2 6.3


3 7.2 7.4 7.4 8.7 8 8.5 7.6 7.1 Scale invariance and operator product expansion Scale invariance Scale transformations Dilatation current Conformal transformations Broken scale invariance General discussion Anomalous breaking of scale invariance Dimensional transmutation Operator product expansion (OPE) Short distance expansion Light-cone expansion The relevance of the light-cone Electron–positron annihilation Deep inelastic hadron leptoproduction Wilson coefficients and moments of the structure function Renormalization group and OPE Renormalization of local composite operators RGE for Wilson coefficients OPE beyond perturbation theory OPE and effective field theories Problems Quantum chromodynamics General introduction Renormalization and BRS invariance.1 7.2 8.3 8.xii Contents Renormalization scheme dependence Effective α in QED Gauge dependence of the β-function Problems 224 226 227 228 230 230 230 233 235 237 237 238 242 243 243 247 249 249 250 254 256 256 259 261 262 269 272 272 272 274 277 279 282 282 284 287 290 290 291 293 293 298 302 7 7. counterterms Asymptotic freedom of QCD The Slavnov–Taylor identities The background field method The structure of the vacuum in non-abelian gauge theories Homotopy classes and topological vacua Physical vacuum -vacuum and the functional integral formalism Perturbative QCD and hard collisions Parton picture Factorization theorem Deep inelastic electron–nucleon scattering in first order QCD (Feynman gauge) Structure functions and Born approximation Deep inelastic quark structure functions in the first order in the strong coupling constant Final result for the quark structure functions 8.5 .

1 .6 10 10.3 12 12.1 11. probabilistic interpretation Light-cone variables.6 8.1 10.2 9.3 9.Contents 8.3 10.2 11.7 Hadron structure functions.4 11 11. spontaneous symmetry breaking Chiral symmetry of the QCD lagrangian Hypothesis of spontaneous chiral symmetry breaking in strong interactions Phenomenological chirally symmetric model of the strong interactions (σ -model) Goldstone bosons as eigenvectors of the mass matrix and poles of Green’s functions in theories with elementary scalars Goldstone bosons as eigenvectors of the mass matrix General proof of Goldstone’s theorem Patterns of spontaneous symmetry breaking Goldstone bosons in QCD Spontaneous and explicit global symmetry breaking Internal symmetries and Ward identities Preliminaries Ward identities from the path integral Comparison with the operator language Ward identities and short-distance singularities of the operator products Renormalization of currents Quark masses and chiral perturbation theory Simple approach Approach based on use of the Ward identity Dashen’s theorems Formulation of Dashen’s theorems Dashen’s conditions and global symmetry broken by weak gauge interactions Electromagnetic π + –π 0 mass difference and spectral function sum rules Electromagnetic π + –π 0 mass difference from Dashen’s formula Spectral function sum rules Results Higgs mechanism in gauge theories Higgs mechanism Spontaneous gauge symmetry breaking by radiative corrections Dynamical breaking of gauge symmetries and vacuum alignment Dynamical breaking of gauge symmetry Examples Problems Standard electroweak theory The lagrangian xiii 304 306 312 314 315 317 317 320 324 327 327 330 333 337 342 342 342 344 347 348 351 353 353 354 356 356 358 362 362 363 366 369 369 373 379 379 382 388 389 391 9 9.4 9.1 9.5 9. light-like gauge Beyond the one-loop approximation Comments on the IR problem in QCD Problems Chiral symmetry.2 10.

3 12.4 Contents Electroweak currents and physical gauge boson fields Fermion masses and mixing Phenomenology of the tree level lagrangian Effective four-fermion interactions Z 0 couplings Beyond tree level Renormalization and counterterms Corrections to gauge boson propagators Fermion self-energies Running α(µ) in the electroweak theory Muon decay in the one-loop approximation Corrections to the Z 0 partial decay widths Effective low energy theory for electroweak processes QED as the effective low energy theory Flavour changing neutral-current processes QCD corrections to CP violation in the neutral kaon system Problems Chiral anomalies Triangle diagram and different renormalization conditions Introduction Calculation of the triangle amplitude Different renormalization constraints for the triangle amplitude Important comments Some physical consequences of the chiral anomalies Chiral invariance in spinor electrodynamics π 0 → 2γ Chiral anomaly for the axial U (1) current in QCD.2 12.1 13.5 12.6 12.2 13.3 13.xiv 12.2 . UA (1) problem Anomaly cancellation in the SU (2) × U (1) electroweak theory Anomaly-free models Anomalies and the path integral Introduction Abelian anomaly Non-abelian anomaly and gauge invariance Consistent and covariant anomaly Anomalies from the path integral in Euclidean space Introduction Abelian anomaly with Dirac fermions Non-abelian anomaly and chiral fermions Problems Effective lagrangians Non-linear realization of the symmetry group Non-linear σ -model Effective lagrangian in the ξa (x) basis Matrix representation for Goldstone boson fields Effective lagrangians and anomalies 394 398 402 403 406 407 407 411 418 419 422 430 435 438 441 445 456 457 457 457 459 464 465 469 469 471 473 475 478 478 478 480 481 484 486 486 488 491 492 495 495 495 500 502 504 12.4 14 14.1 14.7 13 13.

1 15. Wess–Zumino model Supersymmetry breaking Supergraphs and the non-renormalization theorem Appendix A: Spinors and their properties Lorentz transformations and two-dimensional representations of the group S L(2.7 .3 15. n 2 = 0 Convention for the logarithm Spence functions Appendix C: Feynman rules for the Standard Model Propagators of fermions Propagators of the gauge bosons Propagators of the Higgs and Goldstone bosons Propagators of the ghost fields Mixed propagators (only counterterms exist) Gauge interactions of fermions Yukawa interactions of fermions Gauge interactions of the gauge bosons Self-interactions of the Higgs and Goldstone bosons Gauge interactions of the Higgs and Goldstone bosons Gauge interactions of the ghost fields Interactions of ghosts with Higgs and Goldstone bosons xv 505 506 508 509 509 511 513 515 515 519 521 524 531 539 539 546 550 551 552 555 555 556 557 558 559 559 560 560 561 561 562 563 563 564 565 566 567 567 570 571 573 574 578 579 15.Contents Abelian anomaly The Wess–Zumino term Problems 15 15. C) Solutions of the free Weyl and Dirac equations and their properties Parity Time reversal Charge conjugation Appendix B: Feynman rules for QED and QCD and Feynman integrals Feynman rules for the λ 4 theory Feynman rules for QED Feynman rules for QCD Dirac algebra in n dimensions Feynman parameters Feynman integrals in n dimensions Gaussian integrals λ-parameter integrals Feynman integrals in light-like gauge n · A = 0.2 15.4 Introduction to supersymmetry Introduction The supersymmetry algebra Simple consequences of the supersymmetry algebra Superspace and superfields for N = 1 supersymmetry Superspace Superfields Supersymmetric lagrangian.5 15.6 15.

and four-point functions General expressions for the one-loop vector boson self-energies Appendix E: Elements of group theory Definitions Transformation of operators Complex and real representations Traces σ -model References Index 583 583 585 586 591 591 593 593 594 596 599 605 .xvi Contents Appendix D: One-loop Feynman integrals Two-point functions Three.

Marco Roncadelli. Jacek Pawełczyk. Krzysztof Meissner and Michał Spalinski. Stefan Pokorski Warsaw. The book is intended to be a relatively concise reference to some of the field theoretical tools used in contemporary research in the theory of fundamental interactions. Finally my thanks go to Zofia Zi´ łkowska for her contribution to the preparation o of the manuscript. Peter Minkowski. Heinrich u o Leutwyler. for instance at the level of two volumes by Bjorken & Drell (1964. Olivier Piguet. I am also grateful to several of my younger colleagues at the Institute for Theoretical Physics in Warsaw for their stimulating interactions. Wilfried Buchm¨ ller. 1985 xvii . Wojciech Kr´ likowski. and particularly to Paweł Krawczyk for checking a large part of the calculations contained in this book. Marek ¸ Olechowski. It is addressed to graduate students and to young research workers in theoretical physics who have some knowledge of quantum field theory in its canonical formulation. Warsaw University. It is a technical book and not easy reading. Physical problems are discussed only as illustrations of certain theoretical ideas and of computational methods. My warm thanks go to Antonio Bassetto.Preface to the First Edition This book has its origin in a long series of lectures given at the Institute for Theoretical Physics. I am grateful to Wojciech Debski. I am grateful to Wojciech Kr´ likowski. Jacek Prentki. Maurice Jacob and Peter Landshoff o for their interest in this work and strong encouragement. My thanks go to Andrzej Czechowski for his collaboration at the early stage of this project and for numerous useful discussions. Robert Budzynski. 1965). Andrzej Turski. and especially to Peter Landshoff for reading most of the preliminary manuscript. Henri Ruegg and Wojtek Zakrzewski for reading various chapters of this book and for many useful comments. No attempt has been made to review systematically the present status of the theory of fundamental interactions.

C and D are new. I am grateful to Zygmunt Ajduk. Riccardo Guida. Also in Chapter 1 the reader can find a thorough discussion of discrete symmetries C. Jacek Pawełczyk. In particular. I am deeply indebted to Howie Haber for his careful reading of a large part of the new material. in which the modern approach to effective field theories is presented. Carlos Savoy and Kay Wiese for their numerous comments and corrections. Piotr. I would also like to xviii . which was missing in the first edition. including counterterms. His help and collaboration in writing Chapters 1 and 12 and the Appendices was absolutely invaluable. Appendices A. Chapter 12 gives a concise but systematic and self-contained introduction to the electroweak theory.7. The new Appendix A is a substantial extension of the previous Appendix C.Preface to the Second Edition This new edition offers a substantial extension of topics covered by the book. Very special thanks go to Piotr Chankowski. Appendix C contains the complete set of Feynman rules for the Standard Model. Krzysztof Meissner. Several smaller changes and corrections have been made in a number of places in the text. as some intuition based on canonical quantization proves to be very useful even if the main emphasis is on the path integral approach. I hope that the additions leave intact the main feature of the book: it is still not easy reading! I am grateful to many people for their help in the completion of this second edition. Chapter 1 makes the book more self-contained. Thanks. The main additions are Chapter 1 and Chapter 12 and extended Appendices. Marek Olechowski. Ratindranath Akhoury. which is not easily available in the literature.6. An important addition is Section 7. P and T . This is an important completion for a modern book on quantum field theory and fundamental interactions. It gives a systematic introduction to classical field theories and a brief discussion of their canonical quantization. respectively.7 and 15. I also thank Mikolaj Misiak and Janusz Rosiek for their collaboration on Sections 7. to the Second Edition xix thank Peter Landshoff for his constant encouragement and patience during quite a few years before this edition finally materialized. 1999 .edu. Comments and corrections are welcome by e-mail at They will be made accessible at the www page:∼pokorski/ Stefan Pokorski Warsaw.

This page intentionally left blank .

) 1019 years > 1032 years < 6 × 10−22 MeV < 15 × 10−6 MeV From the first three lines of this table we see that the conservation of the electric charge is proved experimentally much more poorly than the conservation of the baryonic charge. nobody is seriously contesting electric charge conservation whereas experiments searching for proton decay belong to the present frontiers of physics. the opposite is not true. A tiny mass for the photon would destroy the gauge invariance of electrodynamics but leave unaffected all its Earth-bound effects. The reason for this lies in the general conviction that the theory of electromagnetism has gauge symmetry† whereas no gauge invariance principle can be invoked to protect baryonic charge conservation. of course. Exact gauge invariance protects the conservation of the electric charge.3 × 1023 years (68% C. It also implies the masslessness of the photon and as seen from the table the present experimental limit on the photon mass is indeed many orders of magnitude better than for the other ‘massless’ particle: the electron neutrino. including the quantum ones. In particular.L. such a theory is also renormalizable (Matthews 1949.0 Introduction 0.1 Gauge invariance It seems appropriate to begin this book by quoting the following experimental information (Review of Particle Properties 1996): electron life-time neutron life-time for the electric charge non-conserving decays (n → p + neutrals) proton life-time photon mass neutrino (νe ) mass > 4. Nevertheless. 1 . † The reader who is not familiar with notions of global symmetry and gauge symmetry is advised to read Chapter 1 first. as long as the electric charge was conserved. It should be stressed at this point that U (1) gauge invariance implies global U (1) invariance but.

Thus one may ask what are the virtues of gauge invariance? We trust electric charge conservation. And non-abelian gauge symmetries are more restrictive and more profound than the U (1) symmetry. Those who remain sceptical should then remember that gauge theories give an economical description of the laws of nature based on well-defined underlying principles which has been phenomenologically successful.2 0 Introduction Boulware 1970. As we know at present. provided that all observers in the universe redefine it in exactly the same way. The standard experimental evidence for gauge theories of weak and strong interactions is briefly summarized in the next section. ∂xµ → exp[ie f (x)/h c] ¯ . because we expect that gauge invariance is behind it. non-abelian gauge bosons carry the group charges and their mass terms in the lagrangian would in general. This attempt proved. nevertheless. This sounds unphysical and we are led to propose that this freedom of convention is present independently at every space-time point or is not present at all as an exact law of nature. The terminology ‘gauge invariance’ can be traced back to Weyl’s studies (Weyl 1919) of invariance under space-time-dependent changes of gauge (scale) in an attempt to unify gravity and electromagnetism. be and are very useful in describing the fundamental interactions. and we doubt baryon charge conservation. Salam & Strathdee 1970) since longitudinally polarized photons decouple from the conserved current. (Approximate global symmetries may. Fock observed (Fock 1926) that one could base quantum electrodynamics (QED) of scalar particles on the operator −ih ¯ ∂ e − Aµ µ ∂x c where Aµ is the electromagnetic four-potential and that the equations were invariant under the transformation Aµ → Aµ + ∂ f (x) . In particular. We do not trust global symmetries as candidates for underlying first principles! A global symmetry gives us a freedom of convention: choice of a reference frame (the phase of the electron wave-function for the U (1) symmetry of electrodynamics). however. destroy not only the gauge symmetry but also the current conservation and therefore the renormalizability of the theory. respectively.) This aesthetical argument may not convince everybody. though this has been much better proved experimentally than electric charge conservation. It can be redefined freely. unsuccessful. unless introduced by spontaneous symmetry breaking. but also for weak and strong interactions successfully described by gauge field theories with non-abelian symmetry groups: SU (2) × U (1) and SU (3). this statement accounts not only for electrodynamics with its U (1) abelian gauge symmetry. We end this section with a short historical ‘footnote’ (Pauli 1933). In 1926.

appeared. If a quark of each flavour has three. however. the known hadrons indeed realize only those q representations of SU (3) which are given by the composite model. Nambu 1966. 2 2 The difficulty can be resolved by postulating a new internal quantum number for quarks which has been called colour (Greenberg 1964. SU (6) is obtained by adjoining the group SU (2) of spin rotation to the internal symmetry group SU (3) for baryons (qqq) and mesons (q¯ ). baryon wave-functions should be antisymmetric in spin and flavour degrees of freedom. Zweig 1964). Fermi statistics is saved by using a totally antisymmetric colour wave-function εabc u a ↑u b ↑u c ↑. 0. Fritzsch & Gell-Mann 1973). Assigning quarks q to the fundamental representation of SU (3). This is not the case in the + original quark model as can be immediately seen from inspection of the ++ ( 3 ) 2 wave-function which must be u↑u↑u↑. not realized by any known hadrons. In 1929. to have difficulties in reconciling the Fermi statistics for quarks with the most natural assumption that in the lowest-lying hadronic states all the relative angular momenta among constituent quarks vanish (s-wave states).2 Reasons for gauge theories 3 which he called gradient transformation.2 Reasons for gauge theories of strong and electroweak interactions We summarize very briefly the standard arguments in favour of gauge theories in elementary particle physics. Both quantum chromodynamics (QCD) and the Glashow–Salam–Weinberg theory are syntheses of our understanding of fundamental interactions progressing over many past years. In particular. and giving them spin one-half one obtains the phenomenologically successful SU (3) and SU (6) schemes. the arrow denotes spin Sz = 1 for each quark. u denotes the quark with electric charge Q = 3 . successful as it was. otherwise indistinguishable. Assuming furthermore that (i) strong interactions are invariant under global SU (3)colour . Bardeen. Weyl studied invariance under this phase change but he kept unchanged his earlier terminology ‘gauge invariance’ (Weyl 1929).0. Similar ideas were also proposed much earlier by Klein (1939) and by Shaw (1955). The latter was postulated as a rationale for the successful SU (3) classification of hadrons (today one should say flavour SU (3)). The quark model for hadrons. QCD QCD emerged as a development of the Gell-Mann–Zweig quark model for hadrons (Gell-Mann 1964. colour states. The concept of gauge transformations was generalized to non-abelian gauge groups by Yang & Mills (1954). Thus. Han & Nambu 1965. London (1927) pointed out the similarity of Fock’s to Weyl’s earlier work: instead of Weyl’s scale change a local phase change was considered by Fock.

The concept of colour certainly underlies what we believe to be the true theory of strong interactions.2) = (including quarks up to b) The experimental value of R is in good agreement with this prediction and in poor agreement with the colourless prediction 11 . 9 Yet another reason for colour is provided by the decay π 0 → 2γ . all flavour-dependent effects can be explained by quark mass differences and the origin of the quark masses. it is assumed that strong interactions act on the colour quantum numbers and only on them. 0. strong arguments. and not qq or q qqqq etc. The concept of colour is also supported by at least two other. transformations (the states may then be classified by their SU (3)colour representation) and (ii) physical hadrons are colourless. Experimentally there is no evidence for any flavour dependence of strong forces.1.4 0 Introduction Fig. exist in nature: the singlet representation appears only in the 3 × 3 × 3 ¯ and 3 × 3 products. we can understand why only qqq and q¯ states. The sum is taken over ¯ all hadronic states in the reaction e+ e− → γ → qq → hadrons. The process e+ e− → γ → hadrons in the parton model. though not satisfactorily understood yet. only colour symmetry can be assumed to be an exact symmetry (flavour .. the theory also has several other basic features which are partly suggested by experimental observations and partly required by theoretical consistency. However.e.1 and the ratio R= is predicted to be R= e2 q 2 e 2 Qq σ (e+ e− → hadrons) σ (e+ e− → µ+ µ− ) (0. In addition. 0. they are singlets under SU (3)colour (quark confinement). Here again the number of quark states matters in explaining the width of π 0 → 2γ . Firstly. The total cross section for this process is then given by the diagram in Fig. is expected to be outside of QCD. This problem will be discussed in more detail in Chapter 13. i.1) = q 2 Qq = 3 × 11 3 4 9 + 1 9 + 4 9 + 1 9 + 1 9 + ··· (0. One is based on the parton model (Feynman 1972) approach to the reaction e+ e− → hadrons. namely QCD.

In recent years there has been a lot of research in QCD perturbation theory neglecting the unsolved confinement problem. Politzer 1973) and presumably.2 Reasons for gauge theories 5 symmetry is evidently broken) and this. Asymptotic freedom means that the forces become negligible at short distances and consequently the interaction between quarks by exchange of non-abelian gauge fields (gluons) is consistent with the successful. as the first approximation. Both are welcome features. Electroweak theory There is. However. Fritzsch. combined with the assumption that it is a gauge symmetry (Han & Nambu 1965. but not with scalars or pseudoscalars. this cannot be fully satisfactory since experimenters collide hadrons and not quarks and gluons. Gell-Mann & Leutwyler 1973). To introduce the Glashow– Salam–Weinberg theory we recall first that the effective Fermi lagrangian for the . A theory with axial-vector gluons based on the SU (3) group cannot be consistently renormalized because of anomalies (Chapter 13). Thus we arrive at vector gluon interaction. has not yet been proved to follow from QCD but it is likely to be true. i. confinement of quarks.e. Given the accuracy of the calculations and of the experimental data one cannot claim yet to have strongly positive experimental verification of perturbative QCD predictions. Coupling of fermions with vector and axial-vector fields. Thus. impressive experimental evidence for the electroweak gauge theory with the gauge symmetry spontaneously broken. experimental results are certainly consistent with QCD (in particular jet physics already provides us with good evidence for the vector nature of gluons) and in view of its elegance and self-consistency there are few sceptical of its chance of being the theory of strong interactions. reflecting strongly singular structure of the non-abelian gauge theory in the IR region. Confinement of the colour quantum numbers. has profound implications: asymptotic freedom (Gross & Wilczek 1973. One can nevertheless argue that it is a justifiable approximation at short distances. is chirally invariant. description of the deep inelastic scattering in the framework of the parton model. of quarks and gluons. confinement of colour is desirable in view of the so far unsuccessful experimental search for free quarks and to avoid a proliferation of unwanted states.0. Once we assume colourful quarks as elementary objects in hadrons. though not proven. at present. An important line of argument in favour of gluons as vector bosons begins with approximate chiral symmetry of strong interactions (Chapter 9). Of course. It has been shown that only non-abelian gauge theories are asymptotically free (Coleman & Gross 1973). at its present stage the theory provides us with calculable corrections to the free-field behaviour of quarks and gluons in the parton model and can be tested in the deep inelastic region.

f L = 1 (1 − γ5 ) 2 νe e− . Both factors . Gasiorowicz (1966) and more recently Abers & Lee (1973) and Taylor (1976)) √ † Leff (x) = −2 2 G µ Jµ (x)J µ (x) (0. 1973).4) it is natural to postulate SU (2) symmetry and consequently the existence of the neutral current corresponding to the third generator of the SU (2) [T + . has conventionally been taken to be (for a systematic account of the weak interactions phenomenology see. However. Neutral-current weak transitions with the expected strength have been discovered at CERN (Hasert et al.6 0 Introduction charged-current weak interactions. The prime superscript indicates the existence of mixing of the quark fields observed in strong interactions (mass eigenstates): d = d cos C + s sin C where the angle C is known as the Cabibbo angle and has been measured in weak decays of strange particles. T − ] = 2T 3 (0. we postulate the existence of a symmetry group. We now extend the Fermi–Cabibbo theory by several additional assumptions. 0. In terms of the lepton and quark fields it can be written as follows Jµ = f f ¯ L γ µT − f L (0. for example. such as.4) with T ± = 1 (σ 1 ± σ 2 ) = T 1 ± iT 2 2 where σ a are Pauli matrices and for weak interactions. Firstly.6) which would induce transitions. those in Fig. Having the current (0. each with V–A structure. νµ µ− . global for the time being. they (i) are not of purely V–A character as expected from the SU (2) model. and (ii) always conserve strangeness to a very good accuracy.3) where the Fermi β-decay constant G µ = 1.165 × 10−5 GeV−2 (h = c = 1) and ¯ the charged current Jµ (x) is composed of several pieces.2 occurring with a similar strength to the charged-current reactions. for instance. valid at low energies. for the weak interactions. u d (0. as known in the 1960s.5) The subscript L stands for the left-handed fermions. According to the existing experimental limits the strangeness non-conserving ¯ neutral-current transitions (like K0 → µ+ µ− or K0 ↔ K0 ) are suppressed by many L orders of magnitude as compared to the standard weak processes.

. 0. Thus.2. Thus. A highly consistent scheme begins to be expected.† has emerged as an important property of weak interactions. Salam & Ward 1964). Augustin et al. Iliopoulos & Maiani (1970) have discovered that the problem of the strangeness non-conserving neutral current is solved if the set of fermionic doublets f L is completed with a fourth one c s . they have predicted the existence of the charm quark discovered later at SLAC (Aubert et al. now further confirmed by the experimental discovery of the ντ τ and t b doublets. Some neutral-current weak transitions. call for further invention in searching for a realistic theory of weak interactions. s = −d sin C + s cos C One can immediately check that with the s orthogonal to d the neutral current is diagonal in flavour. Also the doublet classification of the left-handed fermions with an equal number of lepton and quark doublets. we want the electric charge Q also to be the generator of the symmetry group of our theory. Glashow.0. This has profound implications for a successful extension of the effective model into a nonabelian local gauge field theory: it ensures the cancellation of the chiral anomaly and consequently the renormalizability of the theory. To achieve this in the most economical way we notice that for the left-handed doublets of fermions we can † The t quark was discovered by the CDF and D0 groups in Fermilab in 1994. The next step towards the final form of the Glashow–Salam–Weinberg theory is a ‘unification’ of weak and electromagnetic interactions (Schwinger 1957. 1974). Glashow 1961.2 Reasons for gauge theories 7 Fig. 1974.

The U (1) current reads µ JY = f f f ¯ L γ µY f L + f f ¯ R γ µY f R (0. Therefore the significance of the electroweak unification comes from the choice of a minimal group structure which takes account of both currents. As follows from the formula 1 1 Q = (gT 3 ) + (g Y ) (0. which as we will see.8) where L(R) are left.7).(right-)handed chiral lepton and quark fields. With electromagnetism being described by a gauge field Aµ our minimal model ‘unifying’ electromagnetic and weak interactions requires the α Yang–Mills gauge fields Wµ and Bµ to couple to the SU (2) ×U (1) currents giving the interaction µ α g Jµ W αµ + g JY Bµ (0. this ‘unification’ has a predictive power: it determines the structure of the parity violation in the neutral weak current (see (0. Even with two independent coupling constants.15) below) and gives the W 3 –γ mixing in terms of g and g. Therefore Y commutes with the generators of SU (2) and 2 including the right-handed fermions by the prescription Y = Q (they are singlets with respect to SU (2)) we arrive at the SU (2) × U (1) symmetry group for the electroweak interactions.8 0 Introduction define a new quantum number Y (weak hypercharge) Y = Q − T3 (0.10) Since we are dealing with a direct group product the couplings g and g are independent.7) such that each doublet of the left-handed fermions is an eigenvector of the operator Y (e. reduces the number of free parameters of the theory.11) couple directly to the experimentally observed weak currents.g.9) The SU (2) × U (1) group is the minimal one which contains the electromagnetic and weak currents. the electromagnetic current can be written as follows µ µ Jem = JY + f f ¯ L γ µT 3 f f f L = Qγ µ ¯ f (0. YνL = YeL = − 1 ). According to (0.12) g g . The charged gauge bosons 1 ± 1 2 Wµ = √ (Wµ ± iWµ ) 2 (0.

In the minimal model this is achieved (Weinberg 1967b. Then the boson masses are (at the tree level) MW = 1 vg. b). If initially all vector fields are massless.9) and we have introduced the standard notation in terms of the so-called Weinberg angle (Glashow 1961. we must now introduce masses for intermediate vector boson fields W ± and Z to account for the weakness of the weak interactions as compared to electromagnetism.17) Thus. One knows that the massive Yang–Mills theory.16) (the last relation follows from the fact that the electromagnetic coupling constant is electric charge e. g . 2 so that MW /M Z cos W g2 g2 + g 2 or tan W = g g or e = g sin W = g cos W M Z = 1 v(g 2 + g 2 )1/2 2 =1 (0.13) 1 3 1 W − Bµ g µ g (0. Weinberg 1967b) sin2 W = (0. e > 0). the Fermi constant G µ and the Weinberg angle (parity non-conservation in all neutral-current transitions is consistent with the . v) electroweak theory. For instance.15) µ where Jem is given by (0.14) z couples to the neutral weak current Jµ z Jµ = g cos W 3 (Jµ − sin2 em W Jµ ) (0. knowing the electromagnetic coupling α = e2 /4π .0. Salam 1968) with one complex SU (2) doublet of scalar fields whose neutral component has a non-vanishing vacuum expectation value v.2 Reasons for gauge theories 9 the photon field Aµ (determined by the fact that it couples to the electromagnetic current) must be a combination of the W 3 and B bosons Aµ = The orthogonal combination Zµ = 1 1 + 2 2 g g −1/2 1 1 + 2 2 g g −1/2 1 1 3 Wµ + Bµ g g (0. although less divergent than a theory with arbitrary couplings. apart from the fermion mass matrix and the scalar self-coupling we have a three-parameter (g. is not renormalizable unless the vector boson masses are introduced by means of the Higgs mechanism which breaks the non-abelian gauge invariance spontaneously (’t Hooft 1971a.

at the tree level. However. apparently coupled to each other.15) and with sin2 ≈ 0. .18) Vector bosons with the expected masses were discovered at CERN in 1982! This triumph of the spontaneously broken Yang–Mills theory has since been confirmed by comparing higher accuracy data with higher order calculations. Non-abelian gauge theories are considered at present to be the most promising theoretical framework for fundamental interactions. W MW = M Z cos = πα √ 2G µ 1/2 1 sin W ≈ 80 GeV (0. supergravity) going beyond the so-called standard SU (3) × SU (2) × U (1) model summarized in this section. in this very successful description of the electroweak interactions.23 one predicts. the discovery of the Higgs particle still remains (in 1999) a challenge for future experiments.10 0 Introduction W form (0. They are also extensively studied in the hope of providing the unified description of all interactions (grand unification. We would like to stress the role of gauge symmetry and the unification idea.

symmetries and their breaking Classically. the wave-function (x) = x| (t) (x = (t. It is also reached by quantization of fields describing classical forces.1 Classical fields. Quantization converts classical mechanics into quantum mechanics which describes the behaviour of particles at the quantum level. This logical structure of theories for fundamental interactions is illustrated by the diagram shown below: PARTICLES Classical mechanics (quantization) (wave-function (x) is interpreted as a physical field and quantized) FORCES Classical field theory (quantization) Quantum mechanics QUANTUM FIELD THEORY This chapter is devoted to a brief summary of classical field theory. We then arrive at quantum field theory as the universal physical scheme for fundamental interactions. A state of a particle is described by a vector | (t) in the Hilbert space or by its concrete representation. the ‘classical’ Dirac field which describes the spin 1 particle. As a next step. The forces are described by classical fields. Our ultimate goal is quantum field theory and our classical fields in 11 . The reader should not be surprised by such formulations as. x)) whose modulus squared is interpreted as the density of probability of finding the particle in point x at the time t. The basic physical concept which underlies quantum field theory is the equivalence of particles and forces. the wave-function is interpreted as a physical field and quantized. we distinguish particles and forces which are responsible for interaction between particles. for example. e. such as electromagnetic forces.g. The motion of particles in force fields is subject to the laws of classical mechanics. It is interpreted as a wave function for the 2 Dirac particle.

This ansatz (whose physical sense becomes clearer when classical theory is considered as a limit of quantum theory formulated in terms of path integrals) says: the dynamics of the system evolving from the initial state (t0 . the fields are multiplets under Lorentz transformations and in a space of internal degrees of freedom. equations of motion. (ii) locality of L and its. The action is then the most general functional which satisfies the above constraints. x) is † Later we shall encounter effective lagrangians that may contain terms with more derivatives. In this section we briefly recall the first two results. An ansatz for the action S = dt L = d4 x L can be regarded as a formulation of a theory. In general. x) to the final state (t1 . The equations of motion for a system described by the action S= t1 dt t0 V d3 x L (x). (iv) S has to be real to account for the absence of absorption in classical physics and for conservation of probability in quantum physics.12 1 Classical fields. symmetries and their breaking this chapter are not necessarily only the fields which describe the classical forces observed in Nature. Such terms can usually be interpreted as a perturbation. we always assume implicitly that real physical systems are described by wave packets localized in space. ‡ Although we shall often use plane wave solutions to the equations of motion.1 The action. . From the action we get: equations of motion (invoking Hamilton’s principle). transition from classical to quantum physics (by using path integrals or canonical quantization). conservations laws (from Noether’s theorem).1) (the volume V may be finite or infinite. In classical field theory the lagrangian density L is a function of fields and their derivatives. symmetries and conservation laws Equations of motion All fundamental laws of physics can be understood in terms of a mathematical construct: the action. in the latter case we assume the system to be localized in space‡) are obtained from Hamilton’s principle. bilinearity in the derivatives ∂ν (x) (to get at most second order differential equations of motion). It is our experience so far that in physically relevant theories the action satisfies several general principles such as: (i) Poincar´ invariance e (or general covariance for theories which take gravity into consideration). ∂µ (x) (1.† (iii) invariance under all symmetry transformations which characterize the considered physical system. at most. 1. Path integrals are discussed in detail in Chapter 2.

3) which vanish on the boundary of ≡ ( t. with taken as a Lorentz vector.6) (1. It is important to notice that lagrangian densities which differ from each other by a total derivative of an arbitrary function of fields L = L + ∂µ µ ( ) (1. is are internal quantum number indices).e. the condition δS = 0 gives us the Euler–Lagrange equations of motion for the classical fields: ∂L ∂L =0 µ − ∂ν ∂ i ∂(∂ν iµ ) ν. δS = 0 for arbitrary variations (x) → (x) + δ (x) (1.5) The second term in (1. therefore. . 2. conformal transformations are discussed in . Let us consider a Lie group of continuous global (x-independent) infinitesimal transformations (here we restrict ourselves to unitary transformations.4) can be rewritten as follows: δS = d4 x ∂L ∂L − ∂ν δ ∂ ∂(∂ν ) + ∂ν ∂L δ ∂(∂ν ) (1.8) give the same classical equations of motion (due to the vanishing of variations of fields on the boundary of ). equations of motion. Since in the variation the coordinates x do not change we have: δ(∂ν ) = ∂ν δ and (1. Global symmetries Eq.4) (1. .1.7) (here we keep the indices explicitly. symmetries and conservation 13 such that the action (taken as a functional of the fields and their first derivatives) remains stationary during the evolution. .1 Action. Explicit calculation of the variation δS gives: δS = d4 x ∂L δ ∂ + ∂L δ(∂ν ) ∂(∂ν ) (1. µ = 0. . (1. 3 i = 1.6) is a surface term which vanishes and. 1. V ). i. n (1.6) can also be used to derive conservation laws which follow from a certain class of symmetries of the physical system.2) (summation over all fields and for each field over its Lorentz and ‘internal’ indices is always understood).

9) (for scale transformations see Chapter 7): L( or.) .9) on the fields to distinguish it from the variations δS and δS). ∂µ (x)) = L( (x).12) (x) ≡ L and. Transformations (1.9) the lagrangian density (x). ∂µ ) − L( . Under the change of variables is transformed into L (x).9) where δ0 i (x) = −i a Tia j (1. if (x) describes a motion of a system. of course. (x). Note that a rotation of the reference frame by angle corresponds to active rotation by (− ). symmetries and their breaking Chapter 7) acting in the space of internal degrees of freedom of the fields i .14 1 Classical fields. The T a s are a set of hermitean matrices Tia satisfying the Lie algebra of the group G j T a .3) is again given by (1. ∂µ → Tr(T a T b ) = 1 δ ab 2 (1. the action remains numerically unchanged: δ S=S −S= d4 x L ( . ∂µ (x) (1. Thus.14) Indeed.13) (we use the symbol δ S for the change of the action under transformation (1.10) and i (x) is understood as the ith component of the field (x) in the new reference frame or of the rotated field in the original frame (active). L( (x). (x) is also a solution of the equations of motion for the transformed fields which. Under infinitesimal rotation of the reference frame (passive view) or of the physical system (active view) in that space: i (x) → i (x) = i (x) + δ0 j (x) i (x) (1.9) are symmetry transformations for a physical system if its equations of motion remain form-invariant in the transformed fields. In other words. T b = icabc T c . This is ensured if the density L is invariant under transformations (1. however. ∂µ (x)) (1. and the a s are x-independent. ∂µ ) = 0 (1. ∂µ (x)).11) (x) given by (1. equivalently. For the variations we have δS = δS. ∂µ (x)) = L( (x). variation of the action generated by arbitrary variations of the fields (x) with boundary conditions (1.6) (with → etc.9) remains a solution of the same equations. a solution to the equations of motion after being transformed according to (1. in general are different in form from the original ones.

The hamiltonian of the system reads H= d3 x H( . they satisfy (in an obvious way) the same commutation relation as the matrices T a . these have to be discussed separately). x) = ∂L ∂(∂0 (t. x).10) through Poisson brackets. ∂µ ) = L( . ). y)} = 0 (1. the Lorentz and ‘internal space’ indices of the s and s are hidden. Again.22) . x) ∂ F2 (t. x)(−iT a ) (t. y) {F1 (t. y)} = d3 z − (1.1 Action. x) ∂ F2 (t. such as. x). y)} = { (t.16) are conserved and the charges Q a (t) = a d3 x j0 (t.6) with δ s given by (1. ∂µ ) − L( . Secondly. Moreover. there are interesting exceptions. with = 0. they indeed generate transformations (1.15) is formally given by the integrand in (1. H= ∂0 −L (1. y) ∂ F1 (t.17) are constants of motion.1. F2 (t. ∂µ ) − L( .14) of the lagrangian density under the symmetry transformations: L( . symmetries and conservation 15 and we derive the same equations for (x) as for (x). Let us introduce conjugate momenta for the fields (x): (t.20) ∂ (t.18) (we assume here that = 0.15) and the equations of motion that the currents a jµ (x) = −i ∂L Ta ∂(∂ µ i ) i j j (1. It is very important to notice that the charges defined by (1. It follows from (1.17) (even if they are not conserved) are the generators of the transformation (1.10). (t. the change (1. { (t. equations of motion.17) can be written as Q a (t) = d3 x (t. Firstly. z) ∂ (t. y)} = −δ(x − y) { (t. x) (1. ∂µ ) = 0 (1. z) We get. x)) (1. (t.9).21) The charge (1. z) ∂ (t. provided the currents fall off sufficiently rapidly at the space boundary of . in particular. x) (1. z) ∂ (t. for example. x). (t. x). classical electrodynamics.19) and the Poisson bracket of two functionals F1 and F2 of the fields and is defined as ∂ F1 (t.

Moreover. The well-known examples are.23) Q a (t). i. spatial rotations and Lorentz boosts. we consider transformations (changes of reference frame) which act simultaneously on the coordinates and the fields. 1996)) S = = d4 x L d4 x L (x ).e. for instance. translations. x) i. ∂µ (x) .e.27) ( denotes the image of under the transformation (1. symmetries and their breaking and for its Poisson bracket with the field we get: (1.24) and δ µ is an arbitrary function of the fields) which is a sufficient condition for the new equations of motion to be equivalent to the old ones. ∂µ (x ) = L (x).25) (the Lorentz indices are implicit and α (x ) is understood as the αth component of the field (x ) in the new reference frame). for example.24) and suppose that the fields (x) seen in the first frame transform under some representation T (ε) of the transformation (1. We consider an infinitesimal transformation x µ = x µ + δx µ = x µ + εµ (x) (1.26) (x ) = ∂µ The lagrangian density L after the transformation is defined by the equation (see. generators of symmetry transformations are conserved charges and vice versa. By this we mean that a change of the reference frame has no implications for the motion of the system.16 1 Classical fields. x) = (iT a ) (t. a transformed solution to the original equations remains a solution to the new equations (since (1. there is an arbitrariness in . indeed the transformation (1. ∂µ (x) − ∂µ δ ( (x)) (1. In general det(∂ x /∂ x) = 1 and also L (x). Thus. For infinitesimal transformations we have (∂ denotes differentiation with respect to x ): (x ) = ∂µ (x) + δx µ ∂µ (x) + O(ε 2 ) (x) + δx ν ∂µ ∂ν (x) + O(ε 2 ) (1. Space-time transformations Finally. In this case our formalism has to be slightly generalized.27) implies δS = δS). ∂µ (x ) µ (x). Trautman (1962.24) (which acts on their Lorentz structure) into (x ) in the transformed frame: (x ) = exp[−iT (ε)] (x) (1.10). (t.

28) ( (x)) d4 x (1.30) can be rewritten in the following form: δL = δx µ ∂µ L + Since det ∂ xµ ∂ xν = 1 + ∂µ δx µ (1.1 Action. since δ0 is a functional change. ∂µ (x )) − L( (x).28a). symmetries and conservation 17 the choice of L due to the presence in (1. Symmetry transformations are again defined as transformations which leave equations of motion form-invariant.27) of the total derivative of an arbitrary function δ µ ( (x)). therefore.30) and.31) + ∂L δ0 (∂µ ) + δx µ ∂µ L + O(ε 2 ) ∂(∂µ ) (1. ∂µ (x )) d4 x = L( (x). A sufficient condition is that. ∂µ (x )) = L( (x ). (1. ∂µ (x)) − ∂µ δ µ (x ). δ0 ∂µ = ∂µ δ0 ∂L δ0 ∂(∂µ ) (1. ∂µ (x)) (1. if det(∂ x /∂ x) = 1. ∂µ (x )) (1.26) we get δL = where δ0 (x) = (x) − (x) (1.33) and the equations of motion): 0 = L∂µ δx µ + δx µ ∂µ L + ∂µ = Lδx µ + ∂L δ0 ∂(∂µ ) +δ ∂L δ0 ∂(∂µ ) µ + ∂µ δ µ + O(ε 2 ) (1.35) + O(ε 2 ) . L( (x ).33) we finally get (using (1.34) ∂L ∂L − ∂µ δ0 ∂ ∂(∂µ ) + ∂µ (1.1. equations of motion. for a certain choice of δ . we recover condition (1. The change of the action under symmetry transformations can be calculated in terms of δL = L( ∂L δ0 ∂ (x ).24). The most famous example of symmetry transformations up to a non-vanishing total derivative is supersymmetry (see Chapter 15).28a) Note that. (1.27) satisfies the equation: L( or.14) up to the total derivative.29) where x and x are connected by the transformation (1. the L defined by (1. equivalently. Using (1.32) Eq.

It is easy to check that transformations of the field (x) under translations in time and space are given by {P µ .37) x µ = x µ + εµ implies the conservation law ∂µ where µν µν (x) = 0 (1.18 1 Classical fields. The energy–momentum tensor defined by (1.40) are the total energy of the system (ν = 0) and its momentum vector (ν = 1.39) is the so-called canonical energy–momentum tensor. (x)} = − ∂ (x) ∂ xµ (1. Examples We now consider several examples. It is important to notice that the physical interpretation of the energy–momentum tensor remains unchanged under the redefinition µν (x) → µν (x) + f µν (x) (1.38) is the energy–momentum tensor µν (x) = ∂L ∂ν ∂(∂µ ) − g µν L (1.42) where f µν (x) are arbitrary functions satisfying the conservation law ∂µ f µν (x) = 0 (1. symmetries and their breaking Re-expressing δ0 that ∂µ j µ ≡ ∂µ in terms of δ Lg µ ρ − = (x )− (x) = δ0 +δx µ ∂µ δx ρ + ∂L δ ∂(∂µ ) +δ µ we conclude =0 ∂L ∂ρ ∂(∂µ ) (1.16) to space-time symmetries and both results are the content of the Noether’s theorem which states that symmetries of a physical system imply conservation laws. (x)} εµ .36) This is a generalization of the result (1. note also that det(∂ x /∂ x) = 1). translations Invariance of a physical system under (1. The four constants of motion Pν = d3 x 0ν (x) (1.43) .27) with ≡ 0 satisfies L (x ) = L(x ). 2. 3).39) (we have assumed that the considered system is such that L (x ) defined by (1.41) so that (x) = (x) + {P µ .

This freedom can be used to make the energy–momentum tensor symmetric and gauge-invariant (see Problem 1. Inserting (1.38) and replacing the canonical energy–momentum tensor by a symmetric one and finally using the antisymmetry of ωµν ).1 Action. (x)}.38) and (1. A solution to these constraints is f µν (x) = ∂ρ f µνρ (1.47) where µν is a spin matrix.νρ (x) = x ν for a field with a non-zero spin: M µ. equations of motion.40) remain unchanged.νρ (x) = x ν The conserved charges M νρ (t) = d3 x M 0. x) (1. The total angular momentum three-vector is obtained by constructing the Pauli–Luba´ ski vector n Wµ = 1 εµνρκ M νρ P κ 2 (1.45) (1. 2 The tensor M µν is not translationally invariant. symmetries and conservation 19 with vanishing charges d3 x f 0ν (x) = 0 Then (1.1. we get the following conserved currents: for a scalar field: M µ.47) into (1. The charges M µν are the generators of Lorentz transformations on the fields (x) with {M µν .52) . γ ν ] (for the notation see Appendix A).νρ (t. An infinitesimal Lorentz transformation reads xµ≡ µ ν ν (ω)x ≈ x µ + ωµν x ν µν (1.49) are identified with the total angular momentum tensor of the considered physical system.51) so that (x) = (x) − 1 ωµν {M µν .3).36) (using (1.48) − xρ µν + ∂L ∂(∂µ ) νρ (1.50) µρ µρ − xρ µν (1.44) where f µνρ (x) is antisymmetric in indices (µρ).46) (with ωµν antisymmetric) and it is accompanied by the transformation on the fields (x ) = exp 1 ωµν 2 µν (x) ≈ 1 + 1 ωµν 2 (x) (1.46) and 4 (1. (x)} = −(x µ ∂ ν − x ν ∂ µ + µν ) (x) (1. for Dirac spinors µν = (−i/2)σ µν = 1 [γ µ . For Lorentz scalars = 0.

† Using the formalism of the previous section we derive the equation of motion (since we have two physical degrees of freedom. In units c = h = 1 we have [L] = g4 . 0) to the three-dimensional total angular momentum M k ≡ 1 εi jk M i j : M k = W k /m.54) The lagrangian density which satisfies all the constraints of the previous section reads: 1 1 L = ∂µ ∂ µ − m 2 − λ( )2 − ( )3 + · · · = T − V (1. [m 2 ] = g2 . [ ] = (g/cm)1/2 and ¯ the coefficients [m 2 ] = cm−2 . [λ] = (g · cm)−1 . in order to be able to discuss a theory which is invariant under continuous global symmetry of phase transformations (U (1) symmetry group of one-parameter. 1 . ¯ 1 [m 2 ] = [m 2 ] = g2 and S and λ are dimensionless. the variations in and are independent): ( + m2) +λ + 1 ( m2 1 )2 + ··· = 0 (1. −i χ=√ ( 2 − ) (1.57) † In units c = 1. unitary. the action has the dimension [S] = g · cm = [h ].2 Classical field equations Scalar field theory and spontaneous breaking of global symmetries In this section we illustrate our general considerations with the simplest example: the classical theory of scalar fields (x).55) 2 2 3m 1 where T is the kinetic energy density and V is the potential energy density. [ ] = g.56) In the limit of vanishing couplings (free fields) we get the Klein–Gordon equation. The conserved U (1) current jµ (x) = iq( ∂µ − ∂µ ) (1. It is sometimes also useful to introduce two real fields 1 φ=√ ( 2 + ). symmetries and their breaking (where εµνρκ is the totally antisymmetric tensor defined by ε0123 = −ε0123 = 1) which reduces in the rest frame P µ = (m. 2 1. unimodular transformations): (x) = exp(−iq ) (x) ≈ (1 − iq ) (x) (1.53) Thus. We take these to be complex. Thus [L] = g · cm−3 . The terms which are higher than second powers of the bilinear combination require on dimensional grounds couplings with dimension of negative powers of mass.20 1 Classical fields. the field (x) (its complex conjugate (x)) carries the internal quantum number q (−q).

the U (1) symmetric state is no longer the ground state of this . described by a lagrangian which is symmetric under G) is not invariant under the transformations G. by adding the term ∂ρ (gµν ∂ ρ − gρν ∂ µ ) and using the equation of motion.e.33).1. The minimum of the potential occurs for the classical field configurations such that ∂V ∂V = =0 ∂ ∂ (1.58) 1 λgµν ( 2 are obtained from (1.e. So for an infinite ferromagnet there is an infinite degeneracy of the vacuum.59) For m 2 > 0 the minimum of the potential exists for = 0 = 0 and V ( 0 ) = 0. an infinite array of spin 1 magnetic dipoles.48). i. it can be spontaneously broken. 1. However.16) and (1. The spin–spin interactions between neighbouring 2 dipoles cause them to align. Invariance under Lorentz transformation gives the conserved current (1. for m 2 < 0 (and m 2 > 0. The dependence of the potential on the field is shown in Fig.1. spontaneous symmetry breaking can be realized only in the presence of scalar fields (fundamental or composite). This tensor can be written in other forms. Since we do not want spontaneously to break the Lorentz invariance. for example. the vacuum is filled with scalars (with zero four-momenta) carrying the quantum number (s) of the broken symmetry. The U (1) symmetry of our scalar field theory can be realized in the so-called Wigner mode or in the Goldstone–Nambu mode. Intuitively speaking. The hamiltonian is rotationally invariant but the ground state is not: it is a state in which all the dipoles are aligned in some arbitrary direction. Thus. The phenomenon of spontaneous breaking of global symmetries is well known in condensed matter physics.60) with V ( 0 ) = −(m 4 /2λ).55). study the potential for m 2 < 0!) the solutions 1 1 to these equations are (note that λ must be positive for the potential to be bounded from below) 0 ( ) = exp(i ) −m 2 λ 1/2 1+O m2 λ2 m 2 1 (1.2 Classical field equations 21 and the canonical energy–momentum tensor µν = ∂ µ ∂ν + gµν m 2 + ∂ν ∂µ + − gµν (∂ρ ∂ ρ ) + ··· 2 ) (1. The standard example of a physical theory with spontaneous symmetry breakdown is the Heisenberg ferromagnet. We say that the symmetry G is spontaneously broken if the ground state (usually called the vacuum) of a physical system with symmetry G (i. respectively. Let us return to the question of spontaneous breaking of the U (1) symmetry in the theory defined by the lagrangian (1.

The χ (x) describes the so-called Goldstone boson – a massless mode whose presence ˜ is related to the spontaneous breaking of a continuous global symmetry.17) taken over all space does not exist. as two inequivalent 2 2 representations of the S L(2. φ(x) and χ(x) (see (1. the Poisson brackets (or the commutators) of a charge with a field can be defined provided we integrate over the space after performing the operation at the level of the current (see. Instead.e. 0) and (0. Physical interpretation of this model (and of the negative mass parameter m 2 ) is obtained if we expand the lagrangian (1.22 1 Classical fields.e. as representations ¯˙ ( 1 . We shall return to this subject later (Chapter 9). theory. i. Still. in other words. respectively (see Chapter 11 for more details). positive) and zero.54)). for example. symmetries and their breaking Fig.55) around the true ground state. the integral (1. generically. Spinor fields The basic objects which describe spin 1 particles are two-component anticom2 muting Weyl spinors λα and χ α . C) group of complex two-dimensional matrices M . degenerate set of states (one often calls it a ‘flat direction’) connected to each other by the U (1) rotations. have the mass parameters (−m 2 ) ˜ (i. Bernstein (1974)). 1 ) of the Lorentz group or. the potential has its minimum for an infinite. rewrite it in terms of the fields (we choose = 0 and put m 2 = ∞): 1 ˜ (x) = (x) − 0 . transforming. The degenerate set of vacua (1.1.60) requires massless excitations to transform one vacuum into another one. 1. In this example we can choose any one of these vacua as the ground state of our theory with spontaneously broken U (1) symmetry. respectively. After a short calculation we see that the two dynamical ˜ degrees of freedom. In theories with spontaneously broken symmetries the currents remain conserved but the charges corresponding to the broken generators of the symmetry group are only formal in the sense that.

C) can now be constructed as antisymmetric products ( 1 . for example.64)) and λα = λβ ε β α which ˙˙ transforms as (0. A four-vector is a product of ˙ ˙ ( 1 .16). 0) and (0. Since transition from the representation (1.61) and (1. M = exp −(i/4)ωµν σ µν (1.) 2 It is clear that S L(2.61) (1.62) involves complex conjugation it is natural to denote one of them with a bar. C)-invariant kinetic terms read ¯ ¯¯ L = iλσ µ ∂µ λ = iλσ µ ∂µ λ (1. The corresponding unitary transformations are given by: λα = ε αβ λβ .63) and the two-dimensional matrices σ µν are defined in Appendix A.61) and (1. Its complex 2 ¯˙ ¯ ˙ ˙˙ ¯˙ conjugates are λα ≡ (λα ) (see the text above (1. Thus.1. 0) ⊗ 2 1 ( 2 . (Note. the tensors ε αβ and ¯ ¯˙ ˙˙ ¯˙ ˙ ¯ ˙ ¯˙ ¯ ¯ αβ ˙˙ ε are metric tensors for the spinor representations. The distinction between the undotted and dotted indices is made to stress that the two representations are inequivalent. We use the same symbol for two. 1 ) representation. C) group denoted as λα and χα and transforming through matrices ¯˙ −1 T ∗ (M ) and M are unitarily equivalent to the representations (1. the structure ε αβ λβ φα = λα φα = 2 2 ˙ −λα φ α ≡ λ · φ or χ β ηα εβ α = χα ηα = −χ α ηα ≡ χ · η. Scalars of S L(2. The choice of (1. ε αβ λβ (σ µ )αβ χ β = λα (σ µ )αβ χ β or ˙ ¯ ˙ ¯ 2 2 χα (σ µ )αβ λβ (see Appendix A for the definitions of σ µ and σ µ ). 1 ) representations. ¯˙ ¯ ˙ ¯ We suppose now that a physical system can be described by a Weyl field (or a set of Weyl fields) λα which transforms as ( 1 .64) ˙ where the antisymmetric tensors εαβ and εαβ are defined in Appendix A.61) to (1. Those unfamiliar with two-component spinors and their properties should read Appendix A before starting this subsection. that numerically ε αβ = {εαβ } = iσ 2 . respectively. 0) and (0. 1 ). 0) under the Lorentz group.65) where the two terms are equal up to a total derivative. 1 )⊗(0. Two other possible representations of the S L(2. The factor i has been .62) as fundamental representations is dictated by our convention which identifies (A. for example. The pairwise equivalent representations correspond to lower and upper indices.62).62) χ β (x) ¯ ˙ where x and x are connected by the transformation (1.2 Classical field equations 23 of determinant 1: λα (x ) = Mα β λβ (x) χ α (x ) = M †−1 ¯ ˙ α ˙ ˙ β (1. χ α = χβ ε β α ¯˙ ¯˙ ˙ ˙ ˙ (1.46). They have.32) with the matrix M in (A.and four-dimensional matrices σ µν since it should not lead to any confusion in every concrete context.

24 1 Classical fields. 2 ) (or its unitary equivalent) under the Lorentz transformations. symmetries and their breaking introduced for hermiticity of the lagrangian since one assumes that the spinor components are anticommuting c-numbers. for example. can be rotated away by the appropriate phase transformation. with the equation of motion obtained from the principle of minimal action (variations with respect to ¯ λ and λ are independent): iσ µ ∂µ λ ¯ α ˙ ¯˙ = m λα (1. Then. i. complex α conjugate to R. we remark that a system described by a single Weyl spinor has two degrees of freedom. λiα . one can construct the Lorentz ¯¯ ¯˙¯˙ scalars λλ = λα λα and λλ = λα λα and introduce the so-called Majorana mass term ¯¯ Lm = − 1 m λλ + λλ 2 (1.e. this is what we have to assume for these classical fields when we follow the path integral approach to field quantization (see Chapter 2). Note that in this case the spinors λα (and λα ). Indeed.67) We now consider the case of a system described by a set of Weyl spinors λiα which transform as a complex representation R (see Appendix E for the definition of complex and real representations) of a group of internal symmetry. Remember that ¯ ¯ ¯ λc σ µ ∂µ λc = λc σ µ ∂µ λc (up to a total derivative). the mass term −im (λλc − λλc ). and α the summation over them are always implicit). Such theory is called chiral. so both fields λ and λc enter the . both being ( 1 . There are also interesting physical systems that consist of pairs of independent sets of Weyl fields λα and λc (the internal symmetry indices i. − 1 m 1 (λλ + λλ) − (i/2)m 2 (λλ − λλ). complex conjugate to R. the kinetic lagrangian remains as in (1.e.66) ¯¯ ¯¯ The most general hermitean combination. 2 2 2 1/2 can be recast in the above form with m = (m 1 + m 2 ) by the transformation λ → exp(iϕ)λ with exp(2iϕ) = (m 1 − im 2 )/(m 2 + m 2 )1/2 . Furthermore. 1 i. 0) representations 2 of the Lorentz group but with λc transforming as the representation R .65) but no invariant ¯ ¯ mass term can be constructed since Lorentz invariants λi λ j and λi λ j are not ¯˙ ¯˙ invariant under the symmetry group. The operation of charge conjugation will be discussed in more detail in Section 1. if the field λ carries no internal charges (or transforms as a real representation of an internal symmetry group). Such theory is called vector-like. form the representation R .5. The term Lm does not 1 2 vanish since λα s are anticommuting c-numbers. The mass term can then be introduced and the full lagrangian of such a system then reads: ¯¯ ¯ ¯¯ L = iλσ µ ∂µ λ + iλc σ µ ∂µ λc − m λλc + λλc (1.68) ¯¯ Similarly to the case of Majorana mass. the spinors transforming as (0. We call such spinors charge conjugate to each other. Finally. if present.

0) representations of S L(2. The mass term. α .e.2 Classical field equations 25 lagrangian in a fully symmetric way. 2 ) representations of S L(2. but transforms as a complex representation R under the full SU (2) × U (1) group. Each 2 pair of states described by λiα is connected by the C P transformation (see later) and can therefore be termed particle and antiparticle. transform as 3 and 3 of SU (3). e− . 2 In the electroweak theory (see Chapter 12) all the left-handed physical fields are included in the set of λiα s which transforms then as a reducible. in this case the set λiα includes pairs of left-handed particles with opposite electric charges such as. and are related by charge α conjugation.69) and similarly for λc . of course.68) are: iσ µ ∂µ λ ¯ α ˙ ¯α = m λc ˙ (1. for example. The equations of motion which follow from (1. in the Dirac sea interpretation. λiα s are called left-handed chiral fields. describe massless helicity + 1 states transforming as R . In both cases. In this formulation positive (negative) energy solutions for λi α s describe all + 1 (− 1 ) helicity massless states of the theory (formerly described as 2 2 negative (positive) energy solutions for λiα s).1. Negative energy classical solutions for λiα s. These pairs consist of the states with opposite helicities. the mass term mixes positive energy solutions for λ with ¯α negative ones for λc (which are represented by positive energy solutions for λc ) and ˙ vice versa. i. C) and as R under the internal symmetry ¯ ˙ group can be given. Thus. the left-handed i λα s and λci s. In accord with the discussion of Appendix A. Positive energy solutions for λiα s are identified with − 1 helicity states 2 of massless quarks and positive energy solutions for λci s describe − 1 helicity α 2 states of massless antiquarks.) ¯ ˙ An equivalent description of the same theory in terms of the fields λi α transform1 ing as (0. therefore. is now invariant under the internal symmetry group. the electroweak theory is chiral and quantum chromodynamics is vector-like. C). respectively. the so-called Dirac mass. (Whether within a given reducible representation contained in R the − 1 or + 1 helicity states are called particles is. As we shall discuss in more detail later. it is convenient to take as the fundamental fields of the theory Weyl spinors which transform as ( 1 . e+ and quarks with electric charges ∓ 1 . In the second case (QCD) the fundamental fields of the theory. We recall that (as discussed in Appendix A) 2 positive energy classical solutions for spinors λiα s describe massless fermion states with helicity − 1 and. this set forms a real representation of the electromagnetic 3 3 U (1) gauge group. Negative energy solutions for λiα s and λci s describe. a 2 2 matter of convention. To be more specific. ± 2 . complex representation R of the symmetry group G = SU (2) × U (1) (R = R ). this can be interpreted as a mixing between different helicity states of the same massless particle.

70) transforming as R and R . They are called Dirac matrices in the chiral representation and are given in Appendix A. respectively.68).26 1 Classical fields. −im ¯ γ5 . respectively. Also in Appendix A we derive the Lorentz transformations for the Dirac spinors. mass terms are possible. Since the representation R is 2 real. λα ) to recover the mass term of (1. both transforming as ( 1 . By analogy with the nomenclature introduced for λα and χ α the fields L and R are called left¯˙ and right-handed chiral fields. PR PL L (1. under the group of internal symmetries. Four-component notation can also be introduced and. 0) and (0.71) we define Writing the matrices γ µ .68) and (1. ¯ = † γ 0 and comparing the kinetic terms of (1. The invariants λα λα and λα λα can T T be written as L C L and ¯ L C ¯ L . can be rotated away by the appropriate phase transformation of the spinors λ and λc . in fact. R. Possible Lorentz invariants are of the ¯˙ ¯˙ ¯˙¯˙ form ¯ R L ≡ χ α λα and ¯ L R ≡ λα χ α .71) ¯˙ where we have introduced ¯ ≡ (λcα . Another possible hermitean mass term. like QCD. with the matrix C defined in (A. 0) of the Lorentz group but as R and R 2 representations of an internal symmetry group. We define four-component chiral fermion fields as follows L = λα 0 R = 0 χα ¯˙  =0    =0  (1.72) They satisfy the equations L = R 1 − γ5 2 1 + γ5 = 2 L ≡ PL ≡ PR L.73) R R which define the matrix γ5 in the chiral representation. For a vector-like theory. 1 ) under the Lorentz ¯˙ 2 2 group. with charge conjugate pairs of Weyl spinors λ and λc . symmetries and their breaking respectively. In particular. is quite convenient for Weyl spinors λα and χ α transforming as ( 1 .104). antiquarks and quarks with helicity + 1 . it is convenient to use the Dirac ¯ four-component spinors (in the presence of the mass term in (1.68) can be rewritten as L = i ¯ γ µ ∂µ −m ¯ (1. The lagrangian (1. for a set of Weyl spinors λα and their complex conjugate .68) λ and λc can be interpreted as describing opposite helicity (chirality) states of the same massless particle which mix with each other due to the mass term) = λα ¯ cα λ ˙ or c = λc α ¯˙ λα (1. respectively.

In the latter case it is supplemented by the conditions (1.65) and (1. cj j i ci Lorentz invariants L R and R L (or a combination of them) are generically not invariant under the internal symmetry group but such invariants can be constructed by coupling them.66) reads LM = i ¯ M γ µ ∂µ 2 M − m ¯M 2 M (1. for objects represented by a Weyl field λα with no internal charges or transforming as a real representation of an internal symmetry group we can introduce the so-called Majorana spinor M = λα ¯˙ λα (1. Thus.77) The equations of motion for free four-dimensional spinors have the form iγ µ ∂µ − m =0 (1. The R is then the chiral right-handed positron (electron) field.76) In this notation the lagrangian given by (1.2 Classical field equations 27 ¯˙ λα .74) also transform as R and R .1.73). † The notation in (1.71) but no mass term is allowed. that the superscript ‘c’ ci ¯˙ in the spinors R is now used even though λα and λα are not charge conjugate to each other (since they are in different Lorentz group representations). γ ν } = 2g µν . to scalar fields transforming properly under this group (see later). respectively. transforming as R and R . under the internal symmetry group. Note. in ordinary QED if L has a charge −1 (+1) under the U (1) symmetry group. For a vector-like spectrum we have ci ci ) . For instance. γ 5.78) for the Dirac spinors and with m = 0 for the chiral spinors. unlike the spinors in (1. R R =( .70).74) are defined for both chiral and vector-like spectra of the fundamental fields.70) the spinors in (1. The kinetic part of the lagrangian for the chiral fields is the same as in (1.74) has been introduced by analogy with (1. The solutions to the equations of motion for the Weyl spinors and for the four-component spinors in the chiral and Dirac representations for the γ µ matrices are given in Appendix A.75) Finally. for example. the sets† i L = λiα 0 ci R = 0 ¯ iα λ˙ (1. We note that γ 5 = γ5 = iγ 0 γ 1 γ 2 γ 3 and that the following algebra is satisfied: {γ µ . however. γ µ = 0 (1. respectively. we can identify it with the c left-handed chiral electron (positron) field.

γν ]) gives the conserved angular momentum tensor M µν = d3 x M 0µν (1. † This form of the lagrangian follows after eliminating auxiliary fields via their (algebraic) equations of motion.84) Similar terms can be written down for several spinor fields (λ. ¯¯ λλφ (1. symmetries and their breaking A few more remarks on the free Dirac spinors should be added here. χ.83) The lagrangian (1.85) ¯¯ The absence of otherwise perfectly allowed coupling −g (φ λλ + φ λλ) is due ¯¯ to supersymmetry. the so-called Wess–Zumino model.µν = i ¯ γ κ x µ (1. The angular momentum density i ∂ ∂ M κ. supersymmetry (see Chapter 15). These may be internal symmetries or. see Chapter 15. The simplest supersymmetric theory. Additional symmetries. For instance. for a neutral (with respect to the internal quantum numbers) Weyl spinor λ interacting with a complex scalar field φ (also neutral) the dimension four ([m]4 in units h = c = 1) Yukawa interaction terms are: ¯ λλφ. would further constrain the allowed terms. . .) interacting with scalar fields.65) can be extended to include various interaction terms. describing interactions of one complex scalar field φ(x) with one chiral fermion λ(x) has the following lagrangian† ¯¯ ¯¯ LWZ = iλσ · ∂λ − 1 m(λλ + λλ) + ∂µ φ ∂ µ φ − m 2 φ φ 2 ¯¯ − g(φλλ + φ λλ) − gmφ φ(φ + φ) − g 2 (φ φ)2 (1. if assumed. In the four-component notation we get (φλλ + φ λλ) = ¯ M PL M φ + ¯ M PR M φ . for instance. The momentum conjugate to is α = ∂L =i ∂(∂0 α ) † † α (1.28 1 Classical fields. ¯¯ λλφ . .80) and the energy–momentum tensor density reads µν = i ¯ γµ (1.81) ((1. where M is given by (1. .76).79) For the hamiltonian density we get H= ∂0 −L= γ γ 0 (−iγ · ∂ + m) ∂ ∂ xν (1.81) is obtained after using the equations of motion). λλφ .82) − xν − σ µν ∂ xν ∂ xµ 2 (where four-dimensional σµν = (i/2)[γµ .

3) take the form: j ¯ ¯ λi σ µ (T a )i λ j Aa = µ a j i µ L γ (T )i Lj Aa µ with L given by (1. It is straightforward to include the interaction terms in the equation of motion.71). The terms in (1.3 Gauge field theories 29 In general. we consider gauge invariance in electrodynamics. the interactions with vector fields Aa in the adjoint µ representation of the group (as for the gauge fields. 1 ¯ M Aµ Aµ .70). see Appendix E) can be supplemented by terms like: j ¯ ¯ ¯ ¯ λi σ µ (T a )i λ j Aa + λic σ µ (−T a∗ )i j λc j Aa µ µ ¯ µ ¯ ¯ = λi σ µ (T a )i λ j Aa + λc j σ µ (T a ) j i λcj Aa µ j = ¯ i γ µ (T a )i j j Aa µ In the last form (in which (x) is given by (1. (1. 1 ¯ M2 ¯ 1. For a vector-like theory. see Section 1. for example. the lagrangian (1.68) in which λci transforms as a representation R ∗ of the group (with generators −T a∗ . with a scalar field (x) or a vector field Aµ (x). 1 ¯ M . For the abelian field Aµ this is the same as the minimal coupling of the electron to the electromagnetic field ( pµ → pµ + e Aµ ).70)) this interaction can be added to the lagrangian (1. for example.86) etc.87) The first two terms are Yukawa couplings (scalar and pseudoscalar).72) and the internal symmetry indices are properly contracted.72).3 Gauge field theories U (1) gauge symmetry As a well-known example.87) require dimensionful coupling constants. A Lorentz-invariant and hermitean interaction part of the lagrangian density may consist. Other interaction terms of a Dirac field (1. is called the Fermi interaction. (1. describing the self-interactions of a Dirac field. with the mass scale denoted by M. . The last term. It is often introduced as follows: consider a free-field theory of n Dirac particles with . 1 ¯ γµ A µ M . in the four-component notation the Yukawa terms have the generic j j i i structure ( ¯ R L ± ¯ L R ) k ( k ) where the chiral fields are defined by (1. are possible as well. If λi s transform as a representation R (with hermitean generators T a ) of an internal symmetry group.1. of the operators: ¯ ¯ γ5 ¯ γ 5 γµ Aµ .

the U (1) symmetry of the lagrangian (1.92) is ensured if the gauge field Aµ (x) transforms as: 1 Aµ (x) = Aµ (x) + ∂µ (x) e Covariant derivatives play an important role in gauge theories.1). By Noether’s theorem (see Section 1. . To make the lagrangian invariant one must introduce a new term which can compensate for the extra terms.88).17).89) where the parameter qi is an eigenvalue of the generator Q of U (1) and numbers the representation to which the field i belongs. note that e < 0 in this reference). The lagrangian (1.93) where e is an arbitrary positive constant.91) It is straightforward to verify that lagrangian (1. The derivative Dµ is called a covariant derivative. (1. Equivalently.94) In particular † The sign convention in the covariant derivative is consistent with the standard minimal coupling in the classical limit (Bjorken and Drell (1964).30 1 Classical fields.88) is invariant under that group of transformations.88) is not invariant under gauge transformations because transformation of the derivatives of fields gives extra terms proportional to ∂µ (x).88) Then define a U (1) group of transformations on the fields by i (x) = exp(−iqi ) i (x) (1. one should find a modified derivative Dµ i (x) which transforms like i (x) [Dµ i (x)] = exp[−iqi (x)]Dµ i (x) (1. The transformation rule (1.90) and therefore conservation of the corresponding charge (1. The covariant derivative is constructed by introducing a vector (gauge) field Aµ (x) and defining† Dµ i (x) = [∂µ + iqi e Aµ (x)] i (x) (1. symmetries and their breaking the lagrangian density n L= i=1 ¯ i iγ µ ∂µ i −m ¯i i (1.88) implies the existence of the conserved current jµ (x) = i qi ¯ i γµ i (1.92) and replace ∂µ by Dµ in the lagrangian (1. We now consider gauge transformations (local phase transformations in which is allowed to vary with x) i (x) = exp[−iqi (x)] i (x) (1.

It also implies absence of the gauge field mass term m 2 Aµ Aµ . For the antisymmetric product of two derivatives [Dµ .98) where the T α are the hermitean matrix representations of the generators of the group. The field strength tensor Fµν can be used to complete the lagrangian with the gauge-invariant kinetic energy term for the gauge field itself (assume one fermion field for definiteness) / L = ¯ (iD − m) 2 − 1 Fµν 4 (1. Dν ] i i = Dµ (Dν i) − Dν (Dµ i) = iqi e[∂µ Aν (x) − ∂ν Aµ (x)] i (1. .17).1. appropriate for the fields and satisfying relations (1.3 Gauge field theories 31 one may construct new covariant objects by repeated application of covariant derivatives.95) By comparing the gauge transformation properties of both sides in (1.96) is gauge-invariant.95) we conclude that Fµν (x) = ∂µ Aν (x) − ∂ν Aµ (x) (1. But such a term does not break the global U (1) symmetry. Exactly analogous considerations apply to scalar field theories with U (1) gauge symmetry and to theories with Weyl fermions. We start with the free-field lagrangian for Dirac fields which transform according to a representation of some non-abelian Lie group (x) = exp(−i α / = eq A T α ) (x) (1. Dν ] one gets [Dµ . U (1) gauge invariance implies global U (1) invariance and the conservation of current (1. Non-abelian gauge symmetry To construct a non-abelian gauge field lagrangian we repeat the same steps.90) and the corresponding charge (1.11). Of course.97) The Euler–Lagrange equations of motion are now ∂µ F µν = eq ¯ γ ν (i/ − m) ∂ or / (iD − m) =0 and identifying e (e > 0) with electric charge we recognize in our theory the Maxwell–Dirac electrodynamics.

at the ¯ ¯ quantum level the normalization of the field is fixed by the normalization of the single particle state.105) form a group.101) (1. With the form (1.105) It can be checked that transformations (1. g is a dimensionful parameter and therefore can always be scaled to one (similar arguments apply to (1.93)).100) the condition (1. (x)] where (x) = +i α (1. † The sign convention for the non-abelian case is consistent with the convention (1.93) we expect it to be given by a combination of the normal derivative and a transformation on the fields Dµ (x) = [∂µ + Aµ (x)] (x) where Aµ is a (antihermitean) element of the Lie algebra† Aµ (x) = +ig Aα (x)T α µ (1.99) By analogy with (1.104) is an infinitesimal gauge parameter in the matrix notation. Let us now consider the extension of the group G to a group of local gauge transformations. Equivalently.102) (1. symmetries and their breaking The free-field lagrangian is invariant under global group transformations (1.105) that under global transformations (∂µ = 0) the gauge fields transform according to the adjoint representation of the group with (T α )βγ = −icαβγ . this is convenient for embedding electromagnetism into the electroweak theory. At the quantum level g is relevant since quantum theory contains the Planck constant h and g is dimensionless in units h = 1. . It is also seen from (1.103) (x)T α (1. Generalizing the U (1) case we seek a covariant derivative such that [Dµ (x)] = exp[−i α (x)T α ]Dµ (x) = U (x)Dµ (x) (1.32 1 Classical fields.99) is satisfied if the gauge transformation rule for Aµ (x) is as follows: Aµ (x) = U (x)Aµ (x)U −1 (x) − [∂µU (x)]U −1 (x) = U (x)[Aµ (x) + ∂µ ]U −1 (x) or for an infinitesimal transformation δ Aµ (x) = Aµ (x) − Aµ (x) = ∂µ (x) + [Aµ (x).98).100) Thus we need gauge fields in the number given by the number of generators of the group.93). The constant g is arbitrary.103) and (1. In classical physics. The corresponding transformation for the gauge fields Aα (x) reads µ δ Aα (x) = (1/g)∂µ µ α (x) + cαβγ β (x)Aγ (x) µ (1.

we have [Dµ .1. µν Finally. By applying the covariant derivative Dρ = ∂ρ + Aρ (x) to the object ξ and insisting on the Leibnitz rule Dρ (ξ ) = (Dρ ξ ) we get Dµ ξ = ∂µ ξ + [Aµ .109) (1.113) (1.106) is a covariant quantity transforming under gauge transformations as follows: (1. ξ ] Then the gauge transformation (1.107) that the tensor G µν can be decomposed in terms of group generators G µν = igG α T α µν where G α = ∂µ Aα − ∂ν Aα − gcαβγ Aβ Aγ µν ν µ µ ν (1.113a) (1.112) + ξ(Dρ ) (1.114) (Tr[G µν G µν ] is gauge-invariant because Tr[U G µν G µν U −1 ] = Tr[G µν G µν ]). Aν (x)]) (x) Thus. Dν ] (x) = (∂µ Aν (x) − ∂ν Aµ (x) + [Aµ (x). It reads L=+ 1 / Tr[G µν G µν ] + ¯ (iD − m) 2g 2 (1. for instance.3 Gauge field theories 33 We similarly can construct new covariant quantities by repeated application of the covariant derivative.108) It is obvious from (1. . Dν ]G ρσ = [G µν . the antisymmetric tensor (the field strength) G µν (x) = ∂µ Aν (x) − ∂ν Aµ (x) + [Aµ (x).110) We are ready to write down the gauge-invariant lagrangian for a non-abelian gauge field theory with fermions. G ρσ ] (1.108) and (1. We get.111) The transformation rule for G α (x) follows from (1. Aν (x)] G µν (x) = U (x)G µν (x)U −1 (x) For an infinitesimal transformation we get δG µν (x) = [G µν (x). we can generalize the definition of the covariant derivative to apply it to any Lie algebra element ξ(x) = +iξ α (x)T α .103) can be written simply as δ Aµ (x) = Dµ (x) Moreover.109). [Dµ .107) (1. (x)] (1.

The term G µ µ is zero and the possibility of a parity non-conserving term εµνρδ G µν G ρδ in the lagrangian will be discussed later. One requires the action to be stationary with respect to small variations of the gauge fields Aµ → Aµ + δ Aµ and its derivative δ(∂ν Aµ ) = ∂ν (δ Aµ ) where we formally treat δ Aµ as a covariant quantity transforming in an adjoint representation of the group δ Aµ (x) = U (x)δ Aµ (x)U −1 (x) (1. symmetries and their breaking Terms of higher order in fields are not allowed for quantum field theory if it is to be renormalizable (Chapter 4).116) can be expressed as follows: G µν → G µν + Dµ (δ Aν ) − Dν (δ Aµ ) (1.117) (1.120) and the invariance of Tr[G µν δ Aν ] under gauge transformations (therefore.112) of a covariant derivative acting on a covariant object δ Aµ .115) (we recall our normalization Tr[T α T β ] = 1 δ αβ ). The pure gauge field term must be present in the lagrangian because we need the kinetic energy term to be quadratic in the gauge fields. of (1.120) we may replace the covariant derivative by the ordinary .107) under the variation (1. Its gauge-invariant form then implies the presence of two gauge self-interaction terms of the order g and g 2 .118) Here we have used definition (1. respectively 1 Tr[G µν G µν ] = − 1 (∂µ Aα − ∂ν Aα )2 + gcαβγ Aβ Aγ ∂ µ Aνα ν µ µ ν 4 2g 2 − 1 g 2 cαβγ cασ δ Aβ Aγ Aµσ Aνδ µ ν 4 (1.116) The change of the field strength (1. Using the relation Dµ Tr[(G µν δ Aν )] = Tr[(Dµ G µν )δ Aν ] + Tr[G µν (Dµ δ Aν )] (1. The variation of the gauge field action is δ d4 x 1 2 Tr[G µν G µν ] = 2 2 2g g d4 x Tr[G µν (Dµ δ Aν )] (1. 2 We end this section with a derivation of the Euler–Lagrange equations for gauge fields. on the l.s.34 1 Classical fields.h.119) (antisymmetry of G µν in µ and ν has been used).

Itzykson & Zuber (1980)).122) reads α jµ (x) = i jµ (x)T α with α jµ = ¯ γµ T α (1. procedure (for more details see Bjorken & Drell (1964).121) The remaining term δ can be written as δ / d4 x ¯ (iD − m) =2 d4 x Tr[ jµ δ Aµ ] (1.125) that the non-abelian charge associated with the fermionic current is not a constant of motion. in this section we briefly recall the canonical quantization. However.120) and neglecting the surface terms at infinity on the l.4 From classical to quantum fields 35 derivative.123) Dµ j µ (x) = 0 (1.121) and (1. that δ d4 x 1 2 Tr[G µν G µν ] = − 2 2 2g g d4 x Tr[(Dµ G µν )δ Aν ] (1. this remark also helps to check relation (1.s.h.124) Applying the covariant derivative to (1.1.124) one derives the covariant divergence equation for the current (1. 1. We also see from (1. for the sake of completeness and easy reference. by integrating (1.120)) we conclude.122) / d4 x ¯ (iD − m) where the current jµ defined by the variation (1. .122) we get the equation of motion (1/g 2 )Dµ G µν − j ν = 0 (1.4 From classical to quantum fields (canonical quantization) In this book we use the path integral approach to the quantization of fields (see Chapter 2). since gauge fields are not neutral their contribution must be included to find conserved charges.125) Thus.123) Combining (1. gauge fields can only couple consistently to currents which are covariantly conserved..

replacing the Poisson brackets in (1. } → −i[ .130) (1. y)] =0 (1. j) are internal symmetry indices). for example. the fields and their conjugate momenta = ∂0 are promoted to hermitean operators in the Heisenberg picture (in a Hilbert space of state vectors) by postulating for them canonical equal-time commutation relations. see also Section 2. the a(k) and a † (k) are complex conjugate to each other.7). y)] = −iδi j δ(x − y) =[ i (t.40) can be rewritten in terms of the operators a(k) and a † (k): Pµ = d3 x 0µ (x) = 1 2 d3 k k µ a † (k)a(k) + a(k)a † (k) (2π )3 2k0 (1. In the canonical quantization.128) (the indices (i.21) by commutators. For most physical questions (apart from the problem of the cosmological . ]: [ [ i (t. we see that the vacuum energy is infinite.127) d3 k a(k) exp(−ik · x) + a † (k) exp(ik · x) (2π )3 2k0 (1. for a real field (x). The canonical quantization of interacting fields can be performed along similar lines in the so-called interaction picture (we refer the reader to the standard textbooks. An arbitrary solution to the free Klein–Gordon equation (1.36 1 Classical fields. k0 = E = +(k2 + m 2 )1/2 and.56) may be expanded as a Fourier integral over plane-wave solutions: (x) = where d3 k = (2π )3 2k0 d4 k 2π δ(k 2 − m 2 )θ(k0 ) (2π )4 (1.e. j (t. j (t. x). symmetries and their breaking Scalar fields We begin with a scalar field theory and consider only free fields.126) is a Lorentz-invariant measure. Bjorken & Drell (1965). x). i (t. y)] j (t. a † (k ) = (2π )3 2k0 δ(k − k ) [a(k). a † (k ) = 0 The ground state (the vacuum) is defined by the relations: a(k)|0 = 0. a(k )] = a † (k). { .129) The energy–momentum operator (1. x).128) that the (hermitean conjugate to each other) operators a(k) and a † (k) satisfy the following algebra: a(k). It follows from the postulated relations (1.131) Since a(k)a † (k) = a † (k)a(k) + (2π )3 2k0 δ(0). i. 0|0 = 1 (1.

the a † (k) and a(k) are the creation and annihilation operators. (y)] = 0 (1.135) (1. Therefore. whereas in quantum field theory (x) |0 is a state |x of a particle localized at the space-time point x. consequently.129) and (1. interpreted as a quantum field. we get k| (x)|0 = 0|a(k) (x)|0 = exp(ik · x) (1. It follows from (1. . Hence. for example. with its momentum representation k| (x)|0 = x |k † . a † (k) = k µ a † (k) and. the field (x) (its negative frequency part) acting on the vacuum creates a particle at point x. the positive frequency part of the field (x) annihilates that state into the vacuum and the negative frequency part creates two-particle state. For a product of the fields (x) = (+) (x) + (−) (x) we get.126). P µ a † (k)|0 = k µ a † (k)|0 (1. with continuous momentum spectrum and propagating as a plane wave (with positive energy).† When acting on a particle at the point x. where (+) ( (−) ) are positive (negative) frequency parts of (x) given by (1. (x) = x | (t) . Its only effect is to subtract from the original expressions the infinite vacuum energy. : : = (−) (−) + 2 (−) (+) + (+) (+) . Hence.4 From classical to quantum fields 37 constant) the vacuum energy is unobservable. The field–particle duality now becomes evident.134) Thus. respectively.137) † Note that in quantum mechanics.132) that P µ . the wave-function (x) describes a state in the configuration representation. the state |k ≡ a † (k)|0 is an energy–momentum eigenstate.136) i.1. for a quantum of four-momentum k µ . it is its negative frequency part which creates a particle with positive energy. in quantum mechanics the positive frequency part of (x) describes propagation with positive energy but.e.132) + a(k)a † (k)) → : 1 (a † (k)a(k) + a(k)a † (k)): ≡ a † (k)a(k) (1. Simultaneous measurement of the field strength at points x and y is possible only if [ (x).133) 2 is called ‘normal (or Wick’s) ordering’ and is denoted by a double-dot symbol. we redefine the hamiltonian as P 0 → P 0 − 0|(a † (k)a(k) + a(k)a † (k))|0 = The prescription 1 † (a (k)a(k) 2 d3 k k 0 a † (k)a(k) (2π )3 2k0 (1. Also.

y) = −iδi j δ(x − y) (1. operator-valued functions. † i = ∂0 i (1. Fourier integrals. j (t. analogous to (1. x).138) is a solution of a free Klein–Gordon equation x + m2 (x − y) = 0 (x − y) = − (y − x) Since that (1.e. They can easily be expressed in terms of the hermitean annihilation and † † creations operators a1 (k). i. It follows from the postulated commutation relations (1.140) Thus the field strength is simultaneously measurable in two points with a spacelike interval. from its Lorentz invariance we conclude (x − y) = 0 for all (x − y)2 < 0 (1. the procedure is exactly the same if we work with the two real fields φ and χ defined in (1. symmetries and their breaking By direct calculation we get [ (x). a2 (k) for the quanta of the real fields .139) (y − x) vanishes for t1 − t2 = 0.126).142) and the other commutators vanish. In terms of the complex field and its conjugate momenta i = ∂0 † i. Quantization of a complex scalar field (x) proceeds in quite an analogous way. consistently with the notion of locality and causality (two points with space-like intervals cannot be connected in a causal way). x).128) that the quanta of a scalar field obey Bose–Einstein statistics.141) we postulate i (t. a2 (k) and a1 (k). now read:  d3 k  †  a(k) exp(−ik · x) + b (k) exp(ik · x)  (x) =  (2π )3 2k0 (1. † j (t.54). (y)] = where ε(k0 ) = One checks that +1 −1 for k0 > 0 k0 < 0 d4 k 2π δ(k 2 − m 2 )ε(k0 ) exp[−ik · (x − y)] ≡ i (x − y) (2π )4 (1.38 1 Classical fields. y) = † i (t. non-hermitean. In fact.143)  d3 k  † †  b(k) exp(−ik · x) + a (k) exp(ik · x)  (x) = (2π)3 2k0 where a(k) and b(k) are two independent.

(x)] = −q (x). We consider the wave-function in the configuration space of the one-particle state b† (k)|0 : φ(x) = x|b† (k)|0 = 0| † (x)b† (k)|0 and in the U (1) rotated basis (or of the rotated physical system in the same basis) φ (x) = 0| † (x)U b† (k)|0 . Using (1. The Feynman propagator For the sake of definiteness.142) there follow the commutation relations for operators a(k) and b(k): a(k). i.143) that the field (x) ( † (x)) creates quanta of type b (a) and annihilates those of type a (b). b† (k) are also annihilation and creation operators and it is clear from (1.23) with { . Moreover.145) we get [Q.4 From classical to quantum fields 39 φ and χ . We can introduce the unitary operator U = exp(i Q) ≈ 1 + i Q such that U b† (k)|0 = exp(−i q) b† (k)|0 and U (x)U −1 = exp(−i q) (x) (we have used (1. From (1.144) Thus. One-particle states a (Q = +1) and b (Q = −1) at † The value of the charge q relative to the charges of other fields can only be fixed in the presence of interactions. a(k). These equations can also be interpreted in the following way. i.e. A transparent interpretation of the quanta a and b is obtained by noting that the classical lagrangian (1.† The conserved charge Q (1. b† (k ) = (2π )3 2k0 δ(k − k ) (1. we again consider scalar field theory with abelian U (1) symmetry and q = 1. † → exp(iq ) † .146) Note that the charge of the state created by the field (x) transforming as (x) → exp(−iq ) (x) is Q = −q.146) we get φ (x) = exp(−iq )φ(x). b(k) and a † (k).55) is invariant under abelian global U (1) symmetry: → exp(−iq ) . } replaced by −i[ .1. ]. the fields † carry opposite U (1) charges. the symmetry transformation on the quantum fields is the analogue of the transformation on the classical fields interpreted as the wave-functions of one-particle states.17) can be and expressed in terms of the operators a(k) and b(k): Q = iq =q d3 x : † (x)∂0 (x) − ∂0 † (x) (x) : (1. Therefore. using (1. A generalization to non-abelian global symmetries is straightforward. (1.145) d3 k a † (k)a(k) − b† (k)b(k) ≡ q(Na − Nb ) (2π)3 2k0 This equation defines the operators Na and Nb which are the number operators for the quanta a and b carrying opposite charges: Qa † (k)|0 = qa † (k)|0 Qb† (k)|0 = −qb† (k)|0 (1.e.146)). . a † (k ) = b(k).

Another way of increasing the charge by +1 at (t. x) = (−) (t. x )|0 (t. x ) (t.148) Let the particle a propagate to t > t: the probability amplitude to find it at (t . x) (−) †(+) (t . x ) and to propagate it to (t. x)|0 (t. x)|0 ≡ where †(−) (t. x ).149) This amplitude can also be interpreted as the creation of the particle a at (t.151) = where T denotes the so-called chronological ordering of the field operators. x)|0 (1. symmetries and their breaking the space-time point (t. x) ≡ θ(t − t) 0| (t . x) and lowering it by −1 at (t .150) The Feynman propagator is defined as superposition of the two amplitudes: G (2) (x . x) (t . x) =  d3 k   a † (k) exp(ik · x)   3 2E (2π )  d3 k   b† (k) exp(ik · x)  3 2E (2π ) (1. x) with t > t : θ (t − t ) 0| = θ(t − t ) 0| † (t. Spinor fields Quantization of a Weyl or Dirac field can be based on the phenomenological Pauli’s exclusion principle for spin 1 particles or on the requirement of locality 2 . x ) is given by θ(t − t) 0| (t . x) (t . x ) 0 + θ(t − t ) 0| ≡ 0|T = † † † (t. x) and its reabsorption into the vacuum at (t . x)|0 †(−) (t . x ) |0 d3 k θ(t − t) exp[−ik · (x − x)] (2π )3 2E k + θ(t − t ) exp[ik · (x − x)] i d4 k exp[−ik · (x − x)] 4 k 2 − m 2 + iε (2π ) (1. x ) = θ(t − t) 0| (+) † (t. x)|0 (1.147) (t. x) (t . x )|0 (1. x ) is to create a quantum of charge −1 at (t . x)|0 ≡ †(−) (−) (t.40 1 Classical fields. x) are created by the operators † (t. x)|0 (t. x )|0 (t.

2 and Appendix A. The expansion coefficients become operators with the following anticommutation relations: † † bL (k). dR are expansion ¯ coefficients.81) for the left-handed Weyl fields.154) Other anticommutators vanish.128).152) and for its hermitean conjugation λ† (x) (related to the right-handed field through α ˙ ˙˙ ¯˙ λβ = λ† ε β α ): α ˙ λ† (x) = α ˙ d3 k b† (k)aα (k) exp(ik · x) + dR (k)bα (k) exp(−ik · x) ˙ ˙ (2π)3 2E L (1.157) . It can be also introduced by postulating supersymmetry (a certain symmetry between bosons and fermions.156) we find Pµ = = d3 k † † k µ bL (k)bL (k) − dR (k)dR (k) (2π)3 2E d3 k † † † k µ bL (k)bL (k) + dR (k)dR (k) − dR (k).155) and all other anticommutators vanish. the classical Weyl fields are now promoted to operators but instead of relations (1.153) where aα (k) and bα (k) are positive and negative energy chiral solutions to a free Weyl equation iσ µ ∂µ λ(x) = 0 (normalized as in (A.1.152) into the analogue of (1. Inserting (1. λ†˙ (t. this is the content of the theorem on the connection between spin and statistics). bL (k ) = dR (k). see Chapter 15). As for scalar fields. we postulate (the canonical momentum ˙ conjugate to λα (x) is α = iλ†˙ (σ 0 )βα ) ¯ β λα (t. commutativity at the space-like intervals for observables constructed from the Dirac spinors. dR (k) 3 2E (2π) (1. We are always consistently led to the conclusion that the Weyl and Dirac fields must be quantized by assigning for them certain anticommutation relations. y) = (σ 0 )αβ δ(x − y) ˙ β (1. Using the classical spinors introduced in Section 1. dR (k ) = (2π )3 2k 0 δ(k − k ) (1.e. we can write down the general wave expansion for solutions of the free Weyl equations for the massless left-handed field λα (x) λα (x) = d3 k † bL (k)aα (k) exp(−ik · x) + dR (k)bα (k) exp(ik · x) 3 2E (2π) (1.68)) and bL . x).4 From classical to quantum fields 41 (i. µν ¯¯ = iλσ µ ∂ ν λ (1.

we see that λα (x) creates a helicity h = + 1 state with charge −q at the point x. Indeed.159) i. We redefine the vacuum by subtracting this infinite c-number. The normal ordering for the Weyl fields is defined as: (+) †(−) †(+) :λ† λβ : ≡ λ†(+) λ(+) + λ†(−) λβ + λα λ(−) − λ(−) λα β β β α ˙ α ˙ α ˙ ˙ ˙ (1. symmetries and their breaking It follows from (1. † − NR (k) = dR (k)dR (k) d3 x :λ† σ 0 λ: = q ¯ d3 k † :b† (k)bL (k) + dR (k)dR (k): (2π )3 2E L (1.158) where the subscript ‘+’ (‘−’) denotes positive (negative) frequency parts of λ and λ† given by (1.160) d3 k + − NL (k) − NR (k) (2π)3 2E can be interpreted as the number operators for positive-energy state with U (1) charges +q and −q. h = + 1 = k. k) which. have opposite charges under the global U (1) phase transformations. its energy is not only infinite (as for the scalar fields) but also is not positive definite. Acting with the field λ(x) on the vacuum and † projecting on the momentum eigenstates dR (k)|0 . h = + 1 |λα (x)|0 2 2 = 0|dR (k)λα (x)|0 † † ˙ = b α (k) exp(−ik · x) . It follows from the † † relation (1.155) that. the lagrangian for a single Weyl field has global U (1) symmetry generated by ¯¯ the transformations λ = exp(−iq )λ. This is accomplished by the normal ordering. x) can be expressed in terms of the creation and annihilation operators. there is a minus sign for each interchange of λ and λ† required by the normal ordering. described by a superposition 2 of the plane wave solutions α ˙ x|k. however. The conserved current j µ = q λσ µ λ µ µ 3 0 (∂ j /∂ x = 0) and the corresponding charge Q = d x j (t.152) and (1.159) that operators bL (k) and dR (k) have the interpretation of creation operators of positive-energy states with four-momentum (E. we have redefined the vacuum so that its energy is zero.153). After the normal ordering we get Q=q =q where † + NL (k) = bL (k)bL (k). respectively.155).157) and from the anticommutation relations (1.42 1 Classical fields. with the vacuum defined by bL (k)|0 = dR (k)|0 = 0. In consistency with the anticommutation relations (1.e. After the normal ordering we get Pµ = = ¯¯ d3 x :iλσ 0 ∂ µ λ(x): d3 k † † k µ bL (k)bL (k) + dR (k)dR (k) (2π )3 2E (1.

† β (t. Similar expansions can be written down in terms of the helicity amplitudes.163) where the superscript ‘+’ (‘−’) denotes positive (negative) frequency parts of and ¯ given by (1.161) + d † (k. s)v(k.161). k) with opposite charges under the global U (1) phase transformations = exp(−iq ) . The conserved current . s) are positive and negative energy solutions for the four-component Dirac equation. that b α (k) ∼ a α (k) is the solution of (A. s ) = (2π)3 2k 0 δss δ(k − k ) and all other anticommutators vanish. The functions u(k. y) = δαβ δ(x − y) (1. s) (2π )3 2E The operators b† (k.2 d3 k b† (k. s) and d(k. ˙ 2 α For the Dirac field the procedure described above leads to the field operators (x) = s=1. s)u(k. s ) = d(k. The field λ† (x) α ˙ creates a helicity − 1 state with charge +q at the point x – a state which 2 is a superposition of the plane wave solutions α x|k. s) + d † (k. h = − 1 = k. s)d(k.162) (α and β are spinor indices) with the other anticommutators vanishing. s)γ0 exp(ik · x) ¯ (2π )3 2E + d(k. s) we get b(k. b† (k . s)γ0 exp(−ik · x) ¯ satisfying anticommutation relations α (t.4 From classical to quantum fields 43 ˙ ˙ with four-momenta k µ and positive energy (note. we get for the momentum operator the expression Pµ = d3 x : 0µ (x): = s=1. s). 2 numbers the + 1 and − 1 spin projections 2 2 along the chosen quantization axis (see Appendix A). s)b(k. x). d † (k . s)u(k. Defining the normal ordering for the Dirac fields : ¯α β: (+) = ¯α (+) β (−) + ¯α (+) β (−) + ¯α (−) β − (−) β (+) ¯α (1.60) with positive energy and positive helicity). s) have the interpretation of creation operators of positive-energy states with four-momentum (E. For the operators b(k. s)v(k. s) exp(−ik · x) (2π )3 2E (1.2 d3 k k µ b† (k. s). s = 1.1. s) and v(k. s) and d † (k. s) exp(ik · x) † (x) = s=1.2 d3 k b(k. h = 2 − 1 |λ† (x)|0 † = aα (k) exp(−ik · x).

N − (k. spin s and positive energy. PL. s). s) = b (k. s) exp(−ik · x) For chiral fields in the four-component notation one has d3 k † bL.78). s) α (x)|0 † † α (x)|0 = vα (k.166) u R (k) =  0  0 a (k) L.R (k)u L. s) = b(k. s) = d † (k.R (k) exp(−ik · x) + dR.L . s) = d(k. s)|0 .L ) are positive (negative) energy solutions to the Dirac equation (1. respectively. s) exp(−ik · x) with four-momenta k µ .R . The operator α (x). s)b(k. b(k). Obviously. (We recall that a(k). described by a superposition of the plane wave solutions α x|b(k.R (x) = and similar expressions for vR and vL with a(k) and a (k) replaced by b(k) and b (k).R u L. † Similarly to the case of Weyl fields. symmetries and their breaking j µ = q ¯ γ µ (∂ j µ /∂ x µ = 0) and the corresponding charge Q = can be expressed in terms of the creation and annihilation operators Q=q =q s=1. in the chiral representation for the Dirac matrices we have     a(k) 0 u L (k) =  0 . on the other hand. acting with the field operator α (x) on the † vacuum and projecting on the momentum eigenstates |b(k.2 d3 x j 0 (t. which satisfy the chirality conditions PL. respectively. s) + d(k. s)b(k. x) d3 x : † :=q s=1. s) = b† (k. s) − N − (k.165) where u L. creates a state of particle d with the wave-function given by α x|d(k. s) (2π)3 2E where N + (k.R = u L. one can introduce the unitary operator U which realizes the U (1) transformations on the quantum states and on the quantum Weyl and Dirac fields. a (k). s) † † α (x)|0 † † α (x)|0 = u α (k.44 1 Classical fields. s)d(k.2 3 d3 k :b† (k. s): (2π )3 2E (1. s)| = 0|b(k.164) dk N + (k.L (k)vR. b (k) are solutions to the Weyl .L = vR. As for the scalar fields. s)| = 0|d(k.R vR. s) are interpreted as the number operators for positive-energy states with +q and −q U (1) charges.L (k) exp(ik · x) (2π)3 2E (1. s)d † (k. (1.R (vR. we † see that α (x) creates a state of particle b with the spin s and charge +q at the point x.

167) + exp(iρ)b† (k. It is described by them by the state vectors which form two isomorphic to each other but. s) exp(−ik · x) (2π )3 2E (1. s)v(k.76) (neutral under global symmetry transformations) has the following expansion M (x) = s=1. s) exp(ik · x) † M (x) = s=1. We note already here that the assumption about the identity of the two . We shall make here the following strong assumption which is a necessary (but not sufficient) condition for the change of the reference frame to be a symmetry transformation: the two Hilbert spaces are identical and can be identified with each other (in consequence. s)γ0 exp(ik · x) ¯ (2π )3 2E + exp(−iρ)b(k. different Hilbert spaces. We begin with the Lorentz transformations and consider two observers in two different reference frames O and O . s)v(k.62)). s)u(k.168) In the frame O the state vector |v describes the same physical state as does the state vector |v in O. The state vectors form a representation of the Lorentz group. the operators acting in one of them can also be identified with suitable operators acting in the other space). and by analogy between classical and quantum physics we assume the transformation law (1. The following results apply to free and interacting fields. in general. s)u(k.2 d3 k b† (k. (1.4 From classical to quantum fields 45 equations. Therefore the Hilbert space used by observer O is mapped onto itself by a unitary operator U which is defined by the equation: |v = U |v (1. a chiral left-handed massless field L creates a particle with † charge (−q) and positive helicity and L creates a particle with charge (+q) and negative helicity (see the remarks below (A. it is obvious that a given (the same) physical quantum system is seen by both observers.146)). First of all.25) for wave-functions (in configuration space) of quantum states. A self-conjugate four-component field.) Thus. Symmetry transformations for quantum fields One way to discuss symmetry transformations for quantum fields is to interpret solutions to the classical field equations as wave-functions of quantum states (see the remark below (1.1. s)γ0 exp(−ik · x) ¯ exp(iρ) is a phase factor depending on the choice of the phase for one-particle states (Kayser 1994).2 d3 k b(k.

in the first and the second equation of (1. space reflection (to be discussed later) which is not a symmetry of many physical systems. U α (x)U −1 = exp(− 1 ωµν 2 µν ) β α β (x ) (1.170) the same field operator appears since the Hilbert spaces of both observers are identical). (1. For Poincar´ transformations x µ = µν x ν ≈ x µ + e εµ + ωµν x ν .171) (Internal quantum number indices may be implicit.†) For translations.172) we obtain the infinitesimal form of these transformations ∂ (x) i[Pµ .173) α (x)U −1 = α (x ) (1.170) |v = U (ε.47) interpreted as the equation for the one-particle wave-functions reads: φα (x ) ≡ α x |v = exp( 1 ωµν 2 where α α µν ) β α β x|v (1. .171) simplifies to: U Writing U (ε) = exp iεµ P µ i U (ω) = exp − ωµν M µν 2      (1. ω)|v (the index α refers to the original or to the new reference frame. (1.175) which replace the Poisson brackets (1. one often defines the operator α (x ) = U −1 α (x)U such that the matrices satisfy the equation β v2 | α (x )|v1 = exp( 1 ωµν µν ) α v2 | β (x)|v1 or the operator α (x) = U (x)U −1 such that 2 v2 | (x)|v1 = v2 | (x)|v1 .51). for example.41) and (1. (x)] = ∂xµ and i M µν .174) (x) (1. Therefore.169) x| = 0| x | = 0| α (x) α (x )        (1. Our goal is to find the action of the operator U on the field operators (x) which may carry a Lorentz index α. These equations suggest † Also.46 1 Classical fields. (x) = x µ ∂ ∂ − xν + ∂ xν ∂ xµ µν (1. symmetries and their breaking Hilbert spaces is not necessarily satisfied for such physically relevant changes of the reference frame as.

174) and (1. for example.176).177) In deriving the last equation we have assumed that the vacuum is invariant under the considered group of transformations: U |0 = |0 .174) and (1.131) or (1.4 From classical to quantum fields 47 therefore the identification of the generators P µ and M µν with the four-momentum vector and the angular momentum tensor.1. ω) can be found and the theory. in the Coulomb gauge). The results obtained there remain generally valid for non-abelian groups of transformations and for interacting fields. (1. With U ( ) = exp(i a Q a ). Bjorken & Drell (1965). is Lorentz-invariant.175) follow from the commutation relations imposed in the quantization procedure. the unitary operators U (ε.177) then remains valid for the original fields of the lagrangian but not for the physical fields defined as perturbation around the true vacuum. The Lorentz boosts must be supplemented by additional gauge transformations (to recover the non-covariant gauge in the new frame). although not manifestly so. In every concrete theory those operators can be constructed using equations like (1. We have then: φi (x) ≡ i|v = exp(−i a T a) ij j|v (1. we find the transformation law for the operators U i (x)U −1 = exp(−i a T a) ij j (x) (1.9) as the equation relating the wave-functions of one-particle states in the original and rotated basis built by the action of the field operator (x) on the vacuum: |i = i (x)|0 (the index i refers to the original or rotated basis).157) and one can check explicitly if (1. using (1. If so. Nevertheless.175) is the quantization of the gauge fields in non-covariant gauges (for example.132) or (1. The U (1) transformations in the internal quantum number space on the free quantum fields have already been discussed in the first subsection of this section. An interesting exception to relations (1.176) where the states |v and |v = U |v describe the same physical system before and after the rotation of the basis (we again assume that the two Hilbert spaces can be identified with each other). then the theory is Lorentz-invariant. respectively. Therefore. See. We may follow the strategy of interpreting (1. Eq. . Later we shall also discuss theories with spontaneously broken global symmetries (see Chapter 9) where the ground state of the system (the physical vacuum determined by the equations of motion) is not invariant under the symmetry group of the lagrangian.

J = J. if two right-handed coordinate frames O1 and O2 are connected by the Lorentz transformation with parameters ηi . i. O1 and O2 are connected by the Lorentz transformation with parameters ηi .5 Discrete symmetries Space reflection Under the transformation of space reflection x µ → x µ with t = t. a spinor λα transforming as λβ = M(η . Clearly. x = −x the transformation laws for classical physical observables such as energy E. total angular momentum J. is a dotted spinor λ P as: ˙ ˙ ¯α η ξ ˙ ¯β λ P = [M †−1 (η .178) i. ξ )βα λα . the three-momentum p. they have the same form in the original and in the primed reference frames. η Correspondingly. (1. For scalar fields.23) with { . the Klein–Gordon equations of motion for scalar fields remain form-invariant under those transformations. (1. We may also address the question of the transformation law for classical fields. the most general form of transformation under space reflection is: P (x )= ± (x) exp(iγ ) (x) for a real field for a complex field (1. helicity h = (J · p)/|p| etc. are obvious: E = E. The Weyl spinors transform under space inversion from the one in the ( 1 .48 1 Classical fields. 0) 2 representation of the Lorentz group to the one in the (0.e.180) where the phase γ is arbitrary for a free field (a free scalar complex field has no well-defined parity). 1. ξi (see Appendix A for the definition of the parameters ηi and ξi ) then their images under space reflection. A group G of transformations in the space of internal quantum numbers is a symmetry group provided the action is invariant U L(x)U −1 = L(x). 1 ) representation and vice 2 versa. ] = −T a (1. we get [Q a .179) All internal charges are invariant with respect to space inversion. Indeed. −ξi . symmetries and their breaking where Q a are generators of G. h = −h (1.181) β . ]. } replaced by −i[ . −ξ )]α ˙ λ P . seen from the ˙ ¯ α transforming under Lorentz transformations reflected frame. p = −p.e.

and transforms. −ξ ) and M(η .1. the transformation law follows directly from the transλβ formation of the Weyl spinors (1.183) ¯ because P is numerically equal to the unit matrix. P (x ) defined in (1. in the reflected coordinate frame it describes. 0) in representation R of a group G acting in the space of 2 internal quantum numbers into (0. ¯ γµ γ5 etc.185) satisfies the Dirac’s equation in the reflected frame) provided we choose γ = γ . 1 ) again in representation R (and not R ).185) Such transformation preserves the form of the massive Dirac equation (i. The transformation (1. . therefore. ¯ γ5 . In this case ¯ P (x ) = exp(iγ )γ0 (x) (1.182).5 Discrete symmetries 49 η ξ η since the matrices M †−1 (η . ξ ) are numerically equal. opposite helicity state. according to the representation 2 2 (0. (1. 1 ). in accord with (1.182) satisfies the necessary relation: ˙ ¯˙ η ξ ˙˙ ¯ η [M †−1 (η . −ξ ) = †−1 η M (η . Moreover. Similarly. ξ )]βσ λσ (x) (1.179).184) η ξ and under Lorentz transformations χ Pα (x ) transforms with M(η .182) ¯ where the only role of P ≡ σ 0 is to change the indices from undotted to dotted ¯ ones. The space reflection is a symmetry of the free Dirac fermion. Thus: ˙ ¯˙ ¯P λα (x) → λα (x ) ≡ exp(iγ ) P αβ λβ (x) (1. 2 Therefore parity transformation of the Weyl spinors can be a symmetry of a physical system only if it contains the corresponding pairs of fundamental fields. The positive (negative) energy solution of the Weyl equation iσ µ ∂µ λ = 0 (see Appendix A): ¯ ∼ exp(−ik · x)a(k) (∼ exp(ik · x)b(k)) upon substituting x = −x becomes the solution with positive (negative) energy of the equation iσ µ ∂µ λ = 0. have well-defined properties under the parity transformation. Thus. The Weyl equation is not invariant under space reflection. = ¯ cα˙ .186) where the phase γ is still arbitrary. −ξ )]αβ P βσ λσ (x) = P αβ [M(η . 2 It is very important to notice that the parity transformation on the Weyl spinors transforms fields ( 1 . ¯ γµ . ξ ).184). h = + 1 (− 1 ). For a four-component Dirac spinor in the chiral representation of the Dirac λ matrices. P (x )= ¯ exp(iγ )(σ 0 λc )α ¯ ˙ exp(iγ )(σ 0 λ)β ¯ (x) (1. bilinear forms like ¯ . we get ¯ ¯ χ α (x) → χ Pα (x ) ≡ exp(iγ )Pαβ χ β (x) = exp(iγ )(σ 0 )αβ χ β (x) ¯˙ ˙ ¯ ˙ ¯ ˙ ˙ (1.e.

We can then introduce a unitary linear operator P which transforms the states from one reference frame to another. for example. This also means that the operators P (x) acting in H P can be identified with some operators acting in the original Hilbert space. Bjorken & Drell (1965)). the pion–nucleon interaction can be described. the analogue of (1. −k P|b. there is no choice of the phases γ such that LP ( P ) = L( P ). the pion field must be a scalar or a pseudoscalar (as seen in strong interactions). for example. . From more detailed study of strong interactions it has been determined experimentally that the pion field indeed has spin zero and negative parity (and space inversion is a symmetry of strong interactions. If parity is not a symmetry of the lagrangian. interpreting the transformation on the classical fields as the transformation on the wave-functions. For space reflection to be a symmetry of the physical system. Thus.188) (we assume the vacuum to be invariant under space reflection. and study its action on the field operators.e. it is necessary that these two Hilbert spaces are identical and can be identified with each other. For complex scalar field theory. Werle (1966)). maps the Hilbert space onto itself. i. be different.189) The operator P can be expressed in terms of the creation and annihilation operators (see. in general. see. In quantum theory. for one-particle states we have P|a. this equation is satisfied with P (x)P −1 = exp(−iγ ) (x ) (1. symmetries and their breaking The interaction fixes the relative parities of the different fields provided that space inversion is a good symmetry.50 1 Classical fields. Identifying P (x ) ≡ (x ). For instance.190) (1. −k (1. if the invariance under space reflection is spontaneously broken we face the same situation as for the spontaneously broken global symmetries) and for the creation and annihilation operators we get Pa(k)P −1 = exp(−iγ )a(−k) Pb† (k)P −1 = exp(−iγ )b† (−k) Thus.169) reads: 0| P (x )|v P = exp(iγ ) 0| (x)|v (1. by two Lorentz invariants ¯ and ¯ γ5 . k ≡ Pb† (k)|0 = exp(−iγ )|b. k ≡ Pa † (k)|0 = exp(iγ )|a. assuming spin zero for the pion. the Hilbert spaces H of the observer O and H P of the observer O P (in the reflected coordinate frame) are isomorphic but can.187) (for a real scalar field exp(iγ ) = ±1) where |v P = P|v . if parity is conserved in strong interactions.

187)). 2 2 Thus.192) (1. we get† 0 ¯ Pλα (x)P −1 = exp(−iγ )σαβ λcβ (x ) ˙ ˙ (1. in general.194) up to arbitrary and uncorrelated phases present in the classical solutions of the Weyl equations (A.191) where a (k) and b (k) are positive and negative energy chiral solutions to the free † ¯ Weyl equation iσ µ ∂µ χ(x) = 0 and bR and dL are the annihilation and creation operators. the Weyl fields must be in a real representation. λiα transforming as R and λci α transforming as R . where S P is a constant matrix and parity can be a symmetry of P the theory. we obtain PbL (k)P −1 = bR (−k).61). for example. space reflection cannot be a symmetry of such a theory. respectively). for example. with the expansions analogous to the ones given in (1. with (1.184)) taken as the transformation law for the wave-functions. adjoint representation of SU (N )) then one ˙ ¯˙ can simply identify λα (x ) with S P λα (x ). the system must also contain the right-handed field operators λci α (x). † † PdR (k)P −1 = dL (−k) (1.152) and (1.87) of the classical solutions of the Weyl equations. In this case we can identify the operator P with (they have the same transformation properties under the internal symmetry group) † If the representation R is irreducible and real (like.153):  d3 k  † cα ˙ α ˙ α ˙  ¯  λ (x) = bR (k)a (k) exp(−ik · x) + dL (k)b (k) exp(ik · x)  (2π)3 2E  d3 k  † ¯  bR (k)a α (k) exp(ik · x) + dL (k)b α (k) exp(−ik · x)  λc†α (x) = 3 2E (2π) (1.193) ¯ ˙ ¯ ¯ ˙ P λcα (x)P −1 = exp(−iγ )(σ 0 )αβ λβ (x ) For the creation and annihilation operators. different (the states in H and in HP have opposite helicities − 1 and + 1 . respectively. transforming as (0. 1 ) and R under the Lorentz and internal symmetry group 2 transformations. (A. R = R ⊕ R with. using the properties (A.5 Discrete symmetries 51 For a chiral theory containing left-handed Weyl fields λiα transforming as a complex representation R of the internal symmetry group the Hilbert spaces H P and H mapped into each other by the parity transformation are. . Identifying λα (x ) ≡ λcα (x ) and following the P same arguments as for the scalar fields (see (1. For instance. The action of the parity operator P on the left-handed Weyl ˙ ¯ ˙ field λα (x) can now be defined. For the four-component Dirac fields the parity operation can be realized as a mapping of H onto itself.64).182) (or (1. For this to be ¯ ˙ the case.1. In other words.

γ 0 v(−p. s1 )d † (k2 . s2 )|0 = −|b − k1 . In this case the theory containing the massive Majorana spinor may be parity-invariant.167) the application of the rule (1. s). s1 . The parity of the electromagnetic field is determined by its coupling to the classical current. s1 . symmetries and their breaking so that P (x)P −1 = exp(−iγ )γ 0 (x ) and Pb(p. a Majorana particle at rest (k = 0) is a parity eigenstate with purely imaginary eigenvalue.197) = − exp(−iγ )d (−p. s) † and similarly for b† and d but with the complex conjugate phase exp(iγ ). s2 (1. s) (1. s)P −1 = exp(−iγ )b(−k.52 1 Classical fields.89) Although the phase factor exp(−iγ ) is arbitrary. s)P −1 = − exp(−iγ )b† (−k. s)P −1 = exp(−iγ )b(−p. It is also worth noting that for such a particle P|k. s i. Analogous relations for helicity creation and annihilation operators can be derived using (A.e. The spinor properties γ 0 u(−p. The requirement of the invariance of the classical lagrangian Lint = −ejµ (x)Aµ (x) describing interaction of the charged four-component Dirac field with the electromagnetic field Aµ (x) . s2 ≡ Pb† (k1 .195) (1. One can also check that the free Dirac lagrangian is invariant under space reflection: PL(x)P −1 = L(x ). s)P † −1 (1. dk2 .196) (1. s) have been used (see Appendix A). s) Pb† (k.198) Note that this conclusion does not hold for massless Weyl states because the phases in (1. s) Pd (p. s) = u(p.195) gives two relations Pb(k.194) are arbitrary and uncorrelated (the states with opposite helicities are not related by Lorentz transformations). s = ±i|−k. For a massive Majorana field operator with the expansion given by (1. the (massive) particle and its antiparticle always have negative relative parity (assuming vacuum to be symmetric under the parity operation. jµ (x) = q ¯ (x)γµ (x). P|0 = +|0 ): P|bk1 .199) which are compatible only if exp(−iγ ) = ±i. s) = −v(p. d − k2 .

the system will eventually reach the original initial state. unchanged. i. x)T = − jia (t.5 Discrete symmetries 53 together with the property P (x ) = exp(iγ )γ0 (x) leads to the conclusion that ¯ P (x )γµ P (x )Aµ (x ) = ¯ (x)γµ (x)Aµ (x).57) for the abelian case) we get T (x ) = exp(iγ ) (x) (1. x)) do not change whereas for the space coordinates of the current jia (−t. in quantum theory we identify A0 (x ) ≡ A0 (x ) = P A0 (x)P −1 P A P (x ) ≡ A(x ) = −PA(x)P −1 Time reversal The classical sense of time inversion is clear: the velocities of a system in some final state are reversed and it is moving backwards. We consider now a classical complex scalar field (x).e. time reversal can be described by the change of coordinates† x µ → x µ with t = −t.1. If the fields carry internal quantum number index i and transform as (x) = exp(−i a T a ) (x) then the fields T (x ) transform the same way in spite of the complex conjugation because the change t → −t assures invariance of the charges. h =h (1. x) (since the velocities are reversed). .200) Classical physical observables behave as follows (compare with (1. j0 (t.203) (all time derivatives change sign). Its transformation under time reversal can be inferred from the physical requirement that the charges a (i. Formally. x =x (1. The coupling with the electromagnetic field is invariant under the transformation † Physically. J = −J. From (1. The Klein–Gordon equation is form-invariant under time reversal.e. If the dynamics is invariant under such transformation. of course. a positive energy plane wave with momentum p transforms into a positive energy plane wave in the reflected frame with momentum p = −p. it is however easier to think in terms of the active transformation.179)): E = E.201) A P (x ) = −A(x) (1. Clearly.203). All charges remain. provided P A0 (x ) = A0 (x).16) (or (1. p = −p.204) where the phase factor can actually be only ±1 (since the real component fields cannot change).202) (1. if (∂ 2 + m 2 ) (x) = 0 then also (∂ 2 + m 2 ) T (x ) = 0. in accord with (1. P Thus.

and right-handed Weyl spinors. . −ξi . ξ ) = M(η . −ξ )σ 1 σ 3 . 1 ) under the Lorentz transformations. the right-handed spinors transform into ˙ ˙ χ α → χT (x ) = exp(iγ )(iσ 1 σ 3 )αβ (χ β (x)) ¯˙ ¯α ¯ ¯ ˙ ¯ ˙ (1. similarly to the space inversion.208) Such transformation preserves the form of the massive Dirac equation (i. Moreover. T (x ) satisfies the Dirac equation in the primed variables) provided γ = γ . in the ‘time-reflected’ frame is described by the parameters ηi . In the chiral representation of the Dirac matrices we have (x) → T (x )= ¯ exp(iγ )iσ 1 σ 3 λ (x) 1 3¯c exp(iγ )iσ σ λ (x) ¯ ¯ (1. the transformation on the Weyl fields must involve their complex conjugation.205) where the phase factor exp(iγ ) is arbitrary. the Lorentz transformation properties for the Weyl fields should not change.207) ˙ and χT (x ) transform again as (0.54 1 Classical fields. Taking proper account of the Lorentz indices we finally conclude that λT α (x ) = exp(iγ )(iσ 1 σ 3 )αβ (λ (x))β ¯ (1.e.206) η η ξ is fulfilled because σ 1 σ 3 M (η .204) and the simultaneous transformation of the potential four-vector: A0 (x ) = A0 (x). ¯α 2 The transformation law for the four-component Dirac spinors follows directly from the transformation properties of the left. Similarly. ξi . one can argue that a Lorentz transformation described by parameters ηi . −ξ )αβ exp(iγ )(iσ 1 σ 3 )βκ (λκ ) ¯ (1.208) can be rewritten as T (x † T (x ) = exp(iγ )iγ 1 γ 3 ) = − exp(iγ )i † (x) (x)γ 3 γ 1 (1. Since helicity does not change under time reversal. Thus. symmetries and their breaking (1. from the additional physical requirement that charges are invariant and ¯¯ the space components of the current jµ (x) = λσ µ λ change sign we conclude that the transformation λα → λT α must include the matrix σ 2 . ξ )βκ λκ ¯ η ξ = M(η . Because positive energy solutions with with momenta p must transform into positive energy solutions with −p. the necessary relation η exp(iγ )(iσ 1 σ 3 )αβ M(η . Ai (x ) = −Ai (x). In this case ¯ (1. Furthermore. It is obvious that λT α (x ) ¯ ¯ satisfies the Weyl equation for the left-handed spinors in the primed variables: iσ · ¯ ∂ λT (x ) = 0.209) with the phase γ still arbitrary.

204) as the transformation on the wave-function 0| (x)|v for a state |v we must have: 0T | T (x )|vT = exp(iγ ) 0| (x)|v (1.s.210) to a state vector |v = α1 |v1 + α2 |v2 . with complex coefficients αi .s.h.215) ˙ ¯ ˙ ¯˙ T λα (x)T −1 = exp(−iγ )(iσ 1 σ 3 )α ˙ λβ (x )  ¯ ¯ β † Antiunitarity of the T operator can be justified also by applying it to the time component of (1. T b† (k)T −1 = exp(−iγ )b† (−k) (1.† By applying the antiunitarity condition successively to |v . Therefore. The transformation on the quantum fields does not involve any complex conjugation and the consistency with the classical transformation is ensured due to the antilinearity of T . As for the space inversion.174) with primed variables then T must be antiunitary. |v1 + |v2 and to |w = |v1 + i|v2 one can show that (Wigner 1931) wT |vT = v|w = w|v The l. From (1.212) leads to  T λα (x)T −1 = exp(−iγ )(iσ 1 σ 3 )αβ λβ (x )  ¯ (1.h.214) and similarly for their hermitean conjugates.210) It follows that the operator T must be antilinear (antiunitary): T † T = 1 and T (α1 |v + α2 |w ) = α1 T |v + α2 T |w .210).5 Discrete symmetries 55 In the quantum case we define an operator T such that |vT = T |v . of (1. we conclude that T (x)T −1 = exp(−iγ ) (x ) (1.174): T [H.211) to the state T (T −1 T (x )T |v ). of (1. This can be seen by applying (1.213) for the creation and annihilation operators we get: T a(k)T −1 = exp(−iγ )a(−k). For the Weyl fields we can identify HT with H and the operators λT α with λα .212) where we have applied‡ the rule (1.213) (for a real scalar field we must have exp(−iγ ) = ±1). ‡ Notice that since T is antiunitary 0T | = 0|T −1 .211) )|vT = 0T |T T −1 T (x )T |v = 0|T −1 T (x )T |v (1. . identifying T (x ) with (x ) and comparing with the r. The equation for the matrix elements similar to (1. (x)]T −1 = T −i ∂ (x) T −1 ∂t (x ) is The physically motivated requirement that T H (t)T −1 = H (t ) again leads to the conclusion that if to satisfy (1. the time reflection can be a symmetry of the physical system only if the two Hilbert spaces can be identified with each other.210) can now be written as 0T | T (x (1.1. Interpreting the classical transformation (1.

s)T −1 = i(−1)s exp(−iγ )d † (−k.217) leads to the first relation (1.167) the application of rule (1.219) (1. jµ (x) = q ¯ (x)γµ (x).217) leading. The transformation properties of the electromagnetic field under the operation of time reversal (already discussed) can be determined by its coupling to the classical current. For the massive Majorana field (1. the requirement of the invariance of the lagrangian Lint = −ejµ (x)Aµ (x) together with the properties (1. It can be defined as a transformation on wave-functions (positive energy solutions to the classical wave equations) of quantum systems. s ) T d † (k.218) and the consistency condition exp(−2iγ ) = exp(2iρ). † † T dR (k)T −1 = exp(−iγ )dR (−k) (1.56 1 Classical fields.220) . It is clear that the operator T changes the direction of the momentum of the state leaving its helicity unchanged.216) (up to some arbitrary phase factors associated with the classical solutions for the Weyl spinors) and similar relations for the hermitean conjugate operators. s = 1(2) denotes the spin projection equal + 1 (− 1 ) with 2 2 respect to the chosen quantization axis).93)). For the four-component Dirac fields one gets T (x)T −1 = exp(−iγ )(iγ 1 γ 3 ) (x ) (1. T Thus.91) of the solutions to the Weyl equations we get for the creation and annihilation operators T bL (k)T −1 = exp(−iγ )bL (−k). to the relations T b(k. symmetries and their breaking Using the properties (A. As in the case of the parity transformation.218) one should also remember that T γ µ T −1 = γ µ . We consider a † In verifying the invariance of the quantum lagrangian by using (1. in quantum theory A0 (x ) ≡ A0 (x ) = T A0 (x)T −1 T AT (x ) ≡ A(x ) = −T A(x)T −1 Charge conjugation The transformation of charge conjugation is genuinely connected to quantum physics. upon using the properties of the classical solutions to the Dirac equation (summarized in (A. s)T −1 = −i(−1)s exp(−iγ )b(−k. AT (x ) = −A(x) (1.218) (s = 2(1) for s = 1(2). s ) (1.209) leads to the conclusion† that ¯ T (x )γµ T (x )Aµ (x ) T = ¯ (x)γµ (x)Aµ (x) provided A0 (x ) = A0 (x).

Charge conjugation is a symmetry of the system if the charge conjugate wave function ic (x) can be identified with a solution of the relativistic equation of motion of the system. One should stress that from the fundamental fact of the existence of the negative energy solutions to the relativistic wave equations follows the existence of particle and antiparticle pairs (i. obtained by a transformation on its negative energy solutions and usually also called charge conjugation.223) or ci pos (x) = exp[−i a (−T a )] i j cj pos (x) (1. Thus.1. respectively): the negative energy solutions can always be reinterpreted as antiparticle (positive energy) wave-functions. after such reinterpretation (often and somewhat misleadingly also called charge conjugation) the antiparticle wave-functions may satisfy different equations of motion than the negative energy solutions did and cannot be identified with the ‘charge conjugate’ wave-function of the system.10) for the definition of complex and real representations). complex conjugate to R but has the same space-time Lorentz transformation properties and the same dependence on the space-time variables as the wave-function i (x) c pos (x) = exp[−i(−q) ] c pos (x) (1. This change of the ‘reference frame’ is called the transformation of charge conjugation (we could as well take the active view and replace the original system by a new one with ‘reversed’ charges). This means that the system seen by the observer in the new reference frame satisfies other equations of motion than in the original reference frame and the transformation of charge conjugation is not a symmetry of this particular system. the states whose wave-functions transform as R and R . For an observer using this reference frame the same physical system is described by another wave-function ic (x) which transforms as representation R of G. However.e.224) (see (E.221) = exp(−i T a) i j j pos (x) (1. This transformation may or may not be a symmetry of a given system. .222) We assume that the group G is a global symmetry group of this system.5 Discrete symmetries 57 physical system which carries some internal quantum numbers (charges) and is described by the wave-function i (x) transforming as a representation R of a group of transformations G: abelian or non-abelian i pos (x) pos (x) = exp(−iq ) a pos (x) (1. One can introduce a ‘charge conjugate’ reference frame (a ‘reflection’ in the space of internal charges). ‘opposite’ internal charges are assigned to the same system.

It is easy to check that they are invariant under the operation i (x) → c i (x) = exp(iγ )( i (x)) Aa (x) → Aca (x) : Aca (x)(T a )∗ = −Aa (x)T a µ µ µ µ (1. i. symmetries and their breaking Again.e.225) for the relativistic scalar wave equation provides the interpretation of its negative energy solutions and one predicts the existence of particle–antiparticle pairs described by wave-functions with the same space-time form but opposite internal charges.e. the operation λiα (x) → λiα (x) neg 1 transforms the negative energy solution λiα (x) (transforming as R and ( 2 .e. as predicted by the presence of the negative energy solutions to the equations of motion. the existence of the symmetry operation i (x) and i (1. Moreover. if i (x) are solutions of the Klein–Gordon equation the same is true for the functions ic (x). We now consider several examples. 0) under . transforms solutions i (x) which have no wave-function interpretation into c pos (x) which can be interpreted as wave-functions of antiparticles (with reversed i global quantum numbers).58 1 Classical fields.225) transforms negative energy solutions into positive energy ones and vice versa. Instead. This equation is invariant under the operation i (x) → c i (x) ≡ exp(iγ ) i (x) (1. the wave-function of the antiparticle in the charge conjugate vector potential Aca (x) has the same form as the particle wave-function in the potential Aa (x). neg i. It is trivial but important to notice that the operation (1. The equations of motion follow then from the lagrangian L = ∂ µ + ig Aaµ T a = ∗ ← 2 →µ ∂ µ − ig Aa T a ∂ + ig Aaµ T a µ (in going from the first to the second line the hermiticity of the generators T a has been used). the space-time dependences of the functions pos c pos (x) are the same.226) i. this does not change the fact that particles and antiparticles do exist in pairs. We begin with a set of free complex scalar fields i (x) which satisfy the Klein–Gordon equation. µ µ For the Weyl equation there is no symmetry operation which transforms its neg ˙ c pos ˙ ¯ α neg ¯ cα pos (x)) of negative energy solutions λiα (x) (λi (x)) into solutions λiα (x) (λi the same equation with the same Lorentz transformation properties and the same pos ˙ ¯ α pos space-time dependence as λiα (x) (λi (x)) and with the internal charges reversed pos ¯˙ compared to the charges of λiα (x).225) Thus. We may extend our considerations to scalar fields with gauge symmetries. coupled in a gauge-invariant way to (abelian or non-abelian) gauge vector fields. Thus.

so that c indeed satisfies the Dirac equation.230) with Aca (x) defined in (1. Aa → Aca (x) µ µ (1. s) = ¯ u(p.228) transforms the negative energy solutions neg (x) into positive energy c pos (x) which transform as R and have the same space-time dependence as pos (x). 2 The same conclusion remains valid for the Weyl fields coupled to the abelian or non-abelian gauge fields.1. with the same Lorentz and space˙ pos ¯ α pos time dependence as λi (x) and λiα (x). 0) and R . C −1 ¯ c = − exp(−iγ ) (1. The operation (x) → c (x) = exp(iγ ) (1. transforming as R and (0. 1 )). For a Dirac field coupled to (abelian or non-abelian) gauge fields the lagrangian / L = ¯ i/ − g Aa T a − m ∂ is invariant under simultaneous transformation → c .† However.229) where the 4×4 matrix C (its explicit form and properties are given in Appendix A) satisfies (Cγ0 )γ µ (Cγ0 )−1 = −γ µ . λicα → λiα 2 2 which transforms the negative energy solutions of each of the equations into the positive energy solutions of the other equations. s).and 2 2 pos ˙ ¯ cα pos (x) transforming as R. s) = v(p. For a general Dirac field (a solution to the massive Dirac equation) there exists the symmetry operation c = exp(iγ )Cγ 0 T = exp(iγ )C ¯ T .5 Discrete symmetries 59 ˙ ¯ α pos internal and Lorentz groups. neg (x) is the same as the helicity of λiα (x). respectively.e. for a quantum system described by a set of left. Therefore the positive energy solution to a Weyl equation is a wave-function of a particle with h = − 1 (+ 1 ) and the ‘charge conjugate’ 2 2 form of the negative energy solution is the wave-function of its antiparticle with h = + 1 (− 1 ). . 1 ) there exists the symmetry operation λiα → λiα . s) and C v T (p. ¯ For the four-component spinors we have C u T (p. right-handed Weyl equations with solutions λiα (x). Such a system is equivalent to a Dirac field (x) = λiα (x) ˙ ¯ cβ λi (x) c λiα (x) ˙ ¯β λi (x) (1. λi c ¯˙ ¯ ˙ ( 1 .226) (remember that µ ¯ † Note that the helicity of the solution λi α pos ˙ s are anticommuting c-numbers. respectively) into λi (x) which satisfies the Weyl equation for right-handed spinors (i. (0.227) which satisfies the Dirac equation.

4 we see that. Cb† (k)C −1 = exp(−iγ )a † (k) (1. † It is worth stressing the difference between charge conjugation and time reversal.234) (1. The states in Hc have the same space-time description but opposite internal charges. ¯1 γµ 2 = − ¯2 γµ 1 etc. Remembering our discussion in Section 1. We can then define a linear (unitary) operator C. With few exceptions (see (1. For scalar field theory. we see that. The creation and annihilation operators defined in (1. i. |v c = C|v . we can identify the operators c (x) ≡ † (x) (1. In the presence of interactions this freedom can be used to define the maximal discrete symmetry of the system.232).229) is unphysical and can be taken as zero. After such transformations. the operators acting in H can be identified with some operators acting in Hc ). compare (1.232) of an arbitrary phase reflects the freedom to redefine any quantum state by multiplying it by a phase factor. This operation can be a symmetry only if H and Hc are identical and can be identified with each other (in consequence. Moving now to the description of the physical system in terms of the state vectors (rather than their representation in the configuration space) the operation of charge conjugation maps the original Hilbert space H onto the Hilbert space Hc which describes the physical system in the ‘charge conjugate’ frame.143).229): ¯1 2 = ¯2 1 .243)).233) for complex scalar fields (1. the space-time dependence of the solutions (x) and c (x) remains the same. they have the same space-time dependence. the arbitrary phase γ in (1.225) with (1.213) with (1.e. k) into the antiparticle state with the same fourmomentum.143) satisfy the following relations Ca(k)C −1 = ±a(k) for real scalar fields (1. Next. We use the fact that for the wave-functions 0| c (x)|v c = exp(iγ ) 0| (x)|v .126) and Ca(k)C −1 = exp(−iγ )b(k).60 1 Classical fields.126) and (1. symmetries and their breaking the Dirac lagrangian is invariant up to a total derivative). .231) getting† C i (x)C −1 = exp(−iγ ) † i (x) (1.204) and (1. The transformation properties of the spinor c c c c bilinears follow from (1.232) The presence in (1. charge conjugation transforms a particle state with four-momentum (E. This difference follows from the antilinearity of T . in agreement with the transformation properties under internal symmetry group. consistently with its definition. we would like to find the action of C on the field operators.225) and (1.

with eigenvalues ±1. for abelian charges C Q = −QC. ¯ The K0 –K0 mixing will be discussed in more detail at the end of this section and in Section 12.238) (1.232) we get Cλα C −1 = exp(−iγ )λc α (1. The presence in this equation of the phase factor exp(iγ ) is crucial for a phase-convention-free conclusion. With λα and λc transforming as R and R . It is clear that a particle state with momentum k and helicity h is transformed under charge conjugation into an antiparticle state with the same momentum and helicity. † † CdR (k)C −1 = exp(−iγ )bR (k). For the Weyl fields. (1.235) are eigenvectors of the charge conjugation operator. Assuming the vacuum to be invariant under C.237) and a similar equation for λ† with the opposite phase. the Hilbert space HC can be identified with H (and C can be a symmetry operations) only if the set of those fields transforms as a real representation of the group G. Using the expansions of the α ˙ fields λα and λc given in (1. respectively. it then follows that the states given by the linear superposition of the particle and antiparticle states |kC = exp(iφ± ) † a (k)|0 ± exp(iγ )b† (k)|0 √ 2 (1. . when C (and/or C P) is a good symmetry and the system is neutral in all conserved charges (the latter is necessary since.5 Discrete symmetries 61 It is sometimes very useful to consider the eigenstates of the charge conjugation operator C (similar comments apply to the combined transformation C P).236) In a very similar manner one can construct the states at rest invariant under the C P transformation even when C and P are not conserved separately.7. For instance. and following the same arguments as for the (1.239) up to arbitrary phases present in the classical solutions for the Weyl equations.1. −k . Important physical examples are the positronium state (electromagnetic interactions conserve both C and Q but positronium is ¯ ¯ electrically neutral) and the K0 –K0 (or B0 –B0 ) mixing by electroweak interactions (CP is almost conserved but the flavour charges like strangeness or beauty are not).152) and (1. so no common eigenstates exist) then the description in terms of the C (or CP) eigenstates is convenient. α respectively. Note also that the state |kC can be written as |kC = exp(iφ± ) † a (k)|0 ± Ca † (k)|0 √ 2 (1. For the states with momentum k the C P acts as follows: CP|a.153) we arrive at: α CbL (k)C −1 = exp(−iγ )dL (k). k = exp(iγ )|b. for example.

Thus. s)C −1 = exp(−iγ )d(k.245) It is its own charge conjugate up to a sign. Weyl and Dirac particles described by state vectors in some Hilbert space. For those transformations which are symmetries of the system one can introduce unitary operators acting on quantum fields. We have identified the requirements which are necessary for these transformations to be symmetries of free (or interacting with the gauge fields) quantum systems of scalar.46) for spinors u(k. s) Consistency of these two equations requires exp(−iρ) exp(−iγ ) = ±1 Thus. s)C −1 = exp(−iγ )b† (k. time and charge inversions) on classical fields.242) for the creation and annihilation operators. s) and v(k. s) Cd † (k. s)C −1 = exp(−iρ) exp(−iγ )b† (k. s) Cb(k.199)). We have used the transformation properties (A. Summary and the C P T transformation We have discussed the action of the discrete transformations (space. It is interesting that even free particle systems are not all invariant under the three discrete transformations. s).62 1 Classical fields. Also. We have learned that a system of free Weyl particles which transforms as a complex representation R of some group of internal symmetries is invariant under the time reflection but not under separate . For a Majorana field with expansion (1.240) which leads to (1.241) (1. it follows that a Majorana particle is an eigenstate of the charge conjugation operator with eigenvalue ±1. s)C −1 = exp(−iρ) exp(−iγ )b(k. s) (1.229) we obtain Cb† (k.244) =± M (x) (1. the Majorana field satisfies C M (x)C −1 (1.167). a state of such a particle at rest may be a CP eigenstate with eigenvalues ±i (see (1.243) (1. the action of C on the four-component Dirac field operators is given by c (x) ≡ C ¯ † = exp(iγ )C (x)C −1 Cb(k. using the definition (1. symmetries and their breaking Following the same arguments.

e. i. The most general hermitean mass term of the lagrangian is ¯ ¯B Lmass = −m AB λ A λc − m AB λ A λc B (1. For C P to be conserved m AB must be real. Similar considerations apply to complex Yukawa couplings. Finally. and equal magnetic moments. They include several measurements .152) and (1. The importance of the C P T transformation follows from a fundamental theorem of quantum field theory which says that C P T invariance follows from the basic requirements of locality. It is natural to ask what happens to such a system under simultaneous C and P transformations. Among the consequences of the C P T theorem are equal masses for particles and their antiparticles. which can be used to make the maximal symmetry of the lagrangian explicit. A number of experimental upper limits on violation of the C P T theorem can be found in the Particle Data Book (1996). for C to be conserved the mass matrix must be symmetric. Lorentz invariance and unitarity. −x).248) independently of the form of the matrix m AB . One should also stress that often there exists some arbitrariness in the field phase conventions. −x) and then using (1.5 Discrete symmetries 63 transformations of space reflection and charge conjugation. equal and opposite electric charges. Similarly. For quantized fields CPλα (x)(CP)−1 = exp(−γ )λα (t.e.246) A particle state with momentum k and helicity h is transformed under C P transformation into an antiparticle state with opposite momentum and helicity (as expected from the Dirac sea interpretation of the negative energy solutions to the Weyl equation). equal lifetimes. The discrete symmetries can easily be broken by interactions or even mass terms.247) It is straightforward to check that if Lmass is to conserve P we must have m AB = m B A .1.153) we can infer the transformation properties of the annihilation and creation operators: CPbL (k)(CP)−1 = exp(iγ )dR (−k) † † C PdR (k)C P −1 = exp(iγ )bL (−k) (1. We conclude that all the considered systems are invariant under the C P and C P T transformations. the mass matrix m AB must be hermitean. it can be checked that Lmass remains invariant under the C P T transformation CPT Lmass (CPT )−1 = Lmass (1. i. Consider the theory of Weyl fermions in which each set of λs transforming as R has its counterpart λc transforming as R of the internal symmetry group. It turns out that the Weyl equation is invariant under the replacement λα (x) → λC P (x) = ¯˙ exp(−iγ )λα (t. the field λC P (x) satisfies the same equation as the field ¯˙ λα (x).

we shall discuss the effects of the C P violation in the K0 – ¯ K0 mixing.249) (1.254) (1. The names of KL(ong) and KS(hort) originate from their drastically different lifetimes τL 5 × 10−8 s and τS 9 × 10−11 s. (For a review of speculations on the possibility of violation of the C P T theorem. . Mavromatos & Nanopoulos (1996).255) We choose our phase convention for normalization of the neutral kaon states so ¯ ¯ that CP|K0 = |K0 and CPT |K0 = |K0 . their masses m L mS 498 MeV are almost equal: m ≡ m L − m S 3. The most stringent bound exists ¯ for the mass difference of the K0 and K0 : m K /m K < 9 × 10−19 . On the other hand. More details about C P violation in the neutral kaon system can be found in Jarlskog (1989) and Bernabeu (1997). They were the first processes in which violation of the C P symmetry was observed (Christenson et al. the ¯ neutral kaon states K0 and K0 combine to the following physical ones: |KL |KS = = 1 2(1 + |¯ |2 ) ε 1 2(1 + 1/2 |¯ |2 ) ε 1/2 (1 + ε)|K0 − (1 − ε)|K0 ¯ ¯ ¯ (1 + ε)|K0 + (1 − ε)|K0 ¯ ¯ ¯ (1.5 × 10−12 MeV.251) (1. muons and protons. We recall that both pions and neutral kaons belong to the lightest quark– antiquark bound states (mesons) which are stable with respect to strong interactions.) ¯ C P violation in the neutral K0 –K0 -system In this subsection. In spite of the tremendous accuracy with which the C P T theorem has been verified. down to the Planck scale.64 1 Classical fields. The physical processes under consideration here will be neutral kaon decays to two pions. 1964). see Ellis. it is of great interest if the theorem remains valid at even smaller scales of length.250) (1.253) |π 0 |K0 ¯ |K0 = |d s ¯ ¯ = |s d While the pion states π + . respectively. Their flavour structure is the following: |π + |π − ¯ = |u d = |d u ¯ = 1 ¯ ¯ √ |u u − |d d 2 (1. π − and π 0 are physically observable particles. symmetries and their breaking on electrons and positrons. They decay only due to electroweak interactions.252) (1.

KL could not decay to two pions which are C P-even in the S-wave. Landau & Lifshitz.259) is not hermitean.261) and (1. we find the time evolution equation for a(t) and a(t) ¯ i i d dt a(t) a(t) ¯ = H11 H21 H12 H22 a(t) a(t) ¯ (1. (1959) §43).258) | (t) = (Hs + Hw )| (t) dt where Hw is the electroweak interaction hamiltonian. This explains its long lifetime. In such a case.259) expressed in terms of matrix elements of Hw between strong interaction eigenstates: i (M = M † .262) Here.260) Hi j = Mi j − ij 2 i|Hw |n n|Hw | j 3 + O(Hw ) (1. For t > t0 . because decays to three pions are strongly phase-space suppressed.261) Mi j = m K δi j + i|Hw | j + P m K − En n ij = 2π n 3 i|Hw |n n|Hw | j δ(m K − E n ) + O(Hw ) (1.262) are obtained by sending t0 to (−∞) and turning Hw adiabatically off at t → ±∞ (see. The state | (t) satisfies the Schr¨ dinger equation o d (1.257) and bn (t = t0 ) = 0. ¯ ¯ Hs |K0 = m K |K0 . Hs |n = E n |n (1. It parametrizes the decay rates of the kaons to other strong interaction eigenstates. Then. † Eqs. In reality. This is the reason why the absorptive part i j arises in Hi j . . the parameter ε would vanish. we write it as a linear combination of eigenstates of the strong interaction hamiltonian Hs ¯ | (t) = a(t)|K0 + a(t)|K0 + ¯ n bn (t)|n (1. because the coefficients bn (t) have been removed from the evolution equation.† The matrix Hi j in (1. for example. KL indeed decays much more often to three than to two pions. Let us consider a state which at the initial time t0 is a neutral kaon at rest.5 Discrete symmetries 65 If C P was conserved in Nature. = † ) (1.256) where Hs |K0 = m K |K0 .1. KL and ¯ KS would be C P-odd and C P-even. respectively. Solving this equation perturbatively in Hw . The states |n stand for all the eigenstates of Hs which are ¯ different from |K0 and |K0 . the sum over n denotes the sum-integral over all states different from |K0 ≡ ¯ |1 and |K0 ≡ |2 . while P stands for the principal value. (1.

The observable we would like to consider is the ratio of decay amplitudes of KL and KS into two-pion states with vanishing total isospin εK = A[KL → (π π ) I =0 ] A[KS → (π π ) I =0 ] (1.S and L. masses and decay widths of the physical neutral kaons KL and KS .S − i 2 L.† Once we know the eigenbasis of Hi j . Although the latter equality is violated in Nature.266) with some time-independent coefficients cL and cS . while kaon decays can proceed via only single W ± boson exchange.S are.259) is trivial. then the off-diagonal elements H12 and H21 would be equal. The considered state | (t) can now be written as follows: | (t) = cL exp −im L t − 1 2 Lt 1 2 |KL St + cS exp −im S t − |KS + n bn (t)|n (1. i.17). The parameter ε which occurs in these equations is related to ¯ the matrix Hi j as follows: ε= ¯ 1/2 H12 − H21 H12 + H21 1/2 1/2 1/2 1/2 . This is because the leading contribution to i|Hw | j originates from double W ± boson exchange (Fig.267) † This may not hold only in some really peculiar phase conventions for the parametrization of the Cabibbo– Kobayashi–Maskawa matrix and/or for the complex square root.S = m L. too. we have assumed that H21 tends continuously to H12 when H21 tends to H12 in the limit of no C P violation. 1/2 (1. 12. solving (1. m L.265) Here.263) If electroweak interactions were symmetric under C P. respectively.255). respectively. H11 = H22 ≡ H (1.66 1 Classical fields. The eigenvalues of the matrix Hi j are λL. symmetries and their breaking 2 It is important to note that the O(Hw ) and O(Hw ) parts of Hi j are actually of the same order in the electroweak coupling constants. Thus.264) while the corresponding eigenvectors |KL and |KS are given in (1. it is usually convenient to verify what happens in this limit with the relations we shall discuss below. .254) and (1.e.S = H ± H12 H21 1/2 1/2 (1. The C P T symmetry of electroweak interactions implies that the diagonal elements of the matrix Hi j are equal.

(1. We know from experiment that εK 1 and L 2(m L − m S ) ≡ 2 m. The C P T symmetry implies that A0 = A∗ . 12 is approximately ¯ proportional to A∗ A0 = A∗2 . depending on phase conventions in the parametrization of the CKM matrix.1.271) The advantage of the latter formula is that it is manifestly independent of sign conventions used for complex square roots and of phase conventions in the parametrization of the Cabibbo–Kobayashi–Maskawa (CKM) matrix. Therefore. Next. (1. Using this fact together with 0 (1. This means that the sum representing i j in (1.260) of Hi j into hermitean and antihermitean parts Mi j and i j .262) is dominated by (ππ) I =0 intermediate states.255). we take into account the experimental fact that KS decays dominantly to isospin-zero two-pion states. the phase of the amplitude A0 can take any value between 0 and 2π.269) (1. In the . 2(1 + i) m|A0 |2 (1.267) and (1.268) where δ0 is the strong phase shift in elastic scattering of two pions in the isospin ¯ zero state. 0 0 0 εK Im(M12 A2 ) 0 exp(iπ/4). one finds after a short calculation that εK = H12 A0 − H21 A∗ 0 1/2 1/2 1/2 1/2 H12 A0 + H21 A∗ 0 (1.254).273) In general. Consequently.269). and Im( 12 A2 ) can be neglected. (1. S εK = 2 (1 − εK ) = H12 A2 − H21 A∗2 0 0 εK 2i Im(M12 A2 ) + Im( 12 A2 ) 0 0 .5 Discrete symmetries 67 where 1 |(ππ) I =0 = √ |π + π − + |π − π + − |π 0 π 0 3 It is convenient to express εK in terms of the decay amplitudes A[K0 → (π π ) I =0 ] ≡ A0 exp(iδ0 ) ¯ ¯ A[K0 → (π π ) I =0 ] ≡ A0 exp(iδ0 ) (1.270) which can be equivalently rewritten as H12 A2 − H21 A∗2 0 0 2 (1 − εK ) 1/2 1/2 2 [2(m L − m S ) − i( L − S )]|A0 |2 4H12 H21 |A0 | (1. Now.272) where we have used the decomposition (1. √ 2 m|A0 |2 (1. respectively.265). it is time to make some approximations. Consequently.

. the quantity Im(M12 A2 ) is a 0 0 measure of C P violation in the neutral kaon mixing.274) 2 m The equality of ε and εK for real A0 follows directly from comparing (1.270) and ¯ (1. ¯ Therefore ε is not an observable.254) and (1. Consequently. when C P is violated. in general.272) εK 2i Im M12 + Im 2(1 + i) m 12 (1. as follows from (1. t1 . Calculate the variation S under the transformation x(t) → x(t) + δx(t) ∇ Derive the Newton equation m d2 x/dt 2 = −∇ V from the principle of the minimal action (subject to the constraint δx(t1 ) = δx(t2 ) = 0).68 1 Classical fields. Problems 1. (1. and we can write Im M12 ε εK √ ¯ exp(iπ/4). the value of ε ¯ is bounded from below when C P is violated. then the mass eigenstates KL and KS in (1.27). ε does not even need to be small. ¯ The value of εK is measured quite precisely nowadays. momentum p and orbital momentum L = r×p are constants of motion if the potential is symmetric under translations and rotations.1 Consider the action for a motion of a particle in a potential V (x): S([x]. the parameter ε is.2 Write down the Newton equation d2 x/dt 2 + ∂ V /∂ x = 0 in the new reference frame. A0 is approximately real. In other words.275) However. then either Im M12 or Im 12 must be non-vanishing.255) are not C P eigenstates. if εK differs from zero. t2 ) = t2 dt t1 1 dx m 2 dt − V (x) where [x] denotes a functional of the path x(t). symmetries and their breaking standard phase convention (see Chapter 12). we can write in analogy to (1. On the other hand. Thus. On the other hand.273). As we have observed before. Construct the lagrangians which give both equations and compare with (1. we cannot neglect Im 12 before specifying the phase convention. 1. Riazuddin & Ryan (1969). the phase of 12 is approximately opposite to the phase of A2 . corresponding to the transformation t = t + εt 2 . dependent on phase conventions. However.265). When it is ¯ small. and it is of order 10−3 (for the description of some of the methods of measuring εK see Marshak. unlike εK . Prove that energy E.

6 In abelian gauge theory the dual field strength tensor F µν is defined as ˜ F µν = 1 εµνρσ Fρσ . 2 ˜ Show that Fµν F µν is a total divergence ˜ F µν Fµν = ∂µ K µ where K µ = εµνρσ Aν Fρσ In a non-abelian case a matrix-valued dual field strength is ˜ G µν = 1 εµνρσ G ρσ 2 and ˜ Tr[G µν G µν ] = ∂µ K µ where K ρ = ερσ µν Tr[G σ µ Aν + 2 Aσ Aµ Aν ] 3 ε0123 = −1 . Finally.5 Show that for an abelian gauge theory the Noether current corresponding to the gauge symmetry transformation (1.Problems 69 1. ˜ 1. 1.94) is S µ = ∂ρ (F ρµ (x)) and that the associated conserved charge Qs = d3 x S 0 (x. Thus gauge invariance leads to no new conservation laws as compared to global invariance. 1. where F µν = ∂ µ Aν − ∂ ν Aµ . t) either vanishes or reduces to the electric charge associated with global U (1) invariance provided approaches an angle-independent limit at spatial infinity.4 Consider a closed system of a charged classical particle interacting with an electromagnetic field. Express µν in terms of the electric E and magnetic B fields.3 Use (1. Write down the action in terms of the electromagnetic potential Aµ and derive the equations of motion and the energy–momentum conservation laws.35) to study conservation laws in classical electrodynamics (in the absence of sources) with L = − 1 F µν Fµν . derive the angular momentum vector for the electromagnetic field and split it into orbital and intrinsic orbital momentum parts. Derive the canonical 4 energy–momentum tensor and bring into the symmetric and gauge-invariant form µν ν = F µα Fα + 1 g µν Fαβ F αβ 4 Check that it is traceless. The conservation law ∂0 00 +∂i 0i = 0 connects the change in the energy density to the Poynting vector. In particular show that the energy density is 00 = 1 (E2 + B2 ) and the momentum 2 density 0i = (E×B)i is the Poynting vector.91) and (1.

Show that the charge Qa = a d3 x Y j0 (x. Dµ ]] + [Dρ . t) is time-independent. Since Y ja is not gauge-invariant.124). Dν ]] = 0 and the Bianchi identity Dµ G νσ + Dν G σ µ + Dσ G µν = 0 or ˜ Dµ G µν = 0 1. symmetries and their breaking 1. Dρ ]] + [Dν . Show that the current Y j generates no new charges provided approaches an angle-independent limit at spatial infinity.7 Check the Jacobi identity for the covariant derivatives [Dµ . [Dν .123).8 In pure Yang–Mills theory the Noether current for global symmetry transformations is Y a jµ = cabc Ab G cν ν µ 2 Tr[G νµ Dν ] g2 and for gauge transformations Y µ j = Using field equations Dµ G µν = 0 show that both currents are conserved. Show that Q a is gauge-covariant against gauge transformations which µ approach a definite angle-independent limit as |x| → ∞. [Dρ . provided Yja falls off sufficiently rapidly at large |x|. Using Noether’s theorem derive the conserved symmetry current: a a Jµ = jµ + cabc Ab G cν ν µ a a where jµ is given by (1. Show that the ordinary conservation of Jµ is equivalent a to the covariant conservation of jµ and the gauge field equation (1. Include fermion fields. the fall-off requirement restricts the large |x| behaviour of gauge transformation.70 1 Classical fields. [Dµ . Show that µν ν ∂µ G a = g Ja .

Consequently. ¯ then x|t = dx x| exp[−(i/h )H (t − t)]|x ¯ 71 x |t . The eigenstates of the position operator are introduced as follows X H (t)|x. The quantum system has no definite trajectory and it is necessary to take a sum over all possible ones. t X S |x = x|x with the relation |x = exp[−(i/h )H t]|x. t |x. Transition matrix elements as path integrals Consider a quantum mechanical system with one degree of freedom. t = x|x. time evolution changes the position eigenstate into one in which position is not determined. according to the superposition principle. and is a Green’s function: define |t by H |t = ih (∂/∂t)|t .1 Path integrals in quantum mechanics Our first aim is to show that the matrix elements of quantum mechanical operators can be written as functional (path) integrals over all trajectories. t = x | exp[−(i/h )H (t − t)]|x ¯ (2. t ¯ where H denotes the hamiltonian of the system. the need for integration is obvious: the position operator does not commute with the momentum or the hamiltonian.2 Path integral formulation of quantum field theory 2.1) Heisenberg picture Schr¨ dinger picture o corresponds to the transition from the eigenstate |x at the moment of time t to the state |x at the time t . Intuitively. with the integrand dependent on the action integral. The matrix element x .

Using (2. x j−1 ) + O ε 2 exp h h 2π h ¯ ¯ ¯ (2. xn+1 . x j−1 ) + O ε2 exp h h 2π h ¯ ¯ ¯ dpj i i = p j (x j − x j−1 ) − ε H ( p j . we use the completeness relation at each of the times t j : dx j |x j . respectively.3) where x0 .2) (2.4). x) is now the classical c-number hamiltonian.5) and we obtain the following expression for the matrix element (2.72 2 Path integrals The matrix element (2. t j−1 = x j exp − ε H x j−1 h ¯ iε x j |H |x j−1 + O ε 2 = x j |x j−1 − h ¯ (2. x j−1 )(t j − t j−1 )] (2. . t |x. (2.6) . X ) to be of the form H = f (P) + g(X ) we can write x j |H |x j−1 = = d p j x j | p j p j |H |x j−1 dpj i exp p j (x j − x j−1 ) H ( p j . t j |x j−1 . . t j |x j−1 .1): n n+1 x . . t0 . t j−1 = dpj i i p j (x j − x j−1 ) 1 − ε H ( p j .4) where H ( p. . x j−1 ) 2π h h ¯ ¯ (2.3) becomes x j . tn+1 are to be understood as x. t j x j . t. Choosing the hamiltonian H = H (P. t . t = lim n→∞ dx j j=1 dpj i exp h 2π h ¯ ¯ j=1 n+1 [ p j (x j − x j−1 ) j=1 −H ( p j .1) we shall first represent as a multiple integral which will then be used to define the functional integral by a limiting procedure. First we divide the time interval (t − t) into (n + 1) equal parts of length ε t = (n + 1)ε + t t j = jε + t ( j = 1. n) Next. t j | = 1 together with i x j . x .

i.10). We can use another approach.8). x)] dτ ˙ t (2. There are. theories where the lagrangian is L= 1 2 f (x)x 2 + g(x)x − V (x) ˙ ˙ (2. If the hamiltonian is of a simple form H = (1/2m)P 2 + V (X ) (2. x) dτ is the action integral over the trajectory x(τ ).9) The final result has the form of a functional integral over configuration space: x . t |x. we can choose the path integral formulation as the quantization prescription for a system with the classical hamiltonian in the form (2. where ˙ 1 2 L(x. for instance. This result we shall write in the compact form x . t . to define the quantum theory by functional integral (2.10). systems for which canonical quantization is ambiguous due to non-commutativity of operators P and X .1 Path integrals in quantum mechanics 73 where the limit n → ∞ (ε → 0) has been taken and the O ε 2 terms neglected. t |x. (2. Shifting the integration variables: p j → p j − m( x j /ε) we obtain p2 dpj i j exp ε pj xj − 2π h 2m h ¯ ¯ where x j = x j − x j−1 and 1 = Nj d p j exp − i p2 j ε h 2m ¯ = 1 1 i m exp ε N j 2π h h 2 ¯ ¯ xj ε 2 (2.8).7) is the promised path integral representation of x .8) we have derived path integral representation (2.7) (Dx D p/2π h ) ≡ ¯ ¯ τ [dx(τ ) d p(τ )/2π h ] is called a functional integration over all phase space. namely. x(t ) = x implied in this case.6).10) Here S[x] = t L(x. however. x) = 2 m x − V (x) is the Lagrange function and the normalization factor N ˙ ˙ is given by 1 = N D p exp − i h ¯ t t p dτ 2m Starting with the canonically quantized theory described by hamiltonian (2.2. with the boundary conditions x(t) = x. t |x.8) it is convenient to perform the momentum integrations in (2. Eq. Here belong. t = t 1 N Dx i S[x] exp h 2π h ¯ ¯ (2.11) . t = i Dx D p exp 2π h h ¯ ¯ t [ p x − H ( p.e. Then our derivation proves the equivalence of the path integral and canonical quantization methods for systems described by hamiltonian (2.

2 Theories like (2. If S[x(τ )] h we can evaluate (2. the path integral formulation allows a relatively simple understanding of the classical and semi-classical limit. one can postulate (2. in fact.14) We shall encounter a similar situation in Chapter 3 when quantizing gauge field theories. namely the symmetric ordering. the integrals over p j are to be performed before the x-integration. can be done at least in the framework of perturbation theory. For field theories considered in this book this. ¯ The integral is then dominated by trajectories close to the classical one xc (t). for example. A lagrangian of this kind appears. whenever there is an ambiguity.12) Equivalently.13) ˙ ˙ 2 2 subject to a constraint equation y = c(x).11). The modification of the functional measure Dx into Dx = lim comes from the integral 1 = Nj d p j exp − i −1 f (x j ) p 2 ε j 2h ¯ i→∞ dxi f 1/2 (xi ) i which now depends on x j .12) (see Problem 2. is unambiguously defined by the path integral ansatz in configuration space. if of interest. for a system of two interacting particles m m L = x 2 + y 2 − V (x. In summary.7) with the c-number hamiltonian corresponding to lagrangian (2. which satisfies δS[xc (τ )] = 0: i. Therefore L= m 2 1+ dc dx 2 x 2 − V (x. t ∼ lim j→∞ tj dx(t j ) 1/2 i S[x(t j )] f (x(t j )) exp 2π h h ¯ ¯ (2.e. Integrating over p j one derives (2.11) are of physical interest.11). must be verified case by case. Their correspondence to canonically quantized theories. quantum theory. for instance. insensitive to the ordering of P and X .1). y) (2. .74 2 Path integrals On the other hand. t |x. [ f (x) p]S = 1 [ f (x) p + p f (x)].10) by use of the saddle point approximation. c(x)) ˙ (2. in the following we shall consider quantum theories defined by the path integral formulation. The functional integral (2. It can be shown (Cheng 1972) that the appropriately generalized path integral reads x .12) defines the same dynamics as the Schr¨ dinger equation with one particular ordering of the operators P and o X . which in the classical limit gives the theory (2. supplemented by the rule that.

tn · · · x1 .16) Proceeding exactly as when deriving (2. Also. x) + h J (τ )x(τ )] dτ ˙ ¯ (2. t determines all transition probabilities between quantum mechanical states. . t |X (t1 )X (t2 )|x. t |X (τ1 )X (τ2 )|x. t |xn .15) for the product of two operators: X (τ1 )X (τ2 ) at τ1 > τ2 .6) we obtain (2.h. x)] dτ ˙ × exp h t ¯ (2. t (2.15). τ2 = ti2 and apply the completeness relation at each ti . t . Again we divide the time axis into small intervals. t . the r. . t . X (t N )|x. t = Dx D p x(t1 ) . tn in such a way that τ1 = ti1 . In view of further applications of functional formalism to quantum field theories it is also important to know the path integral representation of the matrix elements of the position operators. .16) is true for τ1 > τ2 . x(t N ) 2π h ¯ i t [ p x − H ( p. of (2. choosing t1 . corresponding to the field operator of quantum field theory.15). t |xn . t |x.15) Let us check (2. t1 < t 2 (2. t |x. . Note that (2. t |X (t2 )X (t1 )|x. t = i dxi xi1 xi2 x . t1 |x.2.18) .1 Path integrals in quantum mechanics 75 Matrix elements of position operators The matrix element x . t |T X (t1 ) . ti2 |X (τ2 )|xi2 −1 . let us note that the transition amplitude in the presence of an external source J (τ ) x .17) As before. . t |X (τ2 )X (τ1 )|x. We obtain x . t J = i Dx D p exp h 2π h ¯ ¯ t t [ p x − H ( p. ti1 |X (τ1 )|xi1 −1 .16) corresponds to the matrix element x . . When τ1 < τ2 . t1 |x. defines the matrix element of the time-ordered product of the position operators Dx D p i t x(t1 ) x(t2 ) exp [ p x − H ] dτ ˙ h t 2π h ¯ ¯ x . it is also possible to change from phase space path integrals to path integrals over configuration space. ti1 −1 · · · × xi2 . For the time-ordered product of N such operators the following expression can be shown to hold x .s. like (2. t1 > t 2 = x . ti2 −1 · · · x1 . t = i dxi x . tn · · · xi1 . Therefore the path integral.

. t) = exp[−(i/h )E 0 t] x|0 = x| exp[−(i/h )H t]|0 = x. can be used as a generating functional of the ¯ matrix elements of the position operators. The energy eigenstates correspond to the wave-functions n (x) = x|n . . . It will be convenient to use 0 (x.21) If the This corresponds to the usual rule of differentiating the monomials. in the case of the functional F[J ] defined as F[J ] = dq1 . Assume that the lagrangian L of the system is time-independent. . qn ) J (q1 ) . δ J (qn ) ∞ n=1 1 n! dq1 . qn ) = δ n [J ] δ J (q1 ) .23) J =0 which compares with the Taylor expansion of the usual functions. 2. . qn ) J (q1 ) . . dqn−1 J (q1 ) . the ground state or the vacuum is described by the function 0 (x) = x|0 . In particular. . t |x. .20) the functional derivative with respect to J is simply ( f can be chosen to be symmetric in all variables) δ F[J ] = δ J (q) dq1 . t|0 ¯ ¯ (2. J (qn−1 ) n f (q1 . . .76 2 Path integrals corresponding to the usual transition amplitude with the hamiltonian modified by a source term: H → H − h J x. for our purpose it is sufficient to know that. . . δ J (t N ) J =0 x . . . . .19) Instead of a formal definition of the functional derivative. t |T X (t1 ) . . Let us first consider the analogous problem in quantum mechanics. . . . qn−1 q) (2. . J (qn ) (2. functional is defined instead by the series [J ] = then φn (q1 .2 Vacuum-to-vacuum transitions and the imaginary time formalism General discussion In field theoretical applications we shall mainly deal with the matrix elements of the field operators taken between the vacuum states: the Green’s functions.24) . which are then given by its functional derivatives with respect to J (τ ) x . t) defined as 0 (x. t = 1 i N δN δ J (t1 ) . X (t N )|x. t J (2. . . dqn f (q1 . J (qn ) (2. . . . dqn φn (q1 .22) (2.

t J W [J ] J =0 (2. . X (t N )|0 = where W [J ] = 0|0 J 1 i N δN δ J (t1 ) . T1 J at any given x1 . t = ¯ ∗ 0 (x . T2 |x . Then we can write x2 . . in fact. t |x. t J x.t ) 0 (x 2 ) . x2 (for instance x1 = x2 = 0) provided that the analytic continuation to the imaginary values of T1 and T2 is performed. t the functional form (2.29) − T1 )] and similarly for x2 . t |T X (t1 ) .26) = dx dx ∗ 0 (x 0 (x. t) (2.28) let us choose that the source J (t) vanishes outside the time interval (t. t) ∗ 0 (x 1 ) ∗ 0 (x 1 ) and. Recalling ¯ that E 0 is the lowest energy eigenvalue we get T1 →+i∞ lim exp[−(i/h )E 0 T1 ] x. T1 (2. t |T X (t1 ) . .18). . . We shall show that W [J ] = exp[(i/h )E 0 (T2 − T1 )] ¯ x2 . t ) x . determined by the transition amplitude x2 . . t) (2.27) with x . X (tn )|x. t ) x . t .28) This implies that W [J ] is. t |x. t|x1 . . . It is very important that the generating functional W [J ] can also be derived in another way. t ) with T2 > t > t > T1 . T1 = x| exp[−(i/h )H (t − T1 )]|x1 ¯ = n n (x) ∗ ¯ n (x 1 ) exp[−(i/h )E n (t J = dx dx x2 . T2 |x1 . t J given by (2. T1 = ¯ = ¯ 0 (x) exp[−(i/h )E 0 t] 0 (x. . T2 |x1 . It reads 0|T X (t1 ) . To derive (2.25) and for the matrix element x . t |x.2. T2 |x1 . T1 where x. t|x1 . T1 ∗ (x1 ) 0 (x2 ) T1 →+i∞ 0 lim T2 →−i∞ J (2. t 0 (x. T2 |x .2 Vacuum-to-vacuum transitions 77 We are interested in the matrix element 0|T X (t1 ) . X (tn )|0 . The considered vacuum expectation value can also be obtained from a generating functional 0|T X (t1 ) . The only T -dependent terms are now the factors exp[−(i/h )E n (t − T1 )] and we can continue to T1 → +i∞ explicitly. . . in the same way T2 →−i∞ lim exp[(i/h )E 0 T2 ] x2 . T2 |x . X (tn )|x. X (tn )|0 = dx dx ∗ 0 (x .15) can be used. t|x1 . δ J (t N ) . t x . .

30) T2 →−i∞ × x2 . including the source term. Notice also that (2. . given by (2. can be written as follows S J [x] = T2 T1 [L + h J x] dt = 1 x x ˙ ¯ 2 T2 T1 + 1 (x.31) J ≡0 Consequently.28) follows. The J -independent factors are irrelevant. .28) we can write W [J ] = T1 →+i∞ T2 →−i∞ lim x(T1 )=x1 x(T2 )=x2 Dx exp (i/h ) ¯ [L(x.33) where ( f. x) ¯ 2 (2. . t ) = − d2 dt 2 + ω2 δ(t − t) (2.26) imply 0|T X (t1 ) .28). T2 |T X (t1 ) . .32) where x1 . T1 We conclude that the vacuum matrix elements can be calculated by taking functional derivatives of the generating functional W [J ]. t ) x(t ) = T2 T1 −x x − ω2 x 2 dt ¨ (2. instead of (2. g) = T2 T1 dt f (t) g(t) and the operator A is defined by the equation T2 T1 dt dt x(t) A(t.34) so that A(t. X (t N )|0 = 0|0 1 i N 1 δN W [0] δ J (t1 ) .78 2 Path integrals After employing (2. . X (t N )|0 = exp[(i/h )E 0 (T2 − T1 )] ¯ ∗ T1 →+i∞ 0 (x 1 ) 0 (x 2 ) lim (2. .29).35) .27) and (2. x2 . because we can always consider quantities like 1 0|T X (t1 ) . . x) + h J x] dt ˙ ¯ (2. Harmonic oscillator As an example of particular interest in view of further field theoretical applications we shall consider the case of the harmonic oscillator L = 1 x 2 − 1 ω2 x 2 = 1 (d/dt)(x x) − 1 x x − 1 ω2 x 2 ˙ ˙ ¨ 2 2 2 2 2 The action integral. . g) denotes the ‘scalar product’ ( f. (2. Ax) + (h J.28) and (2. δ J (t N ) T2 T1 W [J ] (2. X (t N )|x1 . are arbitrary.

36). x0 ) + 1 h 2 (J.39) z(T1 )=0 z(T2 )=0 The important observation is that if the Green’s function (2.2. G J ) exp (i/2h ) xc xc ˙ ¯ ¯ ¯ Dz exp (i/h ) ¯ T2 T2 T1 + h (J. t→±i∞ dG(t)/dt −→ 0 t→±i∞ (2. .37) so that G = −A−1 . z) − z xc ˙ T2 T1 (2.40) † In the space of functions x(t) with zero modes x0 (t) of A excluded. Az) + 1 h (J.2 Vacuum-to-vacuum transitions 79 and Ax ≡ T2 T1 dt A(t. The resulting W [J ] is of the following form W [J ] = T1 →+i∞ T2 →−i∞ lim exp (i/2h )h 2 (J.32) by introducing z(t) x(t) = xc (t) + z(t) where xc (t) is a solution of the classical equation of motion (in the presence of the source) x + ω2 x = h J ¨ ¯ Requiring that xc (t) satisfies the boundary conditions xc (T1 ) = x1 . we have z(T1 ) = z(T2 ) = 0. xc (T2 ) = x2 . x0 ) ¯ × L(z) dt T1 (2. and the special solution of the complete equation xc (t) = x0 (t) + h ¯ dt G(t − t ) J (t ) (2.36) Here G(t − t ) is the Green’s function of the classical equation of motion d2 dt 2 + ω2 G(t − t ) = δ(t − t ) (2. ¯ together with (2.37) is chosen to satisfy the conditions G(t) −→ 0.33) in terms of the classical trajectory and the new integration variable z(t) we get S J [xc + z] = 1 xc xc ˙ 2 T2 T1 + 1 (z. The classical trajectory xc (t) can be written as a superposition of a solution x0 (t) of the homogeneous equation. t ) x(t ) Let us now change the functional integration variable in (2.† Re-expressing the action integral (2. Az) = (Axc .38) where we have used the relations Axc = −h J. G J ) 2 2¯ 2¯ (xc .


2 Path integrals

the only functional dependence of W [J ] on J is given by the factor W [J ] ∼ exp (i/2h )h 2 (J, G J ) ¯ ¯


To prove this assertion we should show that xc xc T2 and (J, x0 ) are J -independent ˙ 1 whenever the boundary conditions (2.40) are satisfied. In fact (2.36) and (2.40) imply xc xc ˙
T2 T −→ x x 2 ˙ T1 T →+i∞ 0 0 T1 1 T2 →−i∞

with x0 obviously J -independent. For (J, x0 ) the argument is somewhat more involved. Consider the solution of the homogeneous equation x0 (t) = A exp(iωt) + B exp(−iωt) with A, B chosen to satisfy the boundary conditions x0 (T1 ) = x1 , x0 (T2 ) = x2 x2 exp(−iωT1 ) − x1 exp(−iωT2 ) , exp[iω(T2 − T1 )] − exp[−iω(T2 − T1 )] x1 exp(iωT2 ) − x2 exp(iωT1 ) B= exp[iω(T2 − T1 )] − exp[−iω(T2 − T1 )] A= In the limit T1 → +i∞, T2 → −i∞ we have A ≈ x2 exp(−ω|T2 |), B ≈ x1 exp(−ω|T1 |)

implying that (J, x0 ) also vanishes in this limit. Note that in (2.41) the limit T1 → +i∞, T2 → −i∞ has in fact been omitted. This could be done because the integral
T2 T2

T1 T1

dt J (t) G(t − t ) J (t ) ≡ (J, G J )

is independent of the limits of integration due to the requirement that J (t) vanishes outside some finite time interval (t1 , t2 ) such that T2 > t2 > t1 > T1 . The dependence on T1 and T2 can now occur only in the factors which are J -independent because of the boundary conditions (2.40), and therefore irrelevant for our discussion. To complete the case of the harmonic oscillator we must find the Green’s function G(t − t ) consistent with the boundary conditions specified by (2.40). ˜ Introducing the Fourier transform G(ν) G(t) =
∞ −∞

dν ˜ G(ν) exp(−iνt) 2π

2.2 Vacuum-to-vacuum transitions


Fig. 2.1. Possible prescriptions for bypassing the singularity at ν = ±ω in the integral (2.42) in the complex ν plane.

we obtain the formal solution of (2.37) G(t) = −
∞ −∞

1 dν exp(−iνt) 2π ν 2 − ω2


This result represents, in fact, four distinct solutions depending on the prescription for bypassing the singularity at ν = ±ω, which may be one of those shown in Fig. 2.1. Cases (a) and (b) are easily seen to be in conflict with the boundary conditions (2.40). It suffices to perform the analytic continuation of (2.42) by deforming the contour as shown in the picture and then to integrate for imaginary t. The two remaining solutions agree with (2.40): we again continue analytically to imaginary t by deforming the contour of the ν-integration from the real to the imaginary axis. They correspond to adding to the denominator of the integrand in (2.42) (+iε) and (−iε), respectively G(t) = −
∞ −∞

dν 1 exp(−iνt) 2 − ω2 ± iε 2π ν


Adding (+iε) to the Green’s function denominator is equivalent to introducing the extra term ±iεx 2 to the lagrangian function. Only for +iεx 2 is the functional integral well defined, so that the final solution for the Green’s function is specified uniquely ∞ 1 dν G(t) = − exp(−iνt) (2.44) 2π ν 2 − ω2 + iε −∞ After the contour integration this becomes (for real t) i [exp(−iωt) (t) + exp(iωt) (−t)] (2.45) 2ω The physical interpretation of the (+iε) prescription is thus made clear: it corresponds to the positive frequency solutions propagating into the future. G(t) =


2 Path integrals

Fig. 2.2. Rotation of the integration contour from the real to the imaginary axis in the complex ν-plane which defines the analytic continuation of G E (τ ) to τ = it.

Euclidean Green’s functions Now let us observe that the same Green’s function can also be derived in another way, i.e. by the analytic continuation from the imaginary time region. To this end we first define the ‘Euclidean’ (i.e. imaginary time) Green’s function G E (τ ), which obeys the equation − d2 dτ 2 + ω2 G E (τ ) = δ(τ ) Consequently, G E (τ ) =
∞ −∞


dν 1 exp(−iντ ) 2 + ω2 2π ν


As the singularities of the integrand appear at ν = ±iω, away from the integration contour, G E (τ ) is uniquely specified to be G E (τ ) = − 1 exp(−ω|τ |) 2ω (2.48)

in agreement with the boundary conditions specified for G(t). We shall now show that the real time Green’s function G(t) is given by the analytic continuation of G E (τ ) to τ = it G(t) = −iG E (it) (2.49) In order to perform this continuation the integration contour in (2.47) should be deformed to keep the integral convergent during the τ = it; such a rotation is illustrated in Fig. 2.2. The result is G(t) = −
∞ −∞

dν 1 exp(−iνt) 2 − ω2 + iε 2π ν

which coincides with (2.44). It is also possible to continue to τ = −it. In this case, however, the resulting real

2.3 Path integral formulation of quantum field theory


time Green’s function appears with the opposite, (−iε), prescription. We conclude that the analytic continuation is correct for τ = it. One way to see the difference is to make the substitution t = ∓iτ under the functional integral. The action changes as follows: i i S[x] = h h ¯ ¯ dt 1 dx 2 dt

− V (x) −→


i (±i) h ¯

dτ −

1 dx 2 dt


− V (x)

1 (2.50) = ± SE [x] h ¯ where SE [x] is the ‘Euclidean’ action. The resulting functional integral is of the form Dx exp{±(1/h )SE [x]} ¯ and only for the (−) sign is it well defined. For the generating functional WE [J ] one gets WE [J ] = Dx exp −(1/h )SE [x] + ¯ dτ J (τ ) x(τ ) ∼ exp
1 h (J, G E J ) 2¯


where G E (τ ) is defined by (2.46). 2.3 Path integral formulation of quantum field theory Green’s functions as path integrals The results of Sections 2.1 and 2.2 are easily generalized to the case of more than one degree of freedom. If the number of degrees of freedom is to be n, the coordinate x should be replaced by an n-component vector. The functional integral would now correspond to the sum over all trajectories in the n-dimensional configuration space, satisfying appropriate boundary conditions. In the field theory, the trajectory x(t) is replaced by a field function (x, t). The degrees of freedom are now labelled by the continuous index x; the number of degrees of freedom is obviously infinite. To define the appropriate path integral one can start from a multiple integral on a discrete and, for a beginning, finite lattice of space-time points. This amounts to defining the quantum field theory as a limit of a theory with only a finite number of degrees of freedom. The limit of an infinite lattice, related to the thermodynamical limit of statistical mechanics, already defines a theory with an infinite number of degrees of freedom. However, this lattice theory does not have enough space-time invariance and a continuous theory must be defined. The latter limit is accompanied by infinities, the ‘UV divergences’ of quantum field theory. The definition of the functional integral in quantum field theory is thus more ambiguous than in the case of quantum


2 Path integrals

mechanics. Nevertheless, the functional formalism in quantum field theory is of great heuristic value. It is a very convenient tool for studying perturbation theory and allows a natural description of some non-perturbative phenomena. The quantum field theory is usually formulated in terms of the vacuum expectation values of the chronologically ordered products of the field operators, the Green’s functions: G (n) (x1 , . . . , xn ) = 0|T (x1 ) . . . (xn )|0 (2.52)

Using our experience from the previous section we shall write down the path integral representation for them. In particular, it is important to remember the role played by (2.28) and (2.30) in getting rid of the vacuum wave functions originally present in (2.25) as the boundary conditions. Thus, by analogy with the results of the previous section, we postulate the following path integral representation G (n) (x1 , . . . , xn ) ∼ D (x1 ) . . . (xn ) exp (i/h ) ¯ d4 x L (2.53)

(D denotes integration over all functions (x, t) of space and time, because, for each value of x, (x, t) corresponds to a separate degree of freedom; L is the lagrangian density). Eq. (2.30) suggests that any boundary conditions are, in fact, irrelevant here, provided that we take the imaginary time limit. In particular, the trajectories may be not constrained at all. In analogy with (2.30), the rotation of the time contour away from the real axis should start somewhere at large absolute values of time, so that the arguments of the Green’s functions, t1 . . . tn , stay real. If instead the whole time axis is rotated, ti = −iτi , the result is a Euclidean Green’s function (see the end of Section 2.2). The latter has a particularly convenient path integral representation, because the weight factor in the integrand, exp(−SE /h ), is then non-negative. This Euclidean path ¯ integral formalism can be used to define the Minkowski space Green’s functions by an analytic continuation of the Euclidean ones. An expression like (2.53) would then be understood as an analytic continuation in the variables t1 , . . . , tn of the analogous Euclidean formula. Equivalently one can work in Minkowski space and use the (+iε) prescription (see (2.62)). Eq. (2.53) should be regarded as the formulation of the theory. What is the relation between the path integral and the usual (canonical) operator formulation of quantum field theory based on the same lagrangian L? For our purposes these are equivalent if they imply the same perturbation theory Feynman rules. This has to be checked in each case of interest. However, the derivation of the perturbation theory rules is much simpler in the functional framework, particularly in the case of gauge field theories. In the following we shall often use the path integral formulation,

2.3 Path integral formulation of quantum field theory


referring to the operator formalism only if it helps to make the presentation more concise, like in the case of the scattering operator discussed in Section 2.7. It is convenient to normalize the Green’s functions by factorizing out the vacuum amplitude G (n) (x1 , . . . , xn ) = 0|T (x1 ) . . . (xn )|0 / 0|0 =N where 1/N = D exp (i/h ) ¯ d4 x L ∼ 0|0 (2.55) D (x1 ) . . . (xn ) exp (i/h ) ¯ d4 x L (2.54)

This removes the extra factors like those appearing in (2.30). The T -product defined by (2.54) is a covariant quantity (see Section 10.1 on ambiguities of T -products). The derivatives of the Green’s functions are functional integrals of derivatives of fields, e.g.
xG (2)

(x, y) ∼



(x) (y) exp (i/h ) ¯

d4 x L


The Green’s functions (2.54) satisfy the standard equations of motion which can be derived using the invariance of the functional integral under a shift of variables (x) → (x) + ε f (x) (see Problem 2.2): for example, in a scalar massless field theory defined by the lagrangian L = 1 ∂µ ∂ µ 2 we have
xG (2)

− V( )

(x, y) = − 0|T V ( (x)) (y)|0 − iδ(x − y)


We observe also that given a theory formulated in terms of the Green’s functions (2.54) quantities like 0|T x (x) (y)|0 remain a priori undefined. We can define them by requiring that they satisfy certain constraints like, for example, for a scalar massless field theory

0|T (x) (y)|0 = 0|T


(x) (y)|0 − iδ(x − y)

where the l.h.s. is given by (2.56a), which amounts to specifying a regularization prescription for them. The Green’s functions (2.54) are given by the functional derivatives of the functional W [J ] equivalent to the vacuum transition amplitude in the presence of the external source J (x) W [J ] = N D exp (i/h ) ¯ d4 x [L + h J (x) (x)] ¯ (2.57)


2 Path integrals

Expanding in powers of J and using (2.54) we can rewrite W [J ] as follows: W [J ] = Consequently, G
(n) ∞ n=0

in n!

dx1 . . . dxn G (n) (x1 , . . . , xn ) J (x1 ) . . . J (xn )

(x1 , . . . , xn ) =

1 i


δn δ J (x1 ) . . . δ J (xn )

W [J ]
J ≡0


The Green’s functions can also be considered as the analytic continuation of those obtained from the generating functional defined in the Euclidean space with ˆ ˆ x0 = −ix4 , where x4 is real: WE [J ] = N D exp −(1/h )SE [ (x)] + ˆ ¯ d3 x d x 4 J ˆ

where, for example, for a free scalar field theory SE [ (x)] = ˆ
1 2

ˆ d3 x dx4

∂ ∂ x4 ˆ


∇ + (∇ )2 + m 2



The Euclidean Green’s functions are given by G (n) (x1 , . . . , xn ) = ˆ E ˆ δ δ ... δ J (x1 ) ˆ δ J (xn ) ˆ WE [J ]
J ≡0

Using path integral formalism one can derive equations of motion for them (Problem 2.2): for example, in free scalar field theory − ∂2 − ∇ 2 + m 2 G (2) (x, y ) = δ(x4 − y4 )δ(x − y) ˆ ˆ E ˆ ˆ ∂ x4 ˆ2

analogous to those in Minkowski space. The Minkowski space Green’s functions are given by analytic continuation G (n) (x1 , . . . , xn ) = (i)n G (n) (x1 , . . . , xn ) ˆ E ˆ where ˆ x = x, x0 = −ix4 ˆ

Since in Euclidean space the exponent in the generating functional WE [J ] is bounded from above we can evaluate the functional integral by the saddle point method (see Problem 2.8). We recall that the saddle point approximation rests on the fact that the integral I = dx exp[−a(x)]

2.3 Path integral formulation of quantum field theory


can be successfully approximated by I ∼ exp[−a(x0 )] = dx exp − 1 (x − x0 )2 a (x0 ) 2

where x0 satisfies a (x0 ) = 0, if a (x0 ) > 0 and if the points away from the minimum do not contribute much. We remark also that we can define the field operator evolution for imaginary time (x) = exp(iH x0 ) (0, x) exp(−iH x0 ) = exp(H x4 ) (0, x) exp(−H x4 ) ˆ ˆ ˆ and also the operator ordering T with respect to x4 analogously to the T -product ˆ definition, for example, ˆ ˆ T (x) ( y ) = ˆ (x4 − y4 ) (x) ( y ) + ˆ ˆ ˆ ˆ ( y4 − x4 ) ( y ) (x) ˆ ˆ ˆ ˆ

whose vacuum expectation value then satisfies the same equation of motion as G (2) (x, y ) defined by the generating functional in Euclidean space. E ˆ ˆ

Action quadratic in fields In the case in which the classical action is quadratic in the field variable S=
1 2

(x) (2.59)

d4 x d4 y

(x) A(x, y) (y)

the generating functional W0 [J ] (index 0 corresponds to the specific form (2.59) of the action) is easily obtained in a closed form. Repeating the steps that led us to the derivation of formula (2.41) we obtain W0 [J ] = exp i h2 ¯ 2h ¯ d4 x d4 y J (x) G(x, y) J (y) (2.60)

where G(x, y) is the Green’s function of the classical field equation A(x, y) (y) d4 y = −h J (x) ¯ (2.61)

satisfying conditions (2.40). One example of a theory with quadratic action is the theory of a free field, with the lagrangian density L= The extra term 1 iε 2
2 1 2

∂µ ∂ µ

− m2


+ 1 iε 2



has been added in accordance with the discussion following


2 Path integrals

(2.43): it makes the functional integral (2.57) well-defined and simultaneously ensures the correct boundary conditions for the Green’s function G(x, y). As ∂µ ∂ µ = ∂µ ∂µ − ∂µ ∂ µ (2.63)

and neglecting in the action the surface term at infinity, the action can be written in the form S= i.e. A(x, y) = − ∂µ ∂ µ + m 2 − iε δ(x − y) The classical field equation ∂µ ∂ µ + m 2 − iε is then solved by (x) = h ¯ d4 y G(x − y) J (y) (2.67) (x) = h J (x) ¯ (2.66) (2.65)
1 2

d4 x d4 y

(x) −∂µ ∂ µ − m 2 + iε δ(x − y) (y)


with the Green’s function G(x − y) satisfying
µ ∂µ ∂(x) + m 2 − iε G(x − y) = δ(x − y)


˜ Introducing the Fourier transform G(k) G(x − y) = we obtain from (2.68) ˜ G(k) = − and therefore G(x − y) = 1 (2π)4 d4 k k2 1 exp[−ik(x − y)] − m 2 + iε (2.71) 1 k 2 − m 2 + iε (2.70) 1 (2π)4 ˜ d4 k G(k) exp[−ik(x − y)] (2.69)

Again, the Green’s function agrees with the analytic continuation from the Euclidean region.

Gaussian integration We end this section with a comment on gaussian integration. Having defined functional integrals as an appropriate limit of multiple integrals, one encounters

2.3 Path integral formulation of quantum field theory


integrals of the type I =
∞ −∞

dx exp −bεx 2 + ibx 2 = 2

dx exp −bx 2 (ε − i)


with ε → 0, where the ε-term corresponds to the (+iε) prescription. Rotating the contour of integration into x = x exp(iϕ) such that exp(2iϕ)(ε − i) = 1 one easily finds I = exp i 1 π − 1 ε (π/b)1/2 −→(iπ/b)1/2 4 2


which is the analytic continuation of the gaussian integral
∞ −∞

dx exp −ax 2 = (π/a)1/2


for complex a (and Re a > 0). A similar result holds for integration over the complex variable z = x + iy dz ∗ dz exp(−az ∗ z) ≡ 2
∞ ∞

−∞ −∞

dy exp −a x 2 + y 2

= 2π/a


for complex a (and Re a > 0). We write dz ∗ dz for 2 d Re z d Im z by convention. For many degrees of freedom z 1 , . . . , z n we can define the complex vector z   z1  .  z= .  . zn and denote by (z ∗ , Az) the scalar product of vectors z and Az, where A is a (n × n)-dimensional complex matrix. Then we have the following generalization of relation (2.75):
∗ dz 1 dz 1 . . . ∗ dz n dz n exp[i(z ∗ , Az)] = (2π )n in / det A


for any positive-definite matrix A which can be diagonalized by unitary transformations. The result is obvious for a diagonal matrix A (it follows from multiple use of (2.75)). For the general case one has to show that the integration measure is invariant under the unitary transformation diagonalizing A (the scalar product (z ∗ , Az) is obviously invariant). Denote the unitary transformation by U = A+iB, with A and B real matrices. From the unitarity of U : UU † = 1 it follows that A AT + B B T = 1, B AT − AB T = 0 (2.77)


2 Path integrals

The change of integration variables z = U z can be written as x y = A −B B A x y =M x y (2.78)

and using (2.77) we conclude that the transformation matrix M is orthogonal (M M T = 1). Therefore the integration measure is indeed invariant under the unitary transformation U . For integration over n real variables x ≡ (x1 , . . . , xn ), one has dx1 . . . dxn exp[i(x, Ax)] = π n/2 in/2 /[det A]1/2 (2.79)

In the limit n → ∞ integral (2.79) is the J -independent path integral in (2.39). Although we do not have to evaluate it explicitly, it is worth remembering its compact form (2.79). Another point is that in addition to results (2.76) and (2.79) we can also easily calculate integrals in which the product (z ∗ , Az) is replaced by an expression (z ∗ , Az) + (b∗ , z) + (z ∗ , b) ≡ F(z) (2.80)

where b is a constant vector, and matrix A is now hermitean. Indeed, expression (2.80) can be written as F(z) = (ω∗ , Aω) − b∗ , A−1 b (2.81)

where ω = z − z 0 and z 0 = −A−1 b is the minimum of F(z). Actually, (2.81) has been used in the derivation of (2.41) and (2.60) for the generating functional W0 [J ]. Finally, we recall that for any matrix L which can be diagonalized by a unitary transformation det(1 − L) = exp Tr ln(1 − L) where Tr ln(1 − L) = − Tr L + 1 L 2 + 1 L 3 + · · · 2 3 2.4 Introduction to perturbation theory Perturbation theory and the generating functional We shall discuss first a simple case of a scalar field theory described by the lagrangian λ 4 L = 1 ∂µ (x)∂ µ (x) − m 2 2 (x) − (x) (2.82) 2 4! where x denotes four coordinates in Minkowski space. We seek a method for calculating an arbitrary Green’s function in such a theory. For that purpose it is

2.4 Introduction to perturbation theory


convenient to use the functional formulation of the theory in which the Green’s functions are given by functional derivatives of the generating functional W [J ] G (n) (x1 , . . . , xn ) = where W [J ] = N D exp (i/h ) ¯ d4 x [L + h J (x) (x)] ¯ (2.84) 1 i

δ δ ... δ J (x1 ) δ J (xn )

W [J ]
J ≡0


and factor N is defined by (2.55). We recall at this point that, so far, we have been able to calculate exactly the generating functional W0 [J ] for a theory of non-interacting scalar fields with the action S0 given by S0 = d4 x Lfree = d4 x
1 2

∂µ ∂ µ

− m2



or, more generally, containing terms at most quadratic in fields. The analogous, exact, solution is not known for the full theory (2.82) and the method to be used for calculating the Green’s functions (2.83) is a perturbative expansion in terms of powers of SI defined as follows SI = S − S0 where S= d4 x L (2.86)

Such an expansion should be understood as an expansion in a neighbourhood of the vacuum state for which (x) = 0, so that the action SI can be regarded as a small parameter. It is convenient to note the following functional identity D exp i S0 [ ] + SI [ ] + h J (x) (x) d4 x ¯ h ¯ i 1 δ i = exp SI S0 [ ] + h D exp ¯ h h i δJ ¯ ¯

J (x) (x) d4 x (2.87)

so that (2.84) can be rewritten as 1 δ i SI W0 [J ] i 1 δ h i δJ ¯ SI W [J ] = = N exp 1 δ i h i δJ ¯ W0 [J ] exp SI i δJ h ¯ J ≡0 exp

W0 [J ]


The perturbation series is generated by expanding the exponential factor

.83) with W [J ] replaced by W0 [J ].89). the following general formula for the Green’s functions (2. . .89) If SI can be written as an integral over polynomials in fields (as in our example) any Green’s function can be calculated term by term using (2. Each term can be written as a respective derivative of the functional W0 [J ].) It is convenient to convert the integrals in (2. .93) . . The case n = 4 is somewhat more complicated. .l (2. j. . (Notice that all G (n) -functions with n odd equal zero. xn ) = D (x1 ) . .k. For the free propagator 2 one immediately gets G (2) (x1 . x2 ) = −ih G(x1 − x2 ) ¯ 0 (2. x4 ) = 0 δ4 δ J1 δ J2 δ J3 δ J4 i h ¯ 2 2 J =0 1 Ji J j Jk Jl G i j G kl 2! i. . where G(x − y) is the Green’s function for the free classical Klein–Gordon equation G(x − y) = − 1 (2π )4 d4 k exp[−ik(x − y)] k 2 − m 2 + iε (2.83) D G (n) (x1 . Wick’s theorem As an introduction to the detailed exposition of perturbative rules we consider again the case of non-interacting scalar fields and prove Wick’s theorem. therefore.90) up to the overall J -independent normalization factor which cancels out in the definition of the Green’s functions. xn ) is given by (2.90). . (xn ) ∞ N =0 ∞ N =0 1 i SI N! h ¯ N N exp i S0 h ¯ i S0 h ¯ 1 i SI N! h ¯ exp (2. In perturbation theory one gets.92 2 Path integrals exp{(i/h ) SI [(1/i)(δ/δ J )]} in powers of SI and performing the functional differ¯ entiations as indicated. 0 From the previous considerations we know that W0 [J ] = exp 1 ih 2 ¯ d4 x d4 y J (x) G(x − y) J (y) (2. . This is equivalent to expanding exp{(i/h )SI [ ]} under the ¯ path integral. The n-point Green’s function G (n) (x1 .92) (the symmetry G(x1 − x2 ) = G(x2 − x1 ) has been taken into account). .91) Referring again to (2.83) one observes that the n-point Green’s function is given by the 1 n th term of the Taylor expansion of exponent (2.90) into sums over discrete points and write i4 G (4) (x1 . . .

One gets So finally (4) G 0 (x1 . x3 )G (2) (x2 . .2.89) in the first order in SI which reads D (x1 ) . . x3 ) 0 0 (2.93) can be represented diagrammatically by the ‘prototype’ diagram ≡ (−ih G i j ) × (−ih G kl ) ¯ ¯ and the differentiation with respect to J1 . . j.4 Introduction to perturbation theory 93 where Ji = J (xi ). (x4 ) (−iλ) 1 4!h ¯ d4 y 4 (y) exp i S0 h ¯ (2. for example. .95) . k. . Let us consider the four-point Green’s function and discuss the numerator of (2. .94) This is the content of Wick’s theorem (see. 0 2 An example: four-point Green’s function in λ 4 Coming back to our main problem we shall now discuss several examples in order to derive general rules for perturbative calculations in the theory of interacting fields. . G i j = G(xi − y j ) The structure of (2. x4 )G (2) (x2 . Bjorken & Drell (1965)) which states that any n-point free Green’s function can be written as a sum over all possible products of 1 n two-point Green’s function G (2) (xi . x4 ) + G (2) (x1 . x2 )G (2) (x3 . . J4 leads to all possible assignments of subscripts 1–4 to subscripts i. l. x4 ) 0 0 0 0 + G (2) (x1 . . x j ). x4 ) = G (2) (x1 . .

97) now provide all possible assignments of points x1 . It is easy to discuss them in a systematic fashion. x3 .96) comes from the term in the Taylor expansion of (2. for example.5 where the dot represents the integral of a product of two propagators taken at the same space-time points y (notice that the combinatorial factors in front of the second and the third diagrams include different . So we have to remember about 1/N ! present in (2. These factors cancel with the factor 1/(2n n!) present in (2. y2 . 2. To make the combinatorics transparent we have again replaced the integrations in (2. we have an additional combinatorial factor following from the ‘horizontal’ symmetry (for example.90) which contains four propagators 4 1 ih 2 ¯ 1 4! 4 d4 x d4 y J (x) G(x − y) J (y) (2. x2 .89). or. 2. y1 = i) which is 2n for n pairs.) which is n! (again for n pairs). p. Then. 2. (x4 . 2.4 with self-evident classification of different possibilities.96) It is clear that the non-zero contribution to (2. y4 .89) (N = 1 in our examples) and about 1/4! present in the definition of the coupling constant and. x1 = i.94 2 Path integrals Fig.97) and this happens for any term of expansion (2. m. remembering that the limit y1 = y2 = y3 = y4 should be understood) to the points i. we have to take account of all possible choices of pairs of points xi and yi joined by propagators. l. y4 ). y2 ). The differentiations of (2. In the limit y1 = y2 = y3 = y4 = y we get the diagrams shown in Fig. The result is as shown in Fig.3. on the other hand. l) etc. and another factor from the ‘vertical’ symmetry ((x1 . y1 = j and x1 = j. (x3 . y1 ) can be assigned to (i.3. Let us first assume that pairs of points joined by the propagators are specified.97) and can be represented graphically by the prototype diagram in Fig. y3 ).97) by discrete sums. k. using the generating functional W0 [J ]. (−iλ) 1 4!h ¯ d4 y δ δ4 δ ··· δ J (x1 ) δ J (x4 ) δ J (y) W0 [J ] J ≡0 (2. j. j) or (k. x4 and four points y (it is convenient to split them into y1 . o. y3 . y1 ). (x1 . (x2 . n.

each of which can be discussed as follows. for example.7. 2. have been specified. 2. . (The proof of this again requires some attention to the combinatorial factors. They define the connected Green’s functions G (n) . In the following we shall understand the central diagram to be the one in which the pairs of external points. y) G (2) (x2 . We observe that. .4.2. . which are joined by propagators to each of the interaction points. Fig. y) G (2) (x4 . x4 ). for a given Green’s function we shall be interested only in connected diagrams like those in Fig. In this sense we have three different central diagrams. The analytic expression for the first diagram reads (−iλ/h ) ¯ d4 y G (2) (x1 . 2. There are twelve fields under the functional integral.8 which give non-trivial S-matrix elements for G (4) (x1 .89). permutations of x1 . the central diagram can be conn easily read off from the numerator of (2.5.4 Introduction to perturbation theory 95 Fig. In general. . In momentum space this corresponds to assigning to each vertex definite external momenta.) So we may ignore the denominator of (2. . the SI2 term in the numerator is given by the diagrams (a)–( j) in Fig.89) can be represented by a product of two infinite series of amplitudes as shown in Fig. The second bracket contains the sum of the so-called vacuum amplitudes which do not depend on the Green’s function considered and. 2. x3 . for example.89) and simultaneously all the diagrams containing vacuum subgraphs.98) It remains to extend our considerations to terms SIN . These problems are related since G (n) involves in N principle a quotient of terms of arbitrary order. it happens to be just the denominator of (2. x2 . N > 1 and to discuss the role of the denominator in (2. The amplitude of.89). Actually.89). y) G (2) (x3 . one can convince oneself that the numerator of (2. y) 0 0 0 0 (2.6. most importantly. x4 ). 2.

We get 4! × 4! to cancel 1/4! × 4! from the definition of the coupling constant. Fig. y2 by any y j = yi (factor 3). in addition. Two external points must be joined with the interaction point at y and the other two with the interaction point at z. so the ‘prototype’ diagram with six propagators is relevant for combinatorics. In addition two propagators must connect y and z.) Since N = 2 there is . No two external points can be joined by a propagator.96 2 Path integrals Fig. the same for z 4 and z 3 (factor 4 × 3). (The factors from the ‘vertical’ and ‘horizontal’ symmetries cancel as usual. 2.7. there are two combinations for the remaining yl joining z k (factor 2) and finally y can be interchanged with z in connections with external points (factor 2). 2. We can have therefore the situation depicted in Fig. y1 can be replaced by any yi (factor 4).6. 2.9 and.

xn → +∞ .100) so that all momenta are treated as outgoing (due to (+iε)) in propagators. . z) 0 0 1 (−iλ/h )2 ¯ 2 ×G (2) (y. xn ) conn (2. .99) Our examples can be generalized to the following rules. z) G (2) (x3 . an extra factor 1 2 (we will call it the symmetry factor) and the amplitude reads (2) d4 y d4 z G 0 (x1 . Attach (−iλ/h ) to each vertex and G 0 to each line. .89).9. positive 0 frequencies propagate into future xn → +∞ (see (2. xn ) = d4 p 1 d4 pn ··· exp(−i p1 x1 ) · · · exp(−i pn xn ) (2π)4 (2π )4 ˜ ×G (n) ( p1 . 2. In actual physical transitions 0 0 x1 . ¯ Momentum space One is usually interested in Green’s functions in momentum space defined as Fourier transforms of the respective Green’s functions in configuration space. . . . 2. z) 0 0 0 G (n) (x1 . .4 Introduction to perturbation theory 97 Fig. . xk → −∞ 0 0 xk+1 . .2. . . y) G (2) (y. y) G (2) (x2 . . .45)). z) G (2) (x4 . We ˜ define G (n) ( p1 . pn ) (2. For each diagram calculate the combinatorial (2) (symmetry) factor. . . . To calculate to order N in SI draw all connected diagrams consistent with the structure of the numerator in (2.8. Fig. . pn ) by the relation G (n) (x1 . . . .

. .101) For the diagram shown in Fig. p2 ) = (2π )4 δ( p1 + p2 )h ¯ (2.10. . . 2. 2. and therefore the momenta p1 . Observe that each vertex contributes a factor h −1 and ¯ .11 we get from (2. the diagrams as above are assumed to represent amplitudes without the propagators on external lines (Feynman diagrams).98) the momentum space amplitude reads 4 4 (2π)4 δ i=1 pi (−iλ/h ) ¯ i=1 h ¯ pi2 i − m 2 + iε (2. .10. . . . . .98 2 Path integrals Fig.103) The Feynman rules in momentum space are now obvious. k i = k + 1. . Usually. 2. Fig. 2.11. pn defined by (2.102) and for the diagram shown in Fig.100) are related to the physical momenta pi ( pi0 > 0) of the initial and final particles as follows ˜ ˜ ˜ pi = − pi ˜ pi = pi i = 1. .92) and (2.99) 4 1 (−iλ/h )2 ¯ 2 i=1 h ¯ i 2 pi − m 2 + iε d4 k1 d4 k2 (2π )4 δ( p1 + p2 + k1 + k2 ) (2π )4 (2π )4 2 k1 ×(2π)4 δ( p3 + p4 − k1 − k2 )h ¯ i i h 2 + iε ¯ k 2 − m 2 + iε −m 2 (2. n i 2 p1 − m 2 + iε In momentum space we get from (2. represented by expression (2.91) ˜0 G (2) ( p1 .

(2. the factor (2N − 1) . . In the resulting perturbation expansion . An important modification of the lagrangian (2.104) This can be rewritten in a form similar to (2. 2. Again. 1 gives the number of different groupings of 2N momenta in N pairs and (N − 1)! gives the number of their different orderings along the loop. this corresponds to a well-defined situation in momentum space. So we again get the symmetry factor 1 for a Feynman diagram in 2 momentum space. remember that under the path integral the fields are c-numbers and the only way of specifying their order is by introducing appropriate counterterms.82) follows if we specify that all products of the field operators in it are normal-ordered L= 1 2 :∂µ (x)∂ µ (x): − 1 m 2 : 2 2 (x): − (λ/4!) : 4 (x): (2. but with a number of counterterms. 1×(N −1)! different graphs in momentum space.106) so that : (x): = (x) − 6G (2) (x.12.7): 2 4 (x) = : (x) = : 4 2 4 (x): + const.105) 2 6G (2) (x.12. For N vertices there are (2N −1)(2N −3) .2. 2.82). .4 Introduction to perturbation theory 99 Fig. assume that the pairs of external points which are joined to each vertex as well as the ordering of vertices along the loop are specified. x) 0 2 (x) − const. Thus a Feynman diagram with L loops is of the ¯ order O h L−1 . (2. (x): + 4 (2. because L = (number of internal lines − number of vertices + 1). . with momenta specified along all the internal lines. for example. From Wick’s theorem it follows that (see Section 2. x) : 0 (x): + const.107) and the other counterterms are constants. We have the factor (1/N !)(1/4!) N from the general formula (2. as in Fig. each internal line gives a factor h . .89) and an extension of our discussion of the N = 2 case gives (3 × 4) N ×2 N −1 × N !. ¯ It is easy to extend these considerations to a loop with N vertices.

112) . 0|T (x1 ) ¯ (x2 )|0 = − 0|T ¯ (x2 ) (x1 )|0 (2.53). 2. or functions. 0 Note. One defines the left and right derivatives by the equation d d η=η =1 dη dη Therefore. 2. that the higher order corrections to this diagram.108) the fermion classical fields in the path integral must be taken to be anticommuting quantities.5 Path integrals for fermions. η} = 0 Any function f (η) can be written as f (η) = f 0 + η f 1 (2. Let us take η to be a Grassmann variable.13. There follows a brief introduction to the subject of Grassmann algebra. The constant counterterms cancel some vacuum diagrams. The anticommuting c-number variables.110) or η2 = 0 (2.13. are said to form Grassmann algebra. Grassmann algebra Anticommuting c-numbers The path integral quantization method involves classical fields only and the theory is formulated directly in terms of Green’s functions given by path integrals like (2.111) (2. d d f (η) = f 1 = − f (η) dη dη ← ← (2. 2. however. the diagram is now cancelled by the −6G (2) (x. then f 0 and f 1 are ordinary and Grassmann numbers. for example. when f 1 is a Grassmann number. Since for Fermi fields we must have. like that shown in Fig.100 2 Path integrals Fig. {η. respectively. are not cancelled.109) For instance if f (η) is taken to be an ordinary number. x) 2 (x) counterterm.

115) d f (η) = f 1 dη (2.116) i.114) Consider now the change of integration variable η → η = a + bη. dηn dηn exp(η∗ . It can be shown that instead of (2.117) follows immediately from the integration rules for Grassmann variables. One gets dη f (η) = dη dη dη −1 f (η(η )) (2.111) we see that the integrals and the left derivatives are identical dη f (η) = The integral of the derivative vanishes: dη d d2 f (η) = 2 f (η) = 0 dη dη (2. . (2. The corresponding formula also holds for an integral over real variables (see Problem 2. we define the integration by the requirement of linearity and the following relations dη 1 = 0. j ηi∗ Ai j η j Eq.113) applies when dη and η are next to each other. In the following chapter the Gaussian integral for complex Grassmann variables will be shown to be very useful. Comparing (2. All the rules can be easily generalized to the case of n real Grassmann variables. Aη) = det A (2.113) with (2. . holds for an integral over anticommuting numbers as well. Grassmann algebra 101 Furthermore. dη η = 1 (2. We also add a specification that rule (2.5 Path integrals for fermions. For a complex Grassmann variable the real and imaginary parts can be replaced by η and η∗ as independent generators of Grassmann algebra.76) we now have ∗ ∗ dη1 dη1 .113) The first relation follows from the requirement that the property of a convergent integral over commuting numbers ∞ −∞ dx f (x) = ∞ −∞ dx f (x + a) valid for any finite a. Aη) = i.4).e. the standard Jacobian appears inverted. The second relation is the normalization convention.2.117) where (η∗ .117) is valid for an arbitrary A. . For a diagonal A the proof of (2.

Dirac propagator The classical Dirac fields (x) and ¯ (x) are taken to be elements of an infinite-dimensional Grassmann algebra.117) we get det M = If we shift z and z ∗ we get det M = det A det B − C A−1 D 1 A−1 D 0 B − C A−1 D A − D B −1 C B −1 C 0 1 det A − D B −1 C det B To evaluate the integral we make a shift in integration variables These results for the superdeterminant are not surprising if we write M= or M= 1 D 0 B A 0 C 1 Several other properties of the superdeterminant are listed in Problem 2. The superdeterminant of a superspace matrix M can be constructed by calculating a generalized Gaussian integral (det M)−1 = dz dz ∗ ∗ dη dη exp[−(z ∗ . Coordinates in superspace are commuting and anticommuting variables. Bη)] 2π η = η − B −1 C z η∗ = η∗ − z ∗ D B −1 and using (2. Az) − (z ∗ . and C and D are anticommuting. The generating functional W0 [α] for a free-fermion field described by the lagrangian ∂ L = ¯ (i/ − m) . respectively.9. Dη) − (η∗ . Vectors in superspace are decomposed in terms of the variables z and η. z and η. We can define a linear transformation in superspace z η = A C D B z z ≡M η η where matrices A and B are commuting.76) and (2. C z) − (η∗ .102 2 Path integrals Gaussian integrals can be used to define the determinant of a matrix acting in superspace.

α] = exp ih ¯ ¯ where (i/ − m)S(x − y) = −1 ∂ lδ(x − y) Therefore S(x − y) = − d4 k /+m k exp[−ik(x − y)] 1 l 4 k 2 − m 2 + iε (2π) D¯ D (x) ¯ (y) d4 x L (x) (2. −A−1 β (2. α] as follows: ¯ W0 [α. α] δ α(x) ¯ δα(y) ← (2.5 Path integrals for fermions.81) but for Grassmann fields Dη Dη∗ exp[i(η∗ . β)] = det(iA) exp i β ∗ .125) 1 i 2 d4 x d4 y α(x) S(x − y) α(y) ¯ (2. Using a formula analogous to ¯ (2.123) reads SF (x − y) = and using (2.126) (2. Grassmann algebra 103 is W0 [α. Aη) + i(β ∗ .122) The Feynman propagator defined as SF (x − y) = 0|T (x) ¯ (y)|0 / 0|0 = × exp (i/h ) ¯ d4 x L (x) D D ¯ exp (i/h ) ¯ δ δ ¯ W0 [α. α] = N D ¯ D ¯ exp (i/h ) d4 x [L (x) + h α(x) (x) + h ¯ (x)α(x)] ¯ ¯¯ ¯ (2.121) (2.120) one gets SF (x − y) = −ih S(x − y) ¯ Imagine now a theory of interacting fermions with the action S= d4 x L (x) + d4 x ¯ (x) V (x) (x) (2.118) where N −1 = 0|0 = W0 [0.120) (2.119) we can rewrite W0 [α. 0] and α(x) and α(x) are sources of fermion fields.2. η) + i(η∗ .124) α=α=0 ¯ = S0 + SI .

From (2. . dxk ¯ (x1 ) V (x1 ) (x1 ) ¯ (x2 ) V (x2 ) (x2 ) . Following Section 2.4 we can develop perturbation theory with the interaction term as the perturbation. V (xk ) (xk ) ¯ (x1 ) This reordering introduces the familiar minus sign for a closed fermion loop.14. The path integral with insertion (2. the vacuumto-vacuum amplitude (the denominator in (2. for example. In the expansion in powers of SI of.129) .128) we must. . V (xk )SF (xk − x1 ) (2.120) and (2. Before making the substitution ¯ 1 δ . α] ¯ given by (2. .127) which correspond to fermion loops like that shown in Fig. . 2. . however.128) we obtain for the closed fermion loop the following result (−)V (x1 )SF (x1 − x2 )V (x2 )SF (x2 − x3 ) . i δ α(xi ) ¯ → (xi ) → ¯ (xi ) → 1 δ i δα(xi ) ← (2. . rearrange the fields as follows V (x1 ) (x1 ) ¯ (x2 )V (x2 ) (x2 ) ¯ (x3 ) .14. ¯ (xk ) V (xk ) (xk ) (2. . 2.120) with respect to α and α. .104 2 Path integrals Fig.89)) we find under the path integral the following products of the and ¯ fields: dx1 .127) can be calculated explicitly by differentiating W0 [α.

. . (xn ) exp{(i/h )S[ ]} ¯ D exp{(i/h )S[ ]} ¯ −1 (2.6 Generating functionals 105 2. . are given by the Feynman diagrams which are one-particle-irreducible. . . including disconnected ones. . . . they remain connected after an arbitrary internal line is cut. . i.130) (2.6 Generating functionals for Green’s functions and proper vertices. .131) n=1 [ ]= ∞ n=1 1 n! d 4 x 1 . . . xn ) J (x1 ) . d4 xn G (n) (x1 .54) G (n) (x1 . . .54) with appropriate generalization to the case of other fields. (xn )|0 / 0|0 = × D (x1 ) . . . These are defined as follows W [J ] = ∞ Z [J ] = h ¯ and n=0 ∞ in n! in−1 n! d4 x1 . . xn ) J (x1 ) .e. . . in this section we shall always consider the case of a single field. . J (xn ) conn (2. . . with the exception of the vacuum diagrams which are cancelled by the normalization factor. . . d4 xn G (n) (x1 . . (xn ) (2. . . also called the oneparticle-irreducible (1PI) Green’s functions. xn ) = 0|T (x1 ) . . The Feynman diagrams contributing to G (n) are all connected. To each type of the Green’s functions we assign the corresponding generating functional. . . for simplicity. . In the lowest order (no loops) the connected proper vertex functions coincide when appropriately normalized with the vertices of the original lagrangian.2. J (xn ) d4 x1 .132) (x). . effective potential Classification of Green’s functions and generating functionals There are three basic sorts of Green’s functions. . conn The connected proper vertex functions (n) (x1 . . The connected Green’s functions G (n) (x1 . . xn ) (x1 ) . xn ) are defined as the connected conn part of G (n) . . . Observe that in the generating functional [ ] we denote the arguments as . . . . The full Green’s functions are defined by (2. xn ). A connected Green’s function is obtained by disregarding all terms which factorize into two or more functions with no overlapping arguments. d4 x n (n) (x1 . For a Fourier transform this implies that no subset of n external momenta is conserved separately. In perturbation theory the full Green’s functions G (n) are given by the sum of all diagrams with n external legs.

. .130)–(2. CL (x) = 0| (x)|0 1 δ Z [J ] = h δ J (x) 0|0 J ¯ J (2. xn ) = conn and (n) 1 1 h i ¯ δ n Z [J ] δ J (x1 ) . .137) W [J ] = exp{(i/h )Z [J ]} ¯ and [ CL ] = Z [J ] − h ¯ CL (x) d4 x J (x) CL (x) (2.139) By differentiating (2.135) ≡0 The generating functional of the full Green’s functions has been studied in Sections 2. .138) where the ‘classical’ field combination is defined as the field which minimizes the d4 x J (x) (x) [ ]+h ¯ or by the equation δ [ ] δ (x) δ Z [J ] = δ J (x) = = −h J (x) ¯ CL (2. .139). xn ) = δn [ ] δ (x1 ) . . . xn ) = 1 i n δ n W [J ] δ J (x1 ) . . . .136) We want to show that the functionals Z [J ] and relations: [ ] are given by the following (2. .106 2 Path integrals not J (x): this is convenient because of the relationship between [ ] and the action integral S[ ] which we shall discuss below. . . δ J (xn ) n−1 (2.133) J ≡0 G (n) (x1 . δ (xn ) (2. . It is given by the path integral formula W [J ] = D exp i S[ ] + h ¯ h ¯ d4 x J (x) (x) D exp i S[ ] h ¯ (2. .140) or.5. It follows from definitions (2.138) with respect to J (x) we also find d4 y δ [ CL ] δ CL (y) +h ¯ δ CL (y) δ J (x) CL (x) +h ¯ d4 y δ CL (y) δ J (x) J (y) (2. . with (2.4 and 2. . . δ J (xn ) (2.141) .134) J ≡0 (x1 .132) that G (n) (x1 .

141). Effective action Consider the classical field equation (we take the case of the λ definiteness) ∂µ ∂ µ + m 2 δ δ (do not confuse c c 4 theory for (2.138) can be written as a functional of CL (x) which determines [ CL ].132).141) to determine J (x) in terms of CL (x) so that the r.143) CL ≡0 because when J ≡ 0. Eq. for example. for the time being.142) The case of a non-zero field vacuum expectation value reflecting spontaneous symmetry breaking will be discussed later in this section. (2.s. Here we shall use another argument which allows a simple understanding of the relationship between [ ] and S[ ].h.139) expresses J (x) in terms of CL (x) and in this sense it is the inverse of (2. ¯ λ>0 Eq.6 Generating functionals 107 Eq. of (2. In the zeroth order in λ we obtain (0) c (x) = d4 y [−ih G(x − y)]iJ (y) ¯ (2. CL takes the value 0 and vice versa.130)–(2. we assume to be zero: δ Z [J ] δ J (x) J ≡0 =0 (2. together with (2.138) is the Legendre transform.144) can be solved perturbatively.145) and CL ). It remains to prove that (2.138) hold for the generating functionals defined by (2.144) follows from the action principle S[ ] + h ¯ c = d4 x J (x) (x) = 0 (2.2.137) and (2. In particular. (2. (2. This can be checked by induction (see. The value of (2.141) and (2.68) .142). (2. it implies that δ [ CL ] δ CL (x) =0 (2. Abers & Lee (1973)). where ∂µ ∂ µ + m 2 − (λ/4!) 4 S[ ] = d4 x − 1 2 Eq.68) ∂µ ∂ µ + m 2 G(x − y) = δ(x − y) (2.141) when the external source is turned off (J (x) ≡ 0) is the vacuum expectation value of the field (x) which. If Z [J ] is known we can use (2.146) where G(x − y) is defined by (2.144) = −(λ/3!) 3 c + h J (x).

138) and (2. if we assume (2. given by (2.s. . 2. δ c (xn )| c =0 generates proper vertex functions in the tree approximation). we . (the iε term should always be remembered). This corresponds to replacing (x) in the integrand of (2. in the classical approximation Z c [J ] reads Z c [J ] = S[ c] +h ¯ d4 x J (x) c (x) (2. in the classical approximation.149) and the expansion (2. Moreover. To obtain the first order correction we insert (0) in the r. Repeating the iterations we obtain the perturbation series for the classical field c (x).108 2 Path integrals Fig.136). Only the numerator then contributes because J ≡ 0 in the denominator.147) which can be represented diagrammatically as shown in Fig. It is easy to see that only tree diagrams appear and that they are all connected.137).147) for the c (x). Consider now the generating functional W [J ].149) and it is indeed given by the connected tree diagrams. . again make in the integrand the ‘classical’ approximation (x) → CL (x) with CL (x) defined by (2. from (2.136) by the classical solution c (x). If in (2.136) we replace the action S[ ] by the functional [ ] and instead of doing the path integration exactly. δ J (xn )| J ≡0 using (2. When the loop corrections are taken into account the connected Green’s functions as well as the generating functional Z [J ] can still be understood as given by the tree diagrams but with the elementary vertices replaced by 1PI Green’s functions. and we get W [J ] = exp (i/h ) S[ ¯ c] +h ¯ d4 x J (x) c (x) (2.141).148) Thus. of (2.h. 2.15.139). . This can be seen by calculating δ n Z /δ J (x1 ) . .144). The result reads c (1) c (x) = (0) c (x) 3 + d4 y [−ih G(x − y)](−iλ/h 3!) ¯ ¯ d4 z [−iG(y − z)]iJ (z) (2. in the classical approximation [ CL ] = S[ c ] so in this approximation [ c ] as given by (2.15.138) is indeed the generating function for the 1PI Green’s functions (because δ n S[ c ]/δ c (x1 ) .

i 1PI functions given directly in terms of Feynman diagrams.155) CL =v . Also (2. for example. . . (n) (2) (x − z) G (2) (z − x) = ih δ(x − y) ¯ conn (2.139). ¯ (xn )|0 (2. We conclude that to all orders W [J ] = exp (i/h ) ¯ [ CL ] +h ¯ d4 x J (x) CL (x) (2. . (n) (x1 . .139) and setting CL (x) = 0 one gets d4 z or. strictly speaking. It is worth observing explicitly that differentiating (2. in momentum space. However. from translational invariance. In this sense [ ] can be treated as the effective action including the loop corrections.152) s are the Spontaneous symmetry breaking and effective action The formalism of the effective action can be extended to the case of a nonvanishing field vacuum expectation value which corresponds to spontaneous symmetry breaking (the latter is discussed in detail in Chapter 9). using (2. . so that δ [ CL ] δ CL (x) =0 (2. . This is because all loop corrections are already included in [ ]. that higher derivatives of Z [J ] at J ≡ 0 are connected Green’s functions of the field ¯ = − v whose vacuum expectation value vanishes: 1 1 h i ¯ n−1 δ n Z [J ] δ J (x1 ) . Z [J ] given by (2.137).141) with respect to CL (y). respectively. by induction. . . ˜ ˜ (2) ( p)G (2) ( p) = ih ¯ (we draw attention to the factor i).154) The effective action is defined by (2. It can be proved.6 Generating functionals 109 shall obtain the exact result. The functionals W [J ] and Z [J ] are taken as given by (2. xn )s are n-point non-local vertices of this effective action.150) and therefore.138) holds to all orders.151) (2. δ J (xn ) J ≡0 = 0|T ¯ (x1 ) .2. we now suppose that (2. as [ CL ] and CL are both given by connected diagrams.138) and (2.153) J ≡0 where v = const. For higher n.136) and (2.142) is replaced by 0| (x)|0 1 δ Z [J ] = h δ J (x) 0|0 ¯ =v (2.137) is indeed connected (the connected diagrams exponentiate).

. . . x n ) CL (x n ) ×( CL (x 1 ) − v) (2. . . for example. . ( (n) ¯ = −v (x 1 . . .153) because when J ≡ 0. CL takes the value v and vice versa. The coefficients of such an expansion δ n [ CL ] δ CL (x1 ) . . . In the limit J ≡ J0 and CL ≡ 0 we get the desired result. d 4 x n − v) . by repeating the inductive arguments leading to (2. . x n ) n f (x)|x=a = am d m! dx m=0 ∞ n+m f (x)|x=0 = ∞ (1/m!)v m (n+m) d4 xn+1 .139). .141) and (2.110 2 Path integrals This follows from (2. . xn ) (2. xn . where δ Z [J ] δ J (x) =0 (2. δ CL (xn ) and [ CL ] = ∞ n=1 CL =v = (n) ¯ = −v (x 1 . d4 xn+m (2.138) generates 1PI vertices for the field ¯ = − v δ n [ CL ] δ CL (x1 ) . . . . |0 conn . This can be proved. . . .139). . . . δ CL (xn ) = (n) CL ≡0 (x1 .156) 1 n! d4 x 1 .157) It is important to observe that even for 0| (x)|0 = v = 0 the effective action [ CL ] can also be expanded in terms of the original fields (x). Also. that [ CL ] given by (2. . . . . (2. . .160) m=0 × (x1 .158) are the 1PI Green’s functions for the original fields (x). xn+m ) . . The relation between the Green’s functions (n) −v and (n) is given by the ¯= analogue of the relation d dx Thus we have (n) ¯ = −v (x 1 . . .156) but for CL = 0 and correspondingly for J (x) = J0 . . . we can prove.159) J ≡J0 One can first check by induction that higher derivatives of Z [J ] taken for J ≡ J0 give connected Green’s functions 0|T . . CL (x n ) in terms of [ CL ] defined by (2. . for example. . .132). see (2. Then we differentiate the appropriate number of times the relation δ Z [J ]/δ J (x) = CL (x) with respect to n CL and express the derivatives δ J (x)/δ CL (x 1 ) . .138) and (2. x n ) (2. by induction.

163) which is more useful than the original one because the effective potential V (ϕ) can be calculated perturbatively in a systematic way. pn ) = ¯ ˜ (n) ∞ (1/m!)v m ˜ (n+m) ( p1 . 0. .132) for the [ ] we get V (ϕ) = − ∞ n=2 (1/n!) ˜ (n) (0. . . . . xn ) Eq. . . . These implications can be discussed more conveniently if we introduce the so-called effective potential V (ϕ). .2. . .161) m=0 where ( p1 . . pn ) is the Fourier transform of the proper vertex (n) (x1 . . (In the tree approximation V is just the ordinary potential. .155) now translates into ϕ=v =0 (2. 0)ϕ . pn )(2π)4 δ( p1 + · · · + pn ) = n d4 x exp(i pi xi ) i=1 (n) (x1 . but it makes use of the . . . . the negative sum of all non-derivative terms in the lagrangian density. . .6 Generating functionals 111 or.161) expresses the 1PI Green’s functions in a theory with spontaneously broken symmetry in terms of the 1PI Green’s functions calculated as in the symmetric mode. . Effective potential Eq.] = −(2π )4 δ(0)V (ϕ) dV dϕ (2. Condition (2. (2. . i.164) so that the nth derivative of V at ϕ = 0 is the sum of all 1PI diagrams for the original fields of the lagrangian. in momentum space. This method of calculation is equivalent to computing directly the derivative of V and demanding that it vanishes. by demanding that all tadpole diagrams sum to zero. . . . xn ) ˜ (n) ( p1 . . . pn .155) has very important implications in studies of spontaneous symmetry breaking: it tells us that the field vacuum expectation value is the value of CL (x) which extremizes [ CL ]. without computing V first. (2. ˜ (n) ( p1 . develop the appropriate Feynman rules and fix v directly from (2.162) (V (ϕ) is a function. . .155) and (2.) Alternatively.e. one can rewrite the lagrangian of the theory in terms of shifted fields ¯ = − v considering v as an extra free parameter. 0. . .156). . Using representation (2. It is defined by the equation [ (x) = ϕ = const. ϕ (2. not a functional). with n vanishing external momenta. 0) m (2. . . .

2. . thus we get (1/2n!)[(2n − 2 n 1)(2n − 3) . A sensible way of organizing this double series is the loop expansion. Knowledge of the effective potential means that one knows the structure of spontaneous symmetry breaking.166) exhibits the 2 2 UV divergence typical for perturbative calculations in quantum field theories. It has been seen at the end of Section 2. 1](n − 1)! Feynman diagrams each with the symmetry factor 1 . In principle. using (2. . and we get V0 (ϕ) = + 1 m 2 ϕ 2 + (λ/4!)ϕ 4 2 (2. .16 which gives the following contribution V1 (ϕ) = i ∞ n=1 1 2n 1 d4 k ϕ2 λ 2 (2π )4 k − m 2 + iε 2 n = −1i 2 n 1 λϕ 2 d4 k ln 1 − 2 2 2 (2π )4 k − m + iε (2.164).112 2 Path integrals Fig. 2. formulation of the theory in which the underlying spontaneously broken symmetry is not explicit (see Problem 4. The integral in (2. . we can calculate the effective potential in perturbation theory but the calculation involves a double sum: over all 1PI Green’s functions and for each 1PI Green’s function there is the expansion in powers of the coupling constant.4: each diagram represents [(2n − 1)(2n − 3) . ¯ ¯ The calculation of the effective potential for the λ 4 theory in the one-loop approximation is simple.4 that loop expansion is an expansion in powers of Planck’s constant h : a Feynman diagram ¯ with L loops is O h L−1 . .1). p) = p 2 − m 2 and ˜ (4) = −λ. To the lowest order (tree approximation) the only nonvanishing 1PI Green’s functions are ˜ (2) ( p.165) To the next order (one-loop approximation) we have the infinite series of diagrams shown in Fig. Indeed expansion in powers of h is an expansion in a parameter that multiplies the ¯ total Lagrange density. the loop expansion preserves any symmetry of ¯ the lagrangian and it is unaffected by shifts of fields and by the redefinition of the division of the lagrangian into free and interacting parts associated with such shifts. 1](n − 1)! 1 = (1/2n) 1 .16.166) The factor (1/2n) 1 can be easily understood by recalling the discussion at the 2 end of Section 2. (1/h )L → L corresponds to a change of units into h = 1. Thus.

Formally. −∞) = T exp −i ∞ −∞ dt HI (t) (2. It can be shown (see. t ) = T exp −i We then define the scattering operator to be (2.166) we refer the reader to Section 11. This interpretation of V (ϕ) is not surprising: it is already suggested by the classical limit of V (ϕ) and by.7 Green’s functions and the scattering operator 113 Its evaluation requires setting up the renormalization programme and for further discussion of V1 given by (2. for example. remark in this section deals with the physical meaning of the effective potential. described by the lagrangian density L= 1 2 ∂µ ∂ µ − m2 2 − V( ) (2.2.169) dt HI (t) t U (t. t )|t t (2.2.170) can be expressed in terms of the interaction picture field operators. but very important. ∂t h I (t) = exp(iH0 t)HI exp(−iH0 t) (2. the operator S can be obtained as an infinite time limit of the evolution operator in the interaction picture.170) S = U (∞. (2. For simplicity.7 Green’s functions and the scattering operator The Green’s functions of quantum field theory can be used to determine the matrix elements of the operator S which are directly related to scattering amplitudes and therefore experimental results. let us consider the case of a single scalar field. Coleman (1974)) that V (ϕ) is the expectation value of the energy density in a certain state for which the expectation value of the field is ϕ. it is only the absolute minimum that corresponds to the true ground state of the theory. In the interaction picture the time evolution of operators is governed by the free. H0 . The interaction hamiltonian HI (t) in (2.171) It transforms the ‘incoming’ (t → −∞) states into the ‘outgoing’ (t → +∞) ones. Our final. 2. If V (ϕ) has several local minima.163) for the field vacuum expectation value 0| (x)|0 = v.168) with the formal solution |t = U (t.172) . t) = exp(iH0 t) (x.167) while the state vectors obey the equation i ∂ t = HI (t)|t . 0) exp(−iH0 t) (2. for example. part of the complete hamiltonian operator H = H0 + HI of the system (from now on we put h = 1) ¯ (x.

t) = ∂L/∂ ˙ (x. l = i 1 = · · · = jn (2.177) × 1 : (xl ): . (xm ) = : (x1 ) .175) to express S in terms of the annihilation and creation operators and calculate the matrix elements between states containing scalar particles. (xm ): + i. (2. t) = exp[−i(k0 t0 − k · x)]. j1 . jn (−i)G(xi1 − x j1 ) . (−i)G(xin − x jn ) when when m = 2n m = 2n + 1..175) where f k are the plane wave solutions of the Klein–Gordon equation f k (x.. i n . t). t)] (2π)3 2k0 (2. t) + [∇ (x)]2 + m 2 2 (x) + d3 x V ( (x)) = H0 + HI (2. When applied to the T -ordered product of m field operators Wick’s theorem has the well-known form (Bjorken & Drell 1965) T (x1 ) . Eq. by the hamiltonian H= 1 2 d3 x 2 ∇ (x. (2. . equivalently. . . with all creation operators standing to the left of the annihilation operators. .176) and a(k). . This is done by using Wick’s theorem. a † (k )] = (2π )3 2k0 δ(k − k ) (see Section 1.173) where (x.167) implies that the interaction picture field operator obeys the Klein– Gordon equation of a free field + m2 (x) = 0 (2. j (xk ): + · · · + i 1 .114 2 Path integrals or. t) + a † (k) f k∗ (x.173) and (2. t) = d3 k [a(k) f k (x. When calculating the matrix elements it is convenient to deal with normalordered products (in the interaction picture). We can now use (2. a † (k) are the annihilation and creation operators of scalar particles with three-momentum k satisfying [a(k).171). . j i= j (−i)G(xi − x j ) : k=i.174) so that at any moment of time we can write (x. k 0 = + k2 + m 2 1/2 (2.4).

(x1 ) (x2 ) (x3 ) (x4 ).179) where J (x) is a c-number function. after performing the functional differentiation in (2. the Klein–Gordon operator. the result is exactly (2. This corresponds to the case of scattering from an external source.183) . We obtain. Now let us take the interaction hamiltonian to be HI (t) = d3 x V ( ) = − d3 x J (x) (x) (2.180) × :exp i = W0 [J ] :exp i d4 x J where W0 [J ] is the previously introduced generating functional of the free-field Green’s functions.180) can be rewritten as follows T exp i d4 x J = : exp d4 x d4 y (x) G −1 (x − y) δ : W0 [J ] δ J (y) (2. the corollary of Wick’s theorem used to derive the perturbation theory Feynman rules in the previous section.182) Here G −1 (x − y) is the inverse operator to G(x − y) d4 y G −1 (x − y) G(y − z) = δ(x − z) i. If we assume F[ ] to be a product of four field operators. Functional differentiation with respect to the operator has been introduced here.178): T exp i d4 x J = exp 1 i 2 d4 y d4 x J (x) G(x − y) J (y) d4 x J : : (2.91).178) where F[ ] is some functional of the field operator and G(x − y) is the Green’s function given by (2. It is a straightforward generalization of the usual functional differentiation: the only difference is that one must keep in mind that the field operators for different values of time do not commute.181) δ J (y) so that (2.2. and take the matrix element of (2.e.7 Green’s functions and the scattering operator 115 A more general and compact formulation is T F[ ] = exp − 1 i 2 δ δ G(x − y) d4 x d4 y :F[ ]: δ (x) δ (y) (2.94).178) between the vacuum states. We observe that δ J (x)W0 [J ] = −i d4 y G −1 (x − y) W0 [J ] (2. (2.

Finally. . . . . .187) It is convenient to assume that ki = k j for all i. c are the normalization constants) |k1 . Let the initial state involve n particles and the final state m particles. defined as the vacuum matrix elements of the T -ordered products of the field operator in the Heisenberg picture. . Of the (n + m) field operators n will act as the annihilation and m as the creation operators: this gives a combinatorial n+m factor . (xn ): The factor N −1 contains all the vacuum-to-vacuum transitions and will be omitted in the following. .185) J ≡0 We have been dealing with fields in the interaction picture but in W [J ] we can recognize.54). see (2. We have (c. j pairs. there are n!m! ways of matching the annihilation and m . .184) Consequently. .88). . a + (kn )|0 k1 . The calculation of matrix elements is now straightforward. . . (xn+m ): will then contribute. the S operator of the interacting scalar field can be written in the form S = T exp −i × exp −i = N −1 :exp d4 x V ( ) = : exp d4 x V 1 δ i δJ W0 [J ] J ≡0 d 4 x d4 y (x) G −1 (x − y) δ : δ J (y) d4 x d4 y (x) G −1 (x − y) δ : W [J ] δ J (y) (2. . . . kn = ca + (k1 ) . yn ) (2.116 2 Path integrals The generalization to arbitrary V ( ) follows from the formula T exp −i d4 x V ( ) = exp −i d4 x V 1 δ i δJ T exp i d4 x J J ≡0 (2. d4 xn d4 yn G (n) (y1 . the generating functional of the previously introduced Green’s functions of the interacting theory. Only the term in the series with : (x1 ) . a(km ) (2.186) × iG (x1 − y1 ) . . by (2. km | = c 0|a(k1 ) . . . . iG −1 (xn − yn ) : (x1 ) . . Expanding the exponential we obtain S = N −1 ∞ n=0 −1 1 n! d4 x1 d4 y1 .

. kn = cc × d4 x1 . . . . kn = lim 1 2 k − m2 i i (2. . obtained from ˜ conn ˜ G (n) by factorizing out the exact two-point functions G (2) (ki ) for each of the ˜ (n) . . external lines of G conn At this point we must observe that in the derivation of (2.192) holds for the field operators both in the interaction and in the Heisenberg pictures. . . Its connected part is proportional to the on-shell limit of the so-called connected truncated Green’s function. . . . .2. .190) we have made the assumption (following from (2. km . yn ) (2. km |S|k1 . −kn ) where ˜ G (n) (k1 . −k1 . . . . f kn (xn+m ) 1 d4 y1 .188) With the usual invariant normalization |k1 . d4 yn+m iG −1 (x1 − y1 ) · · · ×iG −1 (xn+m − yn+m )G (n+m) (y1 . yn+m ) (2. . . (2. kn = a † (k1 ) . . . . . More general cases will be discussed in Chapter 4.189). .192) . . kn ) = d4 y1 exp(ik1 y1 ) . a † (kn )|0 this is just the on-shell limit of the Fourier transformed function k1 . i.192) we shall call the ‘physical’ fields. . d4 xn+m f k∗ (x1 ) . .e. . . . k| = 0|a(k) or k|k = (2π )3 2k0 δ(k − k ) Normalization (2. . .175)) that the one-particle state created by (x) from the vacuum has the standard normalization k| (x)|0 = exp[i(k0 x0 − k · x)] with k| normalized as in (2.191) We see that the S matrix element is equal to the multiple on-shell residue of the Fourier transformed Green’s function. . The result is k1 . . .190) (2. . . d4 yn exp(ikn yn ) G (n) (y1 .7 Green’s functions and the scattering operator 117 creation operators with the external particles. . Fields satisfying (2. km |S|k1 . .193) (2.189) ki2 →m 2 i ˜ ×G (n+m) (k1 .

We conclude therefore that 2 k1 →m 2 ˜ lim G (n) (k1 . . . Among the various terms which ˜ contribute to G (n) (k1 . . (yn )|0 − m2 d3 p |p p| (2π )3 2 p0 (2. . kn ) = 2 k1 i k1 . exp(ikn yn ) 0 0 0 y1 − max y2 . . . . . . . which by themselves can actually be taken as a proof of (2. We can insert a complete set of intermediate states between (y1 ) and the remaining time-ordered product getting ˜ G (n) (k1 .190) can be supplemented by the following arguments. . . (yn )|0 . . . . . kn ) = m d4 y1 . . . t = +∞|T (y2 ) .190). Let us consider the Fourier transform (2. where we have taken 0|0 = 1. and assume that the field (x) is normalized according to (2. . kn ) there is a time-ordering in which (y1 ) stands furthest to the left.118 2 Path integrals The above formal considerations leading to the Lehman–Symanzik– Zimmermann reduction theorem (2. It is important to observe that the pole arises due to the limit 0 y1 → +∞ in the integral (2. .191) of the Green’s function G (n) (y1 . .194) The y1 -integration can now be carried out using 0| (y1 )|n = 0| (0)|n exp(−i pn y1 ) and the integral representation for the 0 y1 − t = -function ∞ 0 dω exp −iω y1 − t ω + iε i 2π −∞ One then gets the following d4 y1 exp[i(k1 − pn )y1 ] 0 y1 − t = iδ(k1 − pn )(2π )3 = i(2π )3 δ(k1 − pn ) 0 k1 1 0 0 exp i k1 − pn t 0 − pn + iε (2. . .192). .195) 0 0 k 1 + pn 0 0 exp i k1 − pn t 2 2 k1 − pn + iε 2 Thus. . . . (yn )|0 + ··· (2.195).196) |1 1| ≡ . . d4 yn exp(ik1 y1 ) . yn ) = 0|T (y1 ) . a single particle state of mass m contributes a pole at k1 = m 2 while no other state |n and none of the remaining terms arising from other time-orderings can have such a pole. yn × 0| (y1 )|m m|T (y2 ) .

p2 |S|k1 . as in Section 2. a † (k2 )|0 4! λ = −i d4 x 0|a(p1 )a(p2 ) + (x) + (x) − (x) − (x) 4 ×a † (k1 )a † (k2 )|0 (2. . is valid for both the path integral and canonical formulations of quantum field theory. kn = k1 . k2 = 0|a(p1 )a(p2 )T exp −i λ 4! d4 x : 4 (2. .130)) and performing the integral over d4 x as well as the integrals over momenta in the field operators (1. . The S-matrix elements can be directly calculated as matrix elements of the operator in (2. . t = +∞|k1 . . k2 = −i λ d4 x 0|a(p1 )a(p2 ) : 4 (x):a † (k1 ).177) that contain contractions do not contribute to the connected amputated part of the matrix element.198) The basic result.188) and (2. (2. p2 |S − 1|k1 . for 2 → 2 scattering in the 4 theory we have: S2→2 = p1 .199) : a † (k1 )a † (k2 )|0 In the lowest order the contribution to the transition matrix is p1 . Commuting the creation and annihilation operators (using (1.190) consider elastic scattering of two identical scalar particles in the 4 theory.129). . kn . for instance.126). . using (2. . or in the (canonical) operator formalism (see. .126) we rederive (2. Bjorken & Drell (1965). . The latter approach is sometimes quite convenient.190). . km |S|k1 . (1.197) By repeating these steps we prove (2. Weinberg (1995)).190) can also be developed either in the path integral approach. In this order the terms in (2.200) where + ( − ) stands for positive (negative) frequency parts of the field operators (1. t = −∞ As an explicit example illustrating (2.198). remembering that k1 .7 Green’s functions and the scattering operator 119 Similarly.102) we get (S − 1)2→2 = −iλ(2π )4 δ (4) ( p1 + p2 − k1 − k2 ) (2. . 2 k1 →m 2 ˜ lim G (n) (−k1 . (yn )|k1 . In the lowest order.171) (expressed in terms of fields satisfying free equations of motion) taken between free particle states. . kn ) = 2 k1 i 0|T (y2 ) . t = −∞ − m2 (2. .2.4. Perturbation theory for calculating the Green’s functions present in (2. km . in particular for quick tree level manipulations. .190).188) and (2. For instance.

7). s4 )..177) and repeating the same steps as in the previous example one arrives at (S − 1) f1 f¯2 → f¯3 f4 = −ig si d3 q 4 . b1 |0 A u(p1 . s1 )b(q1 . s4 )b† (q4 . s2 )d(q2 . (2π )3 2E q4 d3 q 1 (2π )3 2E q1 × d4 x exp[−ix · (q1 + q2 − q3 − q4 )] B v(q3 . s4 ) etc.. f 1 (p1 .202) where we have used the abbreviations b4 ≡ b(p4 . s3 )d † × 0|b4 d3 :[u(q4 . s1 )]: d2 . In the lowest order we have f 4 (p4 . f¯3 (p3 . Using the Wick theorem (2. Also. s)). s1 ) = −ig d4 x 0|b4 d3 : † 0 f 4 (x)γ A f 3 (x) † 0 f 2 (x)γ B f 1 (x) † † : d2 . s2 ). Problems 2. s4 ) ¯ ×[v(q2 . s) (u(k. s2 ) (2. s3 )|S − 1| f¯2 (p2 . 2.1 Derive (2. s) (v(k. integrating over pi s first. it is clear that the absolute sign of the amplitude containing more than one fermion (or antifermion) in the initial and/or final states is purely conventional as it depends on the arbitrary ordering of fermionic creation operators defining these states.2 For a scalar field theory defined by a lagrangian L = 1 ∂ µ (x)∂µ (x) − V ( (x)) 2 . s)) and the initial ¯ (final) antifermion with momentum k and spin s by v(k. b1 |0 (2. s3 )][v(p2 . s2 ) ¯ ×[u(p4 . The relative signs of several different contributions to the transition amplitudes (if present) and various combinatorial factors are easy to fix by operator manipulations.120 2 Path integrals As a second example consider the so-called four-fermion interaction lagrangian of the generic type LI = −g :( ¯ f4 (x) A f 3 (x))( ¯ f2 (x) B f 1 (x)): (2. s1 )] = −ig(2π )4 δ( p4 + p3 − p2 − p1 ) ¯ B v(p3 .201) (where A(B) are some constant matrices in the spinor space) and compute the matrix elements of S − 1 for the process f 1 f¯2 → f¯3 f 4 . The operator ¯ method is convenient to reach this conclusion quickly. s3 )] † † A u(q1 .12) from (2.203) It follows from this example that the initial (final) Dirac fermion with momentum k and spin s is described by the wave-function u(k. s4 ) ¯ (q3 .

3 Using the rules of Section 2. where (η.5 Prove (2. Aη) = 2.6 Prove (2.7 Using representation (2. .54) and (2.4 For real Grassmann variables ηi prove the formula dη1 . 2.156) and (2.56) and the invariance of the generating functional (2.Problems and the corresponding equation of motion +V ( )=0 derive the equation of motion for the Green’s function 0|T (x) (y)|0 : x 121 0|T (x) (y)|0 = − 0|T V ( (x)) (y)|0 − iδ (4) (x − y) using definitions (2. .57) under an infinitesimal change of variables (x) → where f (x) is a function of x D exp iS[ + ε f ] + i d4 x ( + ε f ) J = D exp iS[ ] + i d4 x J (x) + ε f (x) Repeat the same in Euclidean space.158). (2. dηn exp 1 2 (η Aη) = (det A)1/2 i. 2.10) calculate the Green’s function for the harmonic oscillator x | exp[−i(t − t)H ]|x = where H = − 1 ∂ 2 ∂ x 2 + 1 ω2 x 2 2 2 ω 2 S(τ ) = x + x 2 cos ωτ − 2x x 2 sin ωτ ω 2iπ sin ω(t − t) 1/2 exp[iS(t − t)] . j for any antisymmetric matrix A. ηi Ai j η j .154). 2.178).4 calculate the following symmetry factors S in the λ theory 4 2.

204) J 4 ˆ 0d x D exp − 1 2 d4 x d4 y ˆ ˆ (y) − ˆ δ 2 SE [ 0 ] δ (x)δ ( y ) ˆ ˆ (2. respectively: (a) Write det M instead of (det M)−1 as a Gaussian integral.2 with V ( ) = λ 4 calculate the effective potential [ ] by evaluating the generating functional WE [J ] in Euclidean space WE [J ] = NE D exp −SE [ ] + J d4 x ˆ by the method of steepest descent. (b) Show that the supertrace defined as Tr M = Tr A − Tr B satisfies the cyclicity property Tr[M1 M2 ] = Tr[M2 M1 ] .8 For a scalar field theory as in Problem 2.166). The steps are: (a) Expand the exponent around 0 (often called the background field) which is the solution of the classical equation of motion in the presence of the external source J (x) ˆ ˆ −∂ µ ∂ µ + m 2 to write WE [J ] = NE exp −SE [ × 0] + 0 +V ( 0) − J =0 (2.204) perturbatively in powers of λ.9 For a matrix M acting in superspace z η =M z = η A C D B z η where z and η are commuting and anticommuting coordinates.122 2 Path integrals Note that S(τ ) is the classical action (along the classical path). 2. Check that in this approximation [ ] is the classical action. Calculate the effective potential and prove (2. 2. Evaluate the Gaussian integral in (2.205) × [ (x) − ˆ (b) In the classical approximation ˆ 0 ( x)][ ˆ 0 ( y )] + · · · WE [J ] = NE exp −SE [ 0] + d4 x J ˆ 0 Evaluate Z [J ] and [ ] explicitly by solving (2. (c) Calculate [ ] in the saddle point approximation.205) by using det A = exp(Tr ln A) and expanding the logarithm in powers of λ.

as for ordinary matrices. B]]. B] + 1 [A. . 123 B]. B] + · · · 4! The exponent and the logarithm of a matrix in superspace are defined by a series expansion. prove that ln det M = Tr ln M and det(M1 M2 ) = det M1 det M2 (e) Write the Jacobian of a transformation in superspace. [A.Problems (c) Show that Tr ln M1 M2 = Tr ln M1 + Tr ln M2 Hint: use the Campbell–Backer–Hausdorff formula for ordinary matrices exp(A) exp(B) = exp A + B + + 1 1 [A. (d) Writing M= A C 0 1 1 0 A−1 D B − C A−1 D and using property (c). B]] 2 1 [[A. [A. B] + 2! 3! 1 2 [[A.

The integrand of (3.1) is constant along any orbit 124 . which implies multiple counting of the physically equivalent configurations (equivalent up to a gauge transformation). As explained before. 3. gauge-invariant.1) For brevity we write Aµ instead of Aα . the µ straightforward generalization of the integration measure introduced previously for scalar fields.3 Feynman rules for Yang–Mills theories 3. Let us divide the g configuration space {Aµ (x)} into the equivalence classes {Aµ (x)} called the orbits of the gauge group. This construction can be represented graphically as shown in Fig. i. An orbit of the group includes all the field configurations which result when all possible transformations g from the gauge group G are applied to a given initial field configuration Aµ (x). quantity f (Aµ ) D Aµ f (Aµ ) exp i d4 x L (3. where α is the gauge group index.e.1. In this chapter we shall derive Feynman rules for gauge theories.1 The Faddeev–Popov determinant Gauge invariance and the path integral As discussed in the Introduction gauge quantum field theories are of special physical interest in our attempts to describe fundamental interactions. The integration measure D Aµ we assume to be invariant under gauge transformations: i.e. The simplest formal ansatz for the integration measure which has this property is D Aµ = µ.α. L is the µ lagrangian density.x dAα (x).1) runs over all possible configurations Aµ (x). Consider a path integral over gauge field Aµ . Aµ denotes the result of this transformation when applied to Aµ . the path integral in (3. it must have the following property: g D Aµ = D Aµ g where g is an arbitrary transformation from the gauge group. corresponding to a physical.

(3. In fact to fix a gauge we need (at each space-time point) one equation for each parameter of the group.2) [Aµ ] defined by the following equation [Aµ ] g Dg δ F Aµ (3.3. .1) after the infinite constant has been factorized out. [Aµ ]). we assume that the equation g F Aµ = 0 (3. One way to resolve this problem is to apply perturbation theory to the functional integral over the coset space of the orbits of the group (in other words. for any initial field Aµ .4) has exactly one solution.1. i.4) defines a ‘gauge’. Consequently.1 The Faddeev–Popov determinant 125 Fig. where α is the group index of the adjoint representation and x a space-time point.3) Here δ[ f (x)] represents the product of the usual Dirac δ-functions: x δ( f (x)). The problem appears when we want to calculate (3. In the configuration space {Aµ (x)} the equation F[Aµ ] = 0 must then define a surface that crosses any of the orbits of the group exactly once. Let us observe that the functional F[Aµ ] should be of the general form F α (x. This is not an important difficulty in itself. We shall now describe a procedure by means of which this factorization can be achieved. As for the functional F[Aµ ]. g0 . the integral as it stands is proportional to an infinite constant (the volume of the total gauge group G). one at each space-time point. and let us introduce a functional 1= Dg = x dg(x) (3. because the infinity can always be cancelled by a normalization constant.e.1) perturbatively because local gauge symmetry implies that the quadratic part of the gauge field action density has zero modes and therefore cannot be inverted in the configuration space {Aµ (x)} so that the propagator of the gauge field cannot be defined. Let Dg denote an invariant measure on the gauge group G Dg = D(gg ). 3. of the gauge group. over the physically distinct field configurations) which follows from (3.

e.5) where we have used the invariance property of the group measure Dg. From (3. Next we change the order of integrations. [Aµ ] = det g δ F Aµ δg (3.7) which defines the theory as formulated in the F[Aµ ] = 0 gauge. (3. We start by inserting (3. f (Aµ ) and S[Aµ ] to replace them by D Aµ . Unlike (3. the group integration Dg factorizes out to produce an infinite constant: the volume of the full gauge group. Faddeev–Popov determinant We have still to calculate −1 [Aµ ].1). In this case each orbit will contribute only one field configuration and we shall have an integration over physically distinct fields.126 3 Feynman rules for Yang–Mills theories [Aµ ] is invariant under gauge transformations. In effect.6) An important observation is that the complete expression under the Dg integral is in fact independent of g.8) F[Aµ ]=0 g . Our aim is to replace integration over all field configurations by integration restricted to the hypersurface F[Aµ ] = 0.3) we obtain. the result g D Aµ g g g Aµ f Aµ δ F Aµ g exp iS Aµ g can be made manifestly g-independent by a change in notation: Aµ → Aµ . This can be shown as follows −1 g Aµ = gg Dg δ F Aµ g Dg δ F Aµ = = gg D(gg ) δ F Aµ −1 = [Aµ ] (3. The result is Dg D Aµ g [Aµ ] f (Aµ ) δ F Aµ exp{iS[Aµ ]} (3. after formal manipulations DF det g δ F Aµ −1 [Aµ ] = δg δ[F] i. Consequently. g g g g Aµ .7) can be used as a starting point for perturbative calculations.3) (which is equal to one) under the path integral in (3.1). f Aµ and S Aµ : [Aµ ]. the functional [Aµ ] depends only on the orbit to which Aµ belongs. To show this we can use the gauge invariance of D Aµ . We obtain finally Dg D Aµ [Aµ ] δ[F[Aµ ]] f (Aµ ) exp{iS[Aµ ]} (3.

10) is proved. g=1 F[Aµ ] = 0 (3. y) δ F α x.9) in a more explicit form. Aµ in powers of the infinitesimal parameters β (y) F α x. [Aµ ]) δ Aρ (z) ν δ β (y) θ =0 δ Aρ (z) ν δ F α (x. F α (x. Aµ δ β (y) =0 .13) [Aµ ] = det M.3. y).9) Near g = 1 we only have to deal with the infinitesimal transformations: U ( ) = 1 − iT α α (x) (where α (x) 1) and the invariant group measure Dg takes the α simple form D ≡ α. y) = = d4 z δ F α (x. For infinitesimal transformations ≈ + and therefore det(d /d ) = 1. This matrix appears in the expansion of F α x. It is convenient to use the gauge invariance property of [Aµ ] to choose Aµ which already satisfies the gauge g condition F[Aµ ] = 0.8) we can replace the constraint F Aµ = 0 by g = 1. F[Aµ ] = 0 (3. we can also write M αβ (x. so that (3. y) = and αβ (3.10) where U ( ) = U ( )U ( ) and is arbitrary.1 The Faddeev–Popov determinant 127 [Aµ ] is usually called the Faddeev–Popov determinant. [Aµ ]) = 0 (3. We have to calculate a determinant of a matrix in both space-time and the group indices M αβ (x. It is easy to check the invariance property of D D =D (3. First .105) for the infinitesimal transformation of the gauge field. [Aµ ]) 1 (z) d4 z cρβγ Aγ (z) + δρβ ∂ν δ(z − y) ν g δ Aρ (z) ν (3. Then in (3.12) M (x. Aµ δ β (y) =0 β (y) + · · · (3. which simplifies the practical calculations [Aµ ] = det g δ F Aµ δg . We can now rewrite (3. [Aµ ]) + d4 y M αβ (x. with all relevant indices [Aµ ] = det δ F α x. Aµ so that = F α (x.x d (x).14) Recalling (1. β are the gauge group indices.11) g Here Aµ stands for Aµ and α.15) For the subsequent applications the following observations will be helpful.

This is equivalent to replacing the original lagrangian density by 1 Leff = L − (F[Aµ ])2 (3. consequently. For all gauge conditions belonging to this class det M is the same because C(x) is unaffected by g the gauge transformation from Aµ to Aµ . We can make use of this feature in order to replace the δ-functional in (3. The derivation of Feynman rules in gauges fixed by δ-functionals can often be simplified by taking this limit. because the quantity we calculate is gauge-invariant by assumption.16) is the only C-dependent term. but the final results will not. The rules of this perturbation theory will depend on the gauge. of a gauge-invariant quantity can be rewritten in a fixed gauge.18) approaches the δ-functional (up to the α-dependent normalization factor which for our purposes is irrelevant). In particular the generating functional W [J ] in the F[Aµ ] = 0 gauge is given by the following .7) by some other functional of the gauge condition. Consider a class of gauge conditions of the form F[Aµ ] − C(x) = 0. perturbation theory may apply in this case. In the F[Aµ ]−C(x) = 0 gauge (3. Observing that the δ-functional in (3. where C(x) is an arbitrary function of a space-time point. Notice also that in the limit α → 0 (3. we obtain D Aµ det M f (Aµ ) exp{iS[Aµ ]} = DC δ[F[Aµ ] − C] G[C] (3.1). so that the perturbation theory may be set up. because the condition F[Aµ ] = 0 is already enforced by the δ-functional which fixes the gauge. which may be more convenient for practical calculations.16) only by an overall normalization constant.128 3 Feynman rules for Yang–Mills theories let us note that [Aµ ] under a path integral like (3.19) 2α which is no longer gauge-invariant. In the following we shall also deal with gauge-dependent quantities. (3.7) becomes Dg D Aµ det M δ[F[Aµ ] − C(x)] f (Aµ ) exp{iS[Aµ ]} (3.18) where α is a real constant.7) can be replaced by det M. We can integrate (3.16) Being gauge-invariant. this is obviously independent of C(x).16) functionally over DC with an arbitrary weight functional G[C]. the result will differ from (3. We have shown how a path integral representation.17) D Aµ det M f (Aµ ) exp{iS[Aµ ]}G[F[Aµ ]] A popular choice for G[C] is G[C] = exp − i 2α d4 x [C(x)]2 (3.

We shall calculate the free photon propagators G µν (x1 . y)-matrix defined by (3. Although in a sense unphysical. x2 ) corresponding to 0 different gauge conditions. it is in order to mention here that the renormalized S-matrix elements are gauge-independent.7. the Green’s functions are often useful at intermediate stages of calculations and physical quantities may be conveniently defined in terms of them. In such cases one can discuss IR regularized theory but the regulator must not break gauge invariance. The equivalence of different gauges can be explicitly checked for various cases (see. To deal with the δ-functional δ ∂µ Aµ which fixes the gauge we shall (3. the Green’s functions of the gauge field. this is not a gauge-invariant object and the same applies to its functional derivatives with respect to J . the S-matrix suffers from IR divergences and may not even be defined. IR dimensional regularization can be used as an intermediate step to prove gauge independence of the physical. Slavnov & Faddeev (1980).13) is then of the form 1 M(x.21) . finite quantities. One can show that only the renormalization constants attached to each external line depend on the gauge but that is of no consequence for the renormalized S-matrix. First we assume the so-called Lorentz gauge ∂µ Aµ = 0 The M(x. as we know from Section 2. an element of the S-matrix is obtained from the corresponding Green’s functions by removing single-particle propagators corresponding to external lines. The other gauge-dependent terms do not contribute to the poles of the Green’s functions at pi2 = m 2 .1 The Faddeev–Popov determinant 129 expression: W [J ] = N D Aµ [Aµ ] δ(F[Aµ ]) exp i α d4 x L + Jµ Aµ α (3. Abers & Lee (1973). Examples As a simple example let us consider the case of quantum electrodynamics (QED). Although we have not discussed the renormalization programme yet. taking the Fourier transform of the resulting ‘amputated’ Green’s function and placing external momenta on the mass-shell. if not with a spontaneously broken gauge symmetry.3.20) α Due to the presence of fixed external sources Jµ . which is irrelevant. These formal arguments are rigorous except for the fact that in a gauge theory. y) = + ∂ 2 δ(x − y) e where ∂ 2 = ∂µ ∂ µ and the Faddeev–Popov determinant is independent of the field Aµ . for example. In particular. ’t Hooft & Veltman (1973)). consequently it can be included into the overall normalization factor.

We can simply set Aµ = 0.22) is given by W0 [J ] = exp 1 i 2 T d4 x d4 y Jµ (x) D µν (x − y) JνT (y) (3. We L L know that the J -dependence of a path integral like (3. it requires that ∂ 2 = 0.e. The longitudinal fields are of the form AL = ∂µ (‘pure gauge’). Taking this all into account we obtain the following path integral formula for W0 [J ] in the Lorentz gauge (in the following the fermion degrees of freedom are always suppressed): W0 [J ] ∼ D AT exp i µ d4 x 1 2 T AT ∂ 2 g µν AT + Jµ Aµ µ ν T where Pµν = gµν − ∂µ ∂ν /∂ 2 is the projection operator. There is also another sort of gauge ambiguity (the Gribov ambiguity) with more physical implications. Clearly we have (3. This is the residual gauge freedom allowed by (3. however.24) . Such boundary conditions are in fact implicit in the usual derivation of perturbation theory rules from path integrals. In Euclidean space we can do it by assuming that the gauge field Aµ vanishes at Euclidean infinity. i.23) Thus the longitudinal Aµ does not appear in the action and in the Lorentz gauge L δ ∂µ Aµ makes the integration over D AL trivial. The existence of this residual gauge freedom contradicts our previous assumpg tion that the gauge condition F Aµ = 0 has exactly one solution g0 for any Aµ . are separated by finite gauge transformations.22) (the extra term 1 iε AT g µν AT in the action should always be kept in mind). Here µ ν 2 we have made use of the fact that the lagrangian density of a free electromagnetic field can be written as follows − 1 Fµν F µν = 1 Aµ ∂ 2 g µν − ∂ µ ∂ ν /∂ 2 Aν = 1 AT ∂ 2 g µν AT ν 4 2 2 µ (3. let us observe that the residual gauge freedom can be removed by an appropriate choice of the boundary conditions. In the case of the Lorentz gauge it appears only in non-abelian gauge theories. This is the case when the points of intersection of a group orbit with the F = 0 surface are at a finite distance from each other.130 3 Feynman rules for Yang–Mills theories make use of a decomposition of the field Aµ into transverse AT and longitudinal µ AL parts µ AT = Pµν Aν . µ AL = (gµν − Pµν )Aν µ ∂µ A µ = 0 T so that the gauge-fixing δ-functional does not affect the integration over the transverse component of the gauge field. However. The Lorentz gauge condition does not set them exactly µ to zero. unaffected by this ambiguity.21). instead. Perturbation theory is.

summarized in (3. so that we have the following equation for Dµν : ∂ 2 − iε Dµν (x − y) = − gµν − ∂µ ∂ν /∂ 2 δ(x − y) After a Fourier transformation this becomes ˜ −k 2 − iε Dµν (k) = − gµν − kµ kν /k 2 so that ˜ Dµν (k) = and Dµν (x − y) = d4 k gµν − kµ kν /k 2 exp[−ik(x − y)] (2π )4 k 2 + iε (3.1 The Faddeev–Popov determinant 131 where −Dµν (x − y) is the inverse of the quadratic action. Again we can write W0 [J ] = exp 1 i 2 d4 x d4 y J µ (x) Dµν (x − y) J ν (y) (3. Note also that in transverse space. In transverse space the unit operator is equivalent to Pµν δ(x − y).25) The prescription of the k 2 pole in the kµ kν /k 2 term can be arbitrary.27).17)–(3.26) (3.27) 1 kµ kν gµν − 2 2 + iε k k = ˜ Pµν k 2 + iε (3.29) The quadratic action is now no longer proportional to a projection operator and therefore can be inverted over all configuration space. x2 ) = 0 1 i 2 δ 2 W0 [J ] δ Jµ (x1 )δ Jν (x2 ) J ≡0 = −iD µν (x1 − x2 ) (3.19). With F[Aµ ] = ∂µ Aµ we obtain the effective lagrangian Leff = − 1 Fµν F µν − 4 1 ∂µ A µ 2α 2 = 1 Aµ gµν ∂ 2 − 1 − 2 1 ∂µ ∂ν Aν α (3. We can now write the formula for the free electromagnetic field propagator in the Lorentz gauge G µν (x1 . we can always replace Pµν by gµν . as long as we deal only with transverse sources. A generalization of this result follows if we make use of another procedure for fixing the gauge.30) where Dµν (x − y) satisfies the equation ν ∂ 2 gµν − ∂µ ∂ν (1 − 1/α) − iε Dλ (x − y) = −gµλ δ(x − y) (3.3.31) .28) where Dµν (x1 − x2 ) is given by (3.

27) (the Lorentz or Landau gauge).132 3 Feynman rules for Yang–Mills theories with the solution kµ kν 1 exp[−ik(x − y)] k 2 + iε k 2 + iε (3. also called the radiation gauge (Bjorken & Drell 1965). . For the Coulomb gauge one gets ˜ αβ Dµν = and for axial gauges ˜ αβ Dµν = n2 δ αβ kµ n ν + kν n µ + kµ kν gµν − k 2 + iε k·n (k · n)2 (3.33) which allow the quanta of the vector fields to be interpreted as partons.36) δ αβ k · n(kµ n ν + kν n µ ) − kµ kν gµν − 2 + iε k (k · n)2 − k 2 (3. .19). The only modification is that they are diagonal matrices in the non-abelian quantum number space. The case α = 1 corresponds to the Feynman gauge.36) neglects subtle problems raised by the presence of the (n · k) singularity. . Dµν (x − y) = gµν − +α d4 k (2π )4 kµ kν k 2 + iε Non-covariant gauges Non-covariant gauges. α n 2 = 0 or n 2 < 0. depending on an arbitrary four-vector n µ . inverting the term of the action quadratic in the gauge fields and taking the limit α → 0.32). Free propagators of the gauge field in the Coulomb and axial gauges can be formally obtained following the procedure summarized in (3. Our formal derivation of (3. 0. In QCD one often uses axial gauges n µ Aµ = 0.32) In the limit α → 0 this coincides with (3.35) Propagator (3. A more complete treatment of n 2 < 0 and n 2 = 0 gauges has been given in the framework of . 0. For instance the standard canonical quantization of QED uses the Coulomb gauge. The free propagators for non-abelian gauge fields in covariant gauges are also given by (3.34) n µ = (1. 8 (3. α = 1. defined by the condition ∂µ Aµ − n µ ∂ µ n µ Aµ = 0.17)–(3. are also frequently used. 0) (3.35) can be split into the transverse propagator and the static Coulomb interaction term (Bjorken & Drell 1965). .28) and (3. Only in this limit do the propagators behave as k −2 . It is easy to check that in axial gauges the Faddeev–Popov determinant is Aα independent even µ in a non-abelian theory.

The (n · k) singularities are related to the residual gauge freedom still present after the axial gauge condition is imposed.19). Lazzizzera & Soldati 1985. 1) (3.38) follows from (3. 1985). Bassetto. Choice of this condition gives rise to a specific prescription for the spurious singularity in the free boson propagator.37) This prescription has been proposed earlier without any canonical justification as one which allows for Wick rotation (Mandelstam 1983. The treatment of the light-cone gauge n 2 = 0 is quite different (Bassetto et al. For the sake of definiteness we choose a class of covariant gauges ∂µ Aµ = c(x) or. .2 Feynman rules for QCD 133 canonical quantization (Bassetto. 0. = 2 2 n·k k + k⊥ + iε n = (1. The matrix M αβ . the general proof of the renormalizability of the theory in the light-cone gauge is still lacking.38) Relation (3.2 Feynman rules for QCD Calculation of the Faddeev–Popov determinant In order to derive Feynman rules for a non-abelian gauge theory we have to take proper account of the Faddeev–Popov determinant whose presence under the functional integral (3. γ depends on gauge fields Aµ . This time the residual gauge freedom involving functions which do not depend on the variable n µ x µ cannot be eliminated: we are not allowed to impose boundary conditions which would interfere with the time evolution of the system. equivalently. In QCD calculations the principal value prescription has also been extensively used. 3. With both kinds of prescription dimensional regularization is usually adopted (Pritchard & Stirling 1979 and Appendix B). y) = (1/g) δ αβ ∂ 2 + gcαβγ Aµ (x)∂µ δ(x − y) = (1/g) Dµ ∂ µ γ αβ δ(x − y) (3.15).20) is a consequence of quantization with constraints: gauge-fixing conditions. 0. not only on the hypersurface ∂µ Aµ = 0 but also for ∂µ Aµ = c(x) because det ∂µ D µ = det Dµ ∂ µ . Nevertheless a hamiltonian canonical quantization leads to the well-defined prescription for the singularity (n · k): 1 k0 + k3 . Lazzizzera & Soldati 1985). In the space-like case n 2 < 0 one can show (Bassetto. However. Leibbrandt 1984). It then follows from (3.15) that the Faddeev–Popov matrix reads M αβ (x. Dalbosco. Lazzizzera & Soldati 1985) that this residual gauge symmetry is directly connected to the spatial asymptotic behaviour of the potentials and can be eliminated by imposing a specific boundary condition. assume (3.3.

5 we can write det M αβ (x.42) One can define the ‘propagator’ of the ghost field η G 0 (x.40) + β ∗ α (x)ηα (x) + η∗ α (x)β α (x) Using standard methods.134 3 Feynman rules for Yang–Mills theories The standard method of dealing with the Faddeev–Popov determinant det M αβ (x.38).39) and the explicit form (3.39) where C is some unimportant constant (the factor (1/g) in relation (3. y) = and expanding (3. the β-dependence of W0 [β] reads explicitly W0 [β] = exp i where αβ η η d4 x d4 y β ∗ α (x) αβ (x − y) β β (y) (3.41) we get G 0 (x. The ghost field propagator can be calculated from the generating η functional W0 [β] W0 [β] ∼ η Dη Dη∗ exp i d4 x η∗ α (x)δ αβ ∂ 2 ηβ (x) (3. propagator ηαβ G 0 (x. y) = C Dη Dη∗ exp i d4 x d4 y η∗ α (x) M αβ (x. y) we can take account of det M αβ (x. y) = −i Graphically. ηαβ αβ ηαβ 1 i δ δβ ∗ α (x) W0 [β] η δ δβ β (y) ← (3. y) is to replace it by an additional functional integration over some auxiliary complex fields η(x) (ghost fields) which are Grassmann variables.44) y) will be denoted by with an arrow pointing towards the source of η. y) ηβ (y) (3.43) β=β ∗ =0 (x − y) (3.41) (x − y) = δ αβ d4 k 1 exp[−ik(x − y)] 4 k 2 + iε (2π ) 2 (3. The Feynman rules for ghost fields follow in a straightforward way from (3.38) of the matrix M αβ (x.38) is also included in it) and the integration over η and η∗ is equivalent to integration over the real and the imaginary parts of the field η. Using the results of Section 2. .39) and (3. Having (3. y) in the framework of our general perturbative strategy for calculating various Green’s functions.

η] + SG ] α d4 x Jµ (x) Aµ (x) + β ∗ α (x) ηα (x) + η∗ α (x) β α (x) α + α(x) ¯ where S[A.47) + ¯ (x) iγµ ∂ µ − m and into the interaction term SI SI [A. ¯ describe quarks and are vectors in the three-dimensional colour space. α.32).46) SG = −(1/2α) Dirac spinors .2 Feynman rules for QCD 135 Feynman rules The full generating functional W [J. The propagators 0|T Aα (x)Aβ (y)|0 . η] = d4 x gcαβγ ∂µ Aα Aµβ Aνγ − 1 g 2 cαβγ cαρσ Aβ Aγ Aµρ Aνσ ν µ ν 4 − g ¯ γµ Aµα T α + gcαβγ Aγ η∗ α ∂ µ ηβ µ (3. respectively.49) + β ∗ α (x) (x − y)β β (y) µν where the Green’s functions Dαβ . .45) − 1 G α G µν 4 µν α + ¯ iγµ D µ −m ¯ S η [A. The ‘free’ generating functional reads W0 = exp i d4 x d4 y αβ µν 1 α J (x)Dαβ (x 2 µ − y)Jνβ (y) + α(x)α Sαβ (x − y)αβ (y) ¯ (3. (2. ]= d x 4 (x) + ¯ (x) α(x)              (3. We split the action into the term S0 describing non-interacting theory (bilinear in fields and including SG ) S0 [A.122) and (3. . 0|T α (x) ¯ β (y)|0 µ ν and 0|T ηα (x)η∗ β (y)|0 are obtained by multiplying the respective D-. η] = d4 x η∗ α (x) δ αβ ∂ 2 + gcαβγ Aµ (x)∂µ ηβ (x) γ d4 x ∂ µ Aα µ 2 (3. S.42). Sαβ and αβ are given by (3. . α.and -functions by (−i).3.48) Matrices T α are SU (3) colour generators in the fundamental representation. η] = d4 x 1 2 Aα (x)δ αβ ∂ 2 g µν − ∂ µ ∂ ν /∂ 2 Aβ (x) µ ν (x) + η∗ α (x)δ αβ ∂ 2 ηβ (x) + SG (3. β] for QCD reads as follows: W [J. ] + S η [A. β] = N +i D Aα D µ D ¯ Dη Dη∗ exp i[S[A.

. Aβ µ ν ¯ ηη∗ ] N D[A .50) eliminates all vacuum-tovacuum diagrams. ¯ |0 ¯ ηη∗ ] Aα . Aβ µ ν = D[A .2.51) .. ¯ N N (iSI ) N exp(iS0 ) N! (iSI ) exp(iS0 ) N! (3. . We shall now consider several Green’s functions in the first order in SI to derive Feynman rules for vertices in QCD.136 3 Feynman rules for Yang–Mills theories Fig. . Any Green’s function in QCD can be calculated perturbatively from the expansion 0|T Aα . x2 . . . x3 ) ≡ 0|T Aa (x1 )Ab (x2 )Ac (x3 )|0 µ ν λ (3.50) We recall that the presence of the denominator in (3. One is usually interested in Green’s functions in momentum space. . 3. . We begin with G µνλ (3g)abc (x1 .

(e.2 Feynman rules for QCD 137 which defines the full symmetrized triple gluon (3g) vertex by means of the relation G µνλ (3g)abc (x1 . x2 . The final result reads: G µνλ (3g)abc (x1 . κ). In the first order in SI one has G µνλ (3g)abc (x1 . 3. For each assignment there is a 23 3! symmetry factor which cancels with factors coming from the expansion of W0 . y2 . x3 and y1 . r ) as (x1 . x3 ) = igcde f (d) d4 y ∂ρ ad be −iDµκ (x1 − y) −iDνρ (x2 − y) −iDλ (x3 − y) cf κ × (3.3.55) ˜ In momentum space we define G µνλ G µνλ (3g)abc (3g)abc ( p. x3 ) = aa bb d4 y −iDµµ (x1 − y) −iDνν (x2 − y) cc × −iDλλ (x3 − y) Vaµbνcλ (y) (3. κ) and the derivative is with respect to y.52) we get (d) abc (d) (d) Vµνλ = igcde f δ ad δ be δ c f gµλ ∂ν + δ a f δ bd δ ce gµν ∂λ + δ ae δ b f δ cd gνλ ∂µ (d) (d) (d) + δ ad δ b f δ ce gµν ∂λ + δ ae δ bd δ c f gνλ ∂µ + δ a f δ be δ cd gµλ ∂ν (3. Following the standard method we assign points x1 . x2 .54) in the form (3. x2 . x3 ) = 1 δ3 i c a i6 δ Jµ (x1 )δ Jνb (x2 )δ Jλ (x3 ) × (d) d4 y gcde f ∂ρ δ3 W0 [J. ρ) and ( f. x3 ) = d4 p d4 q d4r exp(−i p · x1 ) exp(−iq · x2 ) (2π )4 (2π )4 (2π )4 ˜ (3g)abc × exp(−ir · x3 ) G µνλ ( p. x2 .54) where the sum denotes all permutations of sets of indices (d. q.53) (d) (the superscript ‘d’ in ∂ρ means that the component Ad is differentiated with κ respect to y) and the cubic term in the expansion of W0 contributes to the final result. x2 . r ) (3.2. β.52) illustrated in Fig. Writing the result (3. The factor 1/i6 i3 = (−i)3 is included in the three propagators. q. y3 → y to the prototype diagram For a connected diagram there are six different orderings in pairs joined by propagators.56) . α]| J =β=α=0 d e δ Jκ (y)δ Jρ (y)δ J f κ (y) (3.

138 3 Feynman rules for Yang–Mills theories µνλ and the 3g vertex Vabc ( p. x2 . Writing the propagators in relation (3. (3.57) one gets the final result ˜ abc Vµνλ ( p.57). Similarly. r ) by the relation ˜ (3g)abc ˜ (2g)aa ˜ (2g)bb (q)G (2g)cc (r )V µ ν λ ( p. r ) = −(2π )4 δ( p + q + r )gcabc × [(r − p)ν gµλ + (q − p)λ gµν + (r − q)µ gνλ ] (3. . differentiating according to (3. x3 . r ) = G µµ ( p)G νν λλ abc where ˜ (2g)aa G µµ ( p) = p2 −i pµ pµ gµµ − (1 − α) 2 δ aa + iε p (3. x4 ) = (−i)4 aa bb d4 y Dµµ (x1 − y) Dνν (x2 − y) cc dd ×Dλλ (x3 − y) Dσ σ (x4 − y) Vaµbνcλdσ (3. x3 . integrating over d4 y and finally comparing with (3. y4 to the prototype diagram.61) . x2 . The starting point is the Green’s function G µνλσ (4g)abcd (x1 .52) in terms of their Fourier transforms (3. x3 .56) and (3. There are 24 different assignments of pairs of points x1 . q. r ) ˜ ˜ G µνλ ( p. . x4 . The additional 4!24 symmetry factor is cancelled by the same factor coming from the expansion of W0 .58) (where all momenta are outgoing). . q. . x2 .57) etc.59) Writing G µνλσ (4g)abcd (x1 . x4 ) ≡ 0|T Aa (x1 )Ab (x2 )Ac (x3 )Ad (x4 )|0 µ ν λ σ which in the first order in SI defines the 4g vertex according to the following relation: G µνλσ (4g)abcd (x1 .60) abcd one can calculate Vµνλσ . x4 ) = 1 δ4 c a d i8 δ Jµ (x1 )δ Jνb (x2 )δ Jλ (x3 )δ Jσ (x4 ) δ4 h δ Jκ δ Jρ δ J mκ δ J nρ (y) f ×i d4 y − 1 g 2 ce f h cemn 4 W0 | J =β=α=0 (3. Remembering that the Lorentz contraction in the 4g vertex always occurs between colour indices f and m and h and n one gets the final result: abcd Vµνλσ = −ig 2 [cabe ccde (gµλ gνσ − gµσ gνλ ) + cace cbde (gµν gλσ − gµσ gνλ ) + cade cbce (gµν gλσ − gµλ gνσ )] which is obviously also true in momentum space. q. one can derive the expression for the 4g vertex. q.55). y1 .

written as µ aa d4 y −iDµµ (x1 − y) −i bb (x2 − y) −i cc (x3 − y) Vaµb c (y) (3. The final task is to find expressions for the ghost and fermion vertices in Fig. 3. respectively. Fig. Noticing that the product of two vertices gives us according to (3. respectively. They can be obtained from the Green’s functions 0|T ηc (x3 )ηb (x2 )Aa (x1 )|0 and µ 0|T j (x3 ) ¯ i (x2 )Aa (x1 )|0 .62) The number of different assignments to the prototype diagram is 3 × 2 × 3 × 2 × 4!24 . Following the previous calculations closely we . 3. Following the functional approach one can also easily find that loop diagrams in Fig. 1 and 1 .3. 3.g.3 contribute with additional factors 1 .4. In 2 2 6 particular diagrams (a) and (b) occur in the second order in SI .63) and similarly for the ¯ A vertex. 3.4.58) 36 combinations we are left with an additional factor 1 . the expression corresponding to the first diagram is of the form (iSI )2 1 1 J DJ J DJ J DJ J DJ 2! 4! 24 (3.3.2 Feynman rules for QCD 139 Fig. e. In a similar way one can 2 obtain the other factors.

An important point is that fixing the gauge is usually not sufficient to totally remove the unphysical degrees of freedom from the theory (an exception is the axial gauge with n 2 < 0). Together with equations k · ε = 0. choose another vector n µ such that k · n = 0. this restriction does not prevent one from having unphysical states propagating as virtual states in the intermediate steps of perturbative calculations. In general. de Rafael (1979)). However.65) Of course. k) = −gµν + λ=1. Feynman rules for QED in the covariant gauge can be derived analogously.4(a): Diagram 3. ghosts. A physically sensible theory can be defined in Hilbert space restricted to the transverse photons (see. To make them unique we must sacrifice manifest Lorentz covariance and.4(b): ˜ abc Vµ = −gccab rµ (2π )4 δ(r + q − p) ˜ abc Vµ = −igγµ T a ji (2π )4 δ(r + q − p) (3.140 3 Feynman rules for Yang–Mills theories obtain in momentum space Diagram 3. k) uniquely. the usual on-shell Lorentz condition k · ε = 0 for spin-one massive particles is not enough to determine the photon polarization vectors uniquely. 2. k) = −δλ1 λ2 (3.67) n·ε =0 (3. k). for instance.2 n 2 kµ kν kµ n ν + kν n µ ≡ Pµν − n·k (k · n)2 (3.68) In field theory every particle is associated with a field with definite transformation properties under the homogeneous Lorentz group. k)εν (λ. A well-known example of their presence in the theory is the Gupta–Bleuler formulation of QED in the covariant gauge ∂µ Aµ = 0. The polarization sum reads ∗ εµ (λ. 3. Becchi–Rouet–Stora transformation Unitarity and ghosts A spin-one massless particle has only two physical polarization states εµ (λ. All Feynman rules are given in Appendices B and C. k)ε(λ2 . Since the three four-vectors kµ .64) (3. the . such degrees of freedom can be seen explicitly if we construct the Fock space of the theory: a longitudinal photon state and a negative norm scalar photon state are present. εµ (λ. for example.66) conditions (3.3 Unitarity. λ = 1. ε ∗ (λ1 . k) do not span four-dimensional space. Thus the four-vector gauge field Aµ (x) actually represents only two physical degrees of freedom. Indeed.66) specify the vectors ε(λ.

which reflects the presence of the redundant degrees of freedom. of (3. BRS transformation 141 Fig. Fig. be quite subtle as .h. The question then arises about the unitarity of the theory. however. like in the Feynman gauge.h. On the other hand we can calculate the absorptive part of the scattering amplitude from the r.70) where the sum is taken over all physical states for which transitions to the initial and final states are allowed by quantum number conservation laws.3 Unitarity.s. Here we sum over physical states only. The absorptive part of the scattering amplitude can be calculated perturbatively by means of the so-called Landau–Cutkosky rule: the contribution due to a Feynman diagram with a given intermediate state. gauge boson propagators are gauge-dependent.70) is just the imaginary part of the amplitude Tii and in the general case it is the absorptive part of Tif (see. ghosts. 3.70). Thus the unitarity condition demands that the unphysical intermediate states do not contribute to the amplitude when also calculated by means of the Landau–Cutkosky rule. for the gauge boson propagators in the Feynman gauge (see Section 3. ensures at the same time their decoupling. defined as Sif = δif + i(2π )4 δ( pi − pf )Tif satisfies the relation Tif − Tfi∗ = i n ∗ Tin Tfn (2π )4 δ( pi − pn ) (3.1) αβ Dµν = δ αβ k2 gµν → −2πiδ αβ gµν δ k 2 + iε (k0 ) (3. An explicit verification of this point may.5.71) It is clear that in gauges which do not remove the unphysical degrees of freedom. is obtained from the Feynman amplitude by replacing the propagators in this intermediate state by their imaginary parts. the unphysical intermediate states are included in our calculation. For instance.69) (3. for example.3.s. of the unitarity relation (3.5. Section 18. 3. For the forward elastic scattering the l. Bjorken & Drell (1965).12). The requirement that the S-matrix is unitary implies that the scattering amplitude T . In fact one can expect that gauge invariance.

7 with only physical gluons as the state n (Problem 3.70).73) where α is an arbitrary quantity. (3. the situation is more complicated as can be illustrated by the standard example of the fermion–antifermion scattering in the lowest non-trivial order in the gauge-coupling constant (Feynman 1977. However. amplitudes Tin and Tfn are given by the diagrams in Fig.6) calculated in a covariant gauge (3.39). the so-called ghost term (3. shown in Fig.s. 3.72) . we have learned in Sections 3. 3. Aitchison & Hey 1982). These are examples of Ward identities for a non-abelian gauge theory. In our specific example the cancellation occurs due to the contribution of the diagram with the ghost loop.19) does not satisfy ∗ the unitarity condition (3. leading to a perfectly unitary theory.1) shows that certain relations between the f f¯ → Aα Aβ and f f¯ → ηα ηβ amplitudes.8. for solving the equations of motion we need a gauge-fixing term which breaks the manifest invariance.142 3 Feynman rules for Yang–Mills theories Fig. An explicit perturbative calculation (Problem 3.2 that a consistent quantization of a non-abelian gauge theory requires the addition to the lagrangian of an extra term. are responsible for the cancellation.1).h.6. 3. Nevertheless in QED there is no problem due to the fact that photons always couple to conserved currents kµ M µν = 0 Therefore Pµν given by (3. The sum of the diagrams with only gauge particles in the intermediate state (Fig. It turns out that these additional unphysical degrees of freedom cancel the gauge field unphysical polarization states exactly. 3. where Aµ and η are the gauge and µ ν ghost fields.1 and 3.68) can be replaced by Pµν → −gµν + αkµ kν (3. respectively. where the r. In non-abelian gauge theories.

A(x)] 2α Seff = S + d4 x − 2 + η∗ Mη = S + SG + Sη (3. however. Provided one is able to express the gauge symmetry in a form involving both the classical and ghost fields one can introduce both fields from the very beginning as fundamental fields of a gauge theory and construct its lagrangian requiring invariance under this extended symmetry principle. Fig. Rouet & Stora 1974.7. has indeed been discovered. 3. Such a symmetry. It turns out that given any gauge symmetry one can always find symmetry transformations acting on gauge and ghost fields such that invariance under this symmetry is equivalent to gauge invariance of physical quantities.e. To introduce BRS symmetry (Becchi. BRS and anti-BRS symmetry We have introduced ghosts as a technical device to express the modification of the functional integral measure required to compensate for gauge-dependence introduced by the gauge-fixing term. that it is very natural to adopt the point of view that ghosts. the so-called BRS and anti-BRS symmetry. We shall first follow this historical path and later comment on the extension which is independent of the notion of a lagrangian and remains in the spirit of the above outlined programme. i. ghosts. The discussion in the previous subsection suggests. originally as a symmetry of the Faddeev–Popov lagrangians with specifically chosen gauge-fixing conditions. 3.3 Unitarity. fields with unphysical statistics.74) . 1976) let us consider the action including the gauge-fixing term and the ghost term 1 F α [x.3. BRS transformation 143 Fig.8. are necessary to compensate for effects due to the quantum propagation of the unphysical states of the gauge fields.

Thus it remains to show that δBRS (SG + Sη ) = 0 (3. A(x)] ρβ Dµ (z)δ(z − y) δ Aρ (z) µ d4 z (3. A(x)] where δ Fα is the variation of the functional Fα under a gauge transformation with β (x) replaced by gηβ (x) . y) ηβ (y) δ F α [x. µ s 2 Aα = s 2 µ (3. y) = and ρβ Dµ = ∂µ δ ρβ + gcρβγ Aγ µ d4 y Mαβ (x.76) δBRS ηα = + 1 gcαβγ ηβ ηγ ≡ sηα   2   δ η∗ α = +(1/α) F α ≡ sη∗ α  BRS where is a space-time-independent infinitesimal Grassmann parameter (note that it is not necessary to think of η and η∗ in (3. We note also that the measure D(A ηη∗ ) is invariant under the BRS transformations.78) This can be checked by a straightforward calculation.80) = s 2 ηα = 0 (but s 2 η∗ α = (Mη)α ).144 3 Feynman rules for Yang–Mills theories where (Mη)α = and M αβ (x.74) is invariant under the global transformation  αβ  δBRS Aα = Dµ ηβ ≡ s Aα µ µ     α α δBRS = −igT η ≡ (s ) (3.4).79) and η are nilpotent (see Problem 3.75) Thus (Mη)α = δ Fα [x. We observe that for the classical fields Aµ and the BRS transformation has the form of a gauge transformation with α (x) = +g ηα (x) (3.77) Therefore the BRS transformation leaves the original action invariant without the gauge-fixing and the ghost terms. The full action (3. .2) (3.39) as hermitean conjugates: see Problem 3. As intermediate steps we record the relation δBRS (Mη) = 0 and the fact that the transformations on Aα .

In this case the ghost field is a free real scalar field and the covariant derivative in the first equation in (3. The BRS transformation can be written down in a form more symmetric with respect to the fields η and η∗ if we introduce an auxiliary field b. After taking into account the equation of motion for the field b we see that (3. These anti-BRS transformations read  αβ s A α = Dµ η ∗ β ¯ µ      s η∗ α = 1 gcαβγ η∗ β η∗ γ ¯  2  α α αβγ ∗ β γ (3. BRS transformation 145 The BRS symmetric lagrangian can also be written down in QED. To be specific let us discuss the case of a covariant gauge.81) as our gauge-fixing condition.83) where the minus sign occurs if there is an odd number of ghosts and antighosts in A.3. sbα = 0 (3.86) . Using the auxiliary field b the BRS transformation for η∗ and b µ reads sη∗ α = bα . ghosts. This step is very important for promoting BRS symmetry to the fundamental symmetry of gauge theories.81) is equivalent to the standard gauge fixing term 2 −(1/2α) ∂ µ Aα .76) is replaced by the normal derivative (Problem 3.74) invariant provided the gauge-fixing functional F α is linear in fields.3).82) The action of s on an arbitrary function of the fields follows from their action on field polynomials s(AB) = (s A)B ± As B (3. This also allows formulation of the notion of BRS symmetry in a lagrangian-independent way (transformations (3.3 Unitarity. Using the freedom we have to choose any gauge condition without affecting the S-matrix we can take LG = bα ∂µ Aµ + 1 αbα bα α 2 (3.76) depend on the gauge-fixing term of the lagrangian). Ojima 1980) that in addition to BRS symmetry there is another symmetry which leaves the quantum Yang–Mills action (3.85) s η = −b + gc η η ¯   α αβγ ∗ β γ   s b = gc η b ¯    α ∗α s = −igT η ¯ and ss + ss = s2 = 0 ¯ ¯ ¯ (3.84) It has also been discovered (Curci & Ferrari 1976. The nilpotency of BRS symmetry can now be expressed as s2 = 0 (3.

quartic ghost interactions. which is simultaneously BRS and anti-BRS invariant.86). the latter is a consistent approach only for four-dimensional Yang–Mills theories when one chooses a linear gauge condition. in a BRS-invariant lagrangian four-ghost or higher ghost interactions are present which cannot be obtained from the ordinary Faddeev–Popov procedure. With the auxiliary field b interpreted as a Lagrange multiplier for the gauge-fixing condition. in such cases.5 this is discussed for a less general lagrangian.† In the general case. Indeed. four-ghost interaction terms should have been introduced at the tree level. η ] ¯ sη = − 2 [η. In terms of the fields.84) and (3. . which are matrix-valued in the Lie algebra of the gauge group. Also. the symmetry generators s and s become independent of the notion of ¯ the lagrangian and satisfy the nilpotency relations (3. One can also prove that full BRS invariance of the lagrangian leads to the gauge independence of physics. Thus we are led to construct the quantum lagrangian of a gauge theory as the most general s. absent in the Faddeev–Popov approach. Actually. one can show that starting from a set of infinitesimal gauge transformations building up a closed algebra (possibly with field-dependent structure functions) with a Jacobi identity one can always build an associated nilpotent full BRS algebra. based on the Faddeev–Popov lagrangian generating only quadratic ghost interactions. η] − b  ¯    ∗ sb = 0 s b = −[η .146 3 Feynman rules for Yang–Mills theories We have introduced the BRS and anti-BRS transformations as symmetry transformations of a Yang–Mills lagrangian with a gauge-fixing term linear in fields. have been shown to be necessary (Kallosch 1978. we have  s A µ = Dµ η s A µ = Dµ η ∗ ¯     ∗  s = −η s = −η ¯   1 1 ∗ ∗ ∗ (3.87) s η = − 2 [η . Specifying the general BRS-invariant lagrangian to this special case we indeed recover the lagrangian of Faddeev and Popov. in supergravity. e In fact. η]    sη∗ = b s η = −[η∗ .or/and s -invariant and ¯ Lorentz-invariant function (with ghost number zero) of physical and ghost fields. † In Problem 3. It is not just a formal improvement of the Faddeev–Popov approach since it also gives consistent quantum theories in cases for which the Faddeev– Popov method is inapplicable. Such couplings are in fact necessary for a consistent renormalizable Yang–Mills theory if the gauge-fixing functional F α is chosen to be non-linear in fields. Thus. shows that the four-point ghost function is divergent. one-loop calculation. b] ¯ An important next step is to promote full BRS (BRS or/and anti-BRS) symmetry to the fundamental symmetry of a gauge theory (Alvarez-Gaum´ & Baulieu 1983). This construction simultaneously generates gauge-fixing and ghost terms in the lagrangian.

2 Prove the invariance of (3. 3. Follow the same method in supersymmetric Yang–Mills theory (Falck.39) the Faddeev–Popov term Lη can be made hermitean.5 Check that the most general Yang–Mills and Lorentz scalar lagrangian which is BRSand anti-BRS-invariant and has ghost number zero can be written as follows (Baulieu & Thierry Mieg 1982) L = − 1 (Fµν )2 − 1 (Dµ )2 + ¯ iD / 4 2 + Vinv ( ) + s s ¯ 1 2 2 Aµ + Yukawa terms + εµνρσ Fµν Fρσ + 1 αb2 2 + βη∗ η + γ where α. Problems 3. (Use the relation for the structure constants which follows from the commutation relations T a .and anti-BRS-invariant lagrangians. γ are arbitrary gauge parameters and is the vacuum expectation value of the boson field . Check that due to the phase arbitrariness of the exponent in (3.) 3. i. 3. 3.76). When β = 0 one recovers the usual Faddeev–Popov lagrangian for linear gauges.1 Calculate in a covariant gauge the absorptive part of the amplitude corresponding to the sum of diagrams in Fig. Sterman.3 Check the invariance of the generalized QED lagrangian L = − 1 Fµν F µν + ¯ (iD − m) / 4 − (1/2α) ∂µ Aµ 2 − 1 (∂µ ω) ∂ µ ω 2 where ω(x) is a free scalar field.8 and check the unitarity condition (3. Townsend & van Nieuwenhuizen 1978).70).79) and (3. in particular.e.39) can be any anticommuting fields. they can be taken as real and imaginary parts of η. Check that.6 and Fig. They are naturally generated by determining the quantum lagrangian from the requirement of its BRS invariance. under the transformation δ = +ie ω(x) (x) ∂µ ω(x) δ Aµ = ∂ω = (1/α) ∂µ Aµ 3. 3.6 Use the superspace formulation of the Yang–Mills theory (Bonora & Tonin 1981. . as hermitean fields.80).Problems 147 Nielsen 1978. Thus. full BRS invariance indeed emerges as the fundamental symmetry of the quantum theory associated with any given underlying gauge invariance. 3. Hirshfeld & Kubo 1983) and in supergravity. T b = icabc T c with T a bc = −icabc .4 Note that the fields η and η∗ in (3.74) under transformation (3. β. Hirshfeld & Leschke 1981) to construct the BRS. Check (3.

reflecting the fact that. in general.1 Physical sense of renormalization and its arbitrariness Bare and ‘physical’ quantities Let us consider a theory of interacting fields defined by a given lagrangian density. A regularization procedure is necessary to make sense of the intermediate steps of perturbative calculations. for instance. most likely.1) cannot be used to define a theory? In the following we shall show that this conclusion is not necessarily true.1) are the regularization of divergent integrals and the subsequent renormalization of the parameters and the Green’s functions. the observable quantities are expressed in terms of experimentally measured parameters and our ignorance about the short-distance physics is hidden in those free parameters of the theory determined by experiment. After renormalization.4 Introduction to the theory of renormalization 4. some of the loop momentum integrations in the resulting Feynman diagrams are divergent in the UV. our theory does not properly describe the physics at very short distances. the coefficients of the perturbative series are. Two basic steps in interpreting a theory like (4. the ‘bare’ coupling constant λB is not a correct expansion parameter.1) We know already how to derive the Feynman rules which allow us to calculate the Green’s functions of this theory in the form of the perturbative expansion in powers of the coupling constant λB . The renormalization procedure accounts for the fact that. like λB and the ‘bare’ mess m B (as well as the Green’s functions of the ‘bare’ field B ) are not directly related to observable quantities. infinite. Consequently. let us take the case of a scalar field with quartic coupling L = 1 (∂µ 2 B) 2 − 1 m2 2 B 2 B − λB 4! 4 B (4. 148 . We shall now discuss this problem in some detail. Does this mean that the lagrangian (4. However. for definiteness. This is not so unnatural as it seems: the ‘bare’ parameters of the theory.

Consider thus the two-point Green’s function of the field G (2) (x. agree with the result of the perturbation theory calculation for ‘bare’ Green’s function: G (2) (x.4) F (x) note that 0| (0)|k is a constant from relativistic invariance.1 Physical sense of renormalization and arbitrariness 149 First let us recall the definition of the ‘physical’ field k| F (x)|0 F given in Section 2. If now follows from (2.192) = exp[i(k0 x0 − k · x)].3) (the last step can be easily checked by calculating the r.192) that ˜F G (2) (k) ≈ k 2 →m 2 F it i k 2 − m 2 + iε F (4. first). the Fourier-transformed two-point function near the mass-shell behaves as ˜ G (2) (k) ≈ k 2 →m 2 F i | 0| (0)|k |2 k 2 − m 2 + iε F (x) = (4. Therefore. This is convenient because we prefer to deal with Green’s functions and not field operators throughout.s.189) between the the one-particle intermediate state is then of the form (4. The on-shell truncated Green’s functions of the physical field have a straightforward physical interpretation: they are equal to the appropriate S-matrix elements.5) for the ‘physical’ field F does not. in general. y) = 0|T (x) (y)|0 and insert the complete set of states (2.4. y) = 0|T B (x) B (y)|0 (with some regularization introduced to keep B .h.2) s.7: (2. The normalization of a field operator (x) is directly related to the on-shell residue of its two-point Green’s function in momentum space. The contribution of d3 k [ 0| (x)|k k| (y)|0 (x0 − y0 ) (2π)3 2k0 + 0| (y)|k k| (x)|0 (y0 − x0 )] d4 k (k0 )δ(k 2 − m 2 ){ (x0 − y0 ) exp[−ik(x − y)] = F (2π)3 + (y0 − x0 ) exp[ik(x − y)]}| 0| (0)|k |2 i d4 k = exp[−ik(x − y)]| 0| (0)|k |2 (2π)4 k2 − m 2 + iε F (4.5) The result (4. k0 = (k2 + m 2 )1/2 F The ‘physical’ field F is thus defined by a normalization condition on the one-particle state created by F . Off-shell the equality also holds for analytic continuations.

. . . . the relation between the S-matrix and ‘bare’ Green’s functions involves the renormalization constant Z 3 .7) which implies a simple relation between the Green’s functions (connected or not): (n) G F (x1 . express the original field B in terms of the physical field F B k 2 →m 2 F Z3 k 2 − m 2 + iε F (4. We shall find instead that ˜B G (2) (k) ≈ where Z3 = 1 and is actually divergent in the absence of regularization. however. We can then attempt to express other physical Green’s functions as perturbation series in powers of λF . The ‘bare’ field B is thus not ‘physical’. The ‘bare’ Green’s functions by themselves are not observable. x n ) = Z3 n/2 (4. We expect that a relation between observables should be free from infinities. the ‘renormalization point’. . Suppose that we define the coupling constant λF as the value of the connected (4) proper vertex function ˜ F ( pi ) at some specified point. Let us denote the renormalization point by µ.10) (4) ˜B We can use the regularized bare perturbation expansion to calculate point.190) for the S-matrix elements. xn ) = Z 3 −n/2 G (n) (x1 . The result we shall write as (4) ˜ B |µ = at the same λB Z1 (4. x n ) (4. . the physical Green’s function. . . . in momentum space. the coefficients of the resulting perturbation expansion should be finite for observable quantities if the perturbation theory applies. xn ) B (n) B (x 1 .9) which also holds for the truncated Green’s functions that appear in the expression (2. In this way we would obtain the expression of one observable quantity. Consequently. We can. . . This is not yet the case in the ‘bare’ perturbation expansion in λB . . in terms of other observable quantities: λF and the physical mass. By definition we have (4) ˜ F |µ = λ F (4.11) . . the ‘physical’ Green’s functions are observable and must come out finite in perturbation theory. .150 4 Introduction to the theory of renormalization the loop integrals finite). . On the other hand. Consequently.6) = Z3 1/2 F (4. .8) and its counterpart for the connected proper vertex functions (n) F (x 1 .

between ‘bare’ and ‘physical’ quantities. Equivalently. expressed in terms of the scattering amplitude Ti f .12) The physical mass m 2 is defined as the square of the four-momentum of a freely F propagating particle (see (2.12) and (4. Defined as a pole of the scattering amplitude. Therefore the full propagator must have a pole at p 2 = m 2 F ˜ G (2) ( p) = R + ··· − m2 F p2 and the physical mass can also be defined as the value of p 2 for which the full propagator has a pole. m 2 is an F experimentally measurable parameter.1. (3. it then follows that the amplitude Ti f for the process i → f which can proceed via a one-particle intermediate state has a pole at p 2 = m 2 . we shall define m 2 as the position of the zero of the inverse F propagator (see (2.69) and (3. we 2 can calculate m F in the regularized bare perturbation expansion and define the mass renormalization constant Z 0 by the relation m 2 = (Z 3 /Z 0 )m 2 F B (4.9) we now obtain 2 λB = (Z 1 /Z 3 )λF (4.1 Physical sense of renormalization and arbitrariness 151 Fig.14) Altogether.14). In perturbation theory the general F structure of the amplitude with a one-particle intermediate state corresponds to the diagram in Fig. This definition is convenient because it ˜ avoids multiple poles present in G (2) ( p) calculated to any finite order. Z 1 and Z 3 corresponding to three relations.13) which can be either ‘bare’ or ‘physical’.70). (4. we have three independent renormalization constants: Z 0 . 4.4. Using (4.192)). Using these relations we can now rewrite the original . From the unitarity condition for the S-matrix.1. (4.7). which defines the new renormalization constant Z 1 . Again. 4.152)) ˜ (2) ( p 2 = m 2 ) = 0 F (4. where the central bubble denotes the full propagator.

B F −2 λB = Z 1 Z 3 λF (4. would give rise to divergences analogous to those of the ‘bare’ perturbation expansion. λB = Z 3 (λF + δλ). The perturbation theory based on this part of the lagrangian alone. It is this ‘physical’ perturbation theory and its generalizations described below that will be used in calculations: we shall not refer to the ‘bare’ lagrangian explicitly. After the cancellation of the regulator-dependent terms. and . and for the parameters of the lagrangian (couplings and masses) it is not always convenient. In this case. The former is. respectively. This is not necessary. unless subject to some regularization procedure. Thus. for example.152 4 Introduction to the theory of renormalization F lagrangian (4. a scalar field theory with two fields. renormalized by the fermion loop contributing to the proper 4 vertex function and this contribution is proportional to g 4 but not to λ. is more appropriate. The final important remark is the following.15) = Z3 1/2 F.16) The resulting ‘renormalized’ lagrangian we shall use as the basis for the formulation of the new perturbation expansion parametrized by the ‘physical’ quantities: m F and λF . The latter notation. If this procedure works in any order of perturbation theory with finite number of counterterms. in fact.15) we have divided the lagrangian into two parts. The counterterms can be properly defined only in a regularized theory. A physically important example is. Another important example is a theory of scalar and fermion fields. The first part is identical to (4. an additive −2 renormalization of λ. with the couplings λ 4 and g ) (Yukawa coupling). . −1 −2 2 2 2 m B = Z 3 (m F +δm ) and λB = Z 3 (λF +δλ) is. −1 m2 = Z0 Z3 m2 . Counterterms and the renormalization conditions In (4. λF 1 Z (∂ )2 − 1 Z 0 m 2 2 F F 2 3 µ F 2 2 − (λF /4!) 4 F F 2 − (Z 1 − 1)(λF /4!) 4 F F 2 4 B − (λB /4!) B = Z 1 (λF /4!) 4 = 1 (∂µ F )2 − 1 m 2 F 2 2 F 1 (Z 3 − 1)(∂µ F )2 − 1 (Z 0 − 1)m 2 F 2 2 − 1 m2 2 B (4.1) in terms of new fields L = 1 (∂µ 2 − + where B B) 2 and parameters m F . φ and . We expect that these divergences will be cancelled order-by-order if we take into account the second. we say that the theory is renormalizable. the regulator can be removed.1) with the original parameters replaced by physical ones. for instance. the renormalization of m 2 requires an additive term δm 2 which has a part proportional to M 2 and vice versa. Z i -dependent part of the lagrangian: the ‘counterterms’. with the masses m and M. We have introduced the renormalization constants as multiplicative constants. general enough to include both cases.

4. Observe that we now use m F and λF as free parameters rather than m B and λB : this is also more natural physically.12) and (4.14) is inconvenient because these relations involve ‘bare’ quantities.15) is assumed to be the ‘physical’ field. already used one such condition to define λF : it is (4. present in the Feynman rules of the ‘physical’ perturbation expansion. we shall fix the counterterms by imposing the ‘renormalization conditions’ on the physical Green’s functions. =0 F ( p) F ( p)| p 2 =m 2 F (4. We have.1 Physical sense of renormalization and arbitrariness 153 The theory defined by (4. if the perturbation theory parametrized by m F and λF is finite we would not expect to lose this property . can be determined from the renormalization conditions order-by-order.1) is renormalizable if considered in the space-time of dimension d ≤ 4. Z 1 and Z 0 by relations like (4. The renormalization constants Z 0 . The counterterms. (4. Z 1 and Z 3 are. in fact. of course. Now let us make the following important observation concerning the practical application of our results. To fix all counterterms we need two more conditions.17) F ∂ p2 F which implies (2) ˜ F ( p) −→ p 2 − m 2 F p 2 =m 2 F in accord with the fact that field and F in (4. However. which we choose as follows ∂ (2) ˜ ( p)| p2 =m 2 = 1 (4.18) where (2) ˜ F ( p) = p 2 − m 2 − F (4. not observable quantities and even if expressed in the new ‘physical’ perturbation expansion in terms of λF and m F are divergent order-by-order after the regularization is removed.10). Instead. From this point of view to define Z 3 . Arbitrariness of renormalization We have outlined the procedure which replaces the original ‘bare’ perturbation expansion with divergent coefficients by a new perturbation expansion in which the coefficients are finite. How unique is this procedure? We have used the physical observability of new parameters and Green’s functions as a guide in order to ensure that they are finite and in finite relation to each other.7). We shall always use the ‘physical’ perturbation expansion and never refer to the ‘bare’ parameters at any stage of calculations.19) This choice is exactly equivalent to our previous definitions.

‘physical’ scheme. In other words once we have specified the counterterms which cancel the infinities we can make a finite change in them. which we shall discuss in detail in the following sections. It is possible to do this transformation without affecting Z 3 or Z 0 : the field F is still given by (4.7) and therefore does not change. We can use any definition as long as the transition from λF to the new parameter λR is perturbatively finite: i.11). As m 2 is fixed and the physical content of the theory should not change we must B conclude that the new mass parameter. First of all we can change the renormalization point µ: µ → µ . also the renormalization constant Z 1 . must change: Z 1 → Z 1 . in powers of λF . however. Instead of F we can use the ‘renormalized’ field R provided that their relative normalization is perturbatively . Consider some examples. This renormalization symmetry of the theory is very important. the ‘renormalized mass’.21) where now (2) ˜ R = p2 − m 2 − R R ( p) (4. So far we have kept to the original. n-parameter class of different ‘renormalization prescriptions’ (n is the number of counterterms).154 4 Introduction to the theory of renormalization if we change the parametrization and use parameters which differ from m F and λF by a finite amount. One important example of a renormalization procedure which makes use of this generalization is the minimal subtraction scheme. The Green’s functions are still ‘physical’ but have a new perturbation expansion. As the bare parameters are fixed. and not only because it allows one to choose the most convenient renormalization prescription. it is the basis of the renormalization group and the renormalization group equations. Finally we can relax the requirement of the ‘physical’ renormalization of the two-point renormalized Green’s function. Let us now change the value of the mass renormalization constant Z 0 : Z 0 → Z 0 keeping Z 3 unchanged. We thus have an infinite. As a result. or vice versa. defined by (4. λR can be expressed in terms of λF in the λF -perturbation expansion. it will now have to be expressed in terms of the new parameter λF .20) is no longer the physical mass. defined as m 2 = (Z 3 /Z 0 )m 2 R B (4.e. Similarly. we do not have to require that the coupling constant parameter is equal (4) to the value of ˜ F at any renormalization point.18) by R ( p)|µ =0 (4. In terms of the ‘renormalization conditions’ discussed before this may correspond to replacing (4. the value of λF changes: λF → λF .22) and µ is some renormalization point.

. . .25) G (n) (x1 . . . (Z 1 /Z 3 )λR . . −kn ) 1 2 (k − m 2 ) F i i (4. Of course. One possibility is the ‘physical’ renormalization prescription described before.23) F R where Z R = Z 3 /Z 3 is finite in the perturbation expansion. xn ) R (4.27) which generalizes (2. (Z 0 − 1)m 2 . The counterterms are constructed according to a renormalization prescription.4. kn = lim ki2 →m 2 F i −(n+m)/2 ˜ (n+m) GR ×Z R (k1 .) + L (4. −k1 . . xn ) = Z R −n/2 (4. .29) The Feynman rules of the renormalized perturbation theory follow from the renormalized lagrangian (4. . . . equivalently ∂ (2) 1 ˜ (k)|k 2 =m 2 = 2 R F ∂k ZR Due to the relation (n) G F (x1 .26) the factors of Z R will also appear in the relation between the renormalized Green’s functions and the S-matrix elements. km |S|k1 . km . . (Z 0 /Z 3 )m R . A .28). . . with Z dependent coupling constants: (Z 3 − 1) p 2 . the lagrangian (4. . i. λR .) 2 R . . This does not make the renormalized Green’s functions very ‘unphysical’ because the Z R factors are finite order-by-order. .1 Physical sense of renormalization and arbitrariness 155 finite.24) 2 →m 2 k 2 − m 2 + iε k F F or. Note that R the counterterms are treated strictly perturbatively: when calculating the nth order correction it is enough to know the counterterms up to this order only. . . . . The relation between the S-matrix and the renormalized Green’s functions has the following form k1 .e. we have R = ZR 1/2 F (4.190). (Z 1 − 1)λR . The counterterms provide vertices. .28) theory R) 2 L = 1 (Z 3 − 1)(∂µ 2 − 1 (Z 0 − 1)m 2 R 2 2 R − 1/4!(Z 1 − 1)λR 4 R (4. . . .15) can be rewritten in terms of the new renormalized fields and parameters L = L(Z 3 = L( where in the λ 4 2 2 R . . m R . The value of Z R can be determined if we know the on-shell residue of the two-point renormalized Green’s function iZ R ˜R G (2) (k) ≈ (4.

denoting −i R (k). We use this condition R F ˜R in preference to the one involving the propagator G (2) (k). to any finite order. This is the case in the minimal subtraction scheme. Final remarks Let us finish this section with some remarks concerning the calculation of the scattering amplitudes in the renormalized perturbation expansion. This procedure is particularly convenient because the counterterms can be determined from the divergent parts of the diagrams alone.3. we determine the physical mass m 2 from the requirement that the inverse F (2) propagator ˜ R (k) = k 2 − m 2 − R (k) vanishes at k 2 = m 2 .s. We shall discuss this scheme in detail in Section 4. To obtain a single pole one must sum the whole geometric series (4. For instance ( R is defined by (4. after the introduction of dimensional regularization. We see that. The reason is that the ˜R perturbation expansion for G (2) (k) has the form ˜R G (2) (k) = i i i i + 2 [−i R (k)] 2 + 2 [−i 2 2 2 − mR k − mR k − mR k − m2 R i i × [−i R (k)] 2 + ··· = 2 2 2 k − mR k − m R − R (k) k2 R (k)] 2 k i − m2 R (4.h. (2) Assume that ˜ R (k) calculated up to order n in some renormalization scheme is regular at k 2 = m 2 and expand F R (k 2 )= 0 + (k 2 − m 2 ) F 1 + (k 2 − m 2 )2 F 2 + ··· (4. Another possibility is to assume some trial form for the counterterms and adjust the parameters to make the renormalized Green’s functions finite order-by-order.22)) ∂ (2) ˜ (k)|µ = 1.31).32) where m 2 is understood as the value of k 2 at which the nth order approximation to F . ˜ G (2) (k) has multiple poles at k 2 = m 2 instead of a single pole at k 2 = m 2 required R F by unitarity.30) Clearly the number of the equations must be equal to the number of the renormalization constants.31) corresponding to the diagrams with the blobs on the r. (4) ˜ R (k)|µ = λR (4. ∂k 2 R R (k)|µ = 0.156 4 Introduction to the theory of renormalization more general case corresponds to specifying finite values for some renormalized Green’s functions at a renormalization point. As explained before.

2 Classification of the divergent diagrams Structure of the UV divergences by momentum power counting In the previous section we have described the procedure of renormalization. They appear to have the same form as the counterterms induced by the renormalization described in Section 4. We shall often refer to this rule in one-loop 2 calculations. which are of the same general structure as vertices of the original lagrangian. km . −k1 . . . .25) ZR = Consequently.1. We obtain F R 1 (2) ˜ R (k) = (k 2 − m 2 )[1 − F − (k 2 − m 2 ) F 2 − · · ·] (4. .35) In the lowest non-trivial order (one-loop approximation) we can equivalently replace (1 − 1 )−(n+m)/2 by (1 + 1 (n + m) 1 ) so that we then have the correction 2 (+ 1 1 ) for each external line.4. However. . also called the 1PI Green’s functions.33) and.2 Classification of the divergent diagrams (2) ˜ R (k) vanishes. . we have not given the proof that this procedure actually leads to finite results: we still have to show that a theory like λ 4 is renormalizable. −kn ) ˜ = (1 − 1 )−(n+m)/2 G (n+m) R(trunc) (4. are enough to cancel UV divergences due to loop momentum integrations. . In other words. using (4. . . hence 0 157 = m 2 − m 2 . . kn 1 2 (ki − m 2 ) (1 − = lim F 2 →m 2 i ki F i 1 1− (4. We shall concentrate on the structure of UV divergences which appear in Feynman diagrams of λ 4 perturbation theory (with counterterms not yet included) and use the results to determine the form of counterterms from the requirement that these divergences are cancelled.34) 1 (n+m)/2 1) ˜R ×G (n+m) (k1 . 4. (4. The basic blocks of a renormalization programme are the connected proper vertex functions (n) . .27) becomes k1 . Let us recall that these are truncated Green’s functions which remain connected when an arbitrary internal line is cut. . In a renormalization programme these are important for two . Some examples from QED and λ 3 theories will also be considered. Here we shall give only a general idea of the proof. km |S|k1 . we should still prove that the counterterms induced by the renormalization.

any other Green’s function is made up of 1PI Green’s functions joined together to form a tree. each corresponding to a d4 ki momentum integration.158 4 Introduction to the theory of renormalization reasons. This is because the connected proper vertex functions can themselves be understood as generalized vertices of an effective lagrangian which includes all loop corrections: the generating functional [ ]. → ∞. states that a Feynman integral converges if the degree of divergence of the diagram as well as the degree of divergence associated with each possible subintegration over loop momenta is negative.36) where L is the number of independent loops. The integrand will then behave as a power of .2 the subdiagrams (b) and (c) correspond to the subintegrations over l and k. For a diagram with V vertices and I internal lines the number of independent loops is L = I − V + 1. For example. We can introduce the superficial degree of divergence D S of this subintegration: D S = 4L S − N S . Consequently. This fundamental result will be the basis of our discussion. One defines the so-called superficial degree of divergence of a Feynman diagram D = 4L − N (4. Suppose that all integration momenta become large simultaneously: ki → ki . say −N . A theorem due to Weinberg. The UV divergences we are concerned with come from the region of integration where some of the integration momenta are large. all loops and therefore all divergences are present already in the 1PI Green’s functions. It is easy to see that D S can also be regarded as a superficial degree of divergence of a subdiagram built up of all the internal lines of the complete Feynman diagram with momenta depending on those in S: L S is then the number of loops in this subdiagram. respectively. The second reason is that the identification of the necessary counterterms which are to be interpreted as additional vertices in the renormalized lagrangian is particularly simple if one looks at the divergences of the 1PI functions. Some information about the possible structure of these divergences can be obtained by the simple procedure of counting the powers of momentum in the Feynman integrand. where L S is the number of integration momenta belonging to S. so that the momentum integrations in the two distinct parts are independent of each other. The integrand will now be proportional to another power of : −N S . First. In the above example diagram (a) has D = 2 while (b) and (c) both have D = 0. . in the case of the two-loop diagram (a) in Fig. each loop included entirely in one of the 1PI parts. The first thing to note is that in λ 4 theory (and any other interacting quantum field theory in four dimensions) diagrams with D ≥ 0 are present. Let us next assume that only some subset S of the integration momenta is large and of the order . 4. In this way we can assign a subdiagram and a degree of divergence to each loop subintegration.

[ ] = [m](n−1)/2 (4. In this formula the restriction to 1PI diagrams is due to the fact that propagators corresponding to the internal lines which do not belong to loops are independent of the integration momenta and consequently do not contribute to N : in 1PI diagrams such lines are absent. Then.37) and (4. which are.38) Here IB . Classification of divergent diagrams To find out what counterterms are required we must examine all 1PI Green’s functions with D ≥ 0 which appear in the theory.2 Classification of the divergent diagrams 159 Fig. for any 1PI diagram (4. Consider the case of a scalar field theory without derivative terms in the interaction (λ 4 is one example). It ∂ follows that [ ] = [m](n−2)/2 .37) which is applicable to the case of fermion QED. We can also write a generalization of (4. The argument is based on dimensional analysis so let us first recall the dimensions of some quantities in quantum field theory. Some . The canonical dimensions of Bose and Fermi fields are determined from the requirement that the kinetic terms in the action. Note that by replacing 4L with n L we obtain a generalization of (4.36). of the form dn x ∂ 2 and dn x ¯ i/ . while length and mass are measured in inverse units: [x]−1 = [m]. IF are the numbers of photon and fermion internal lines.37) where I is the number of internal lines. We shall see that D is determined by the number of external lines of the diagram.2.38) to the case of an n-dimensional space.39) In the same manner the dimensions of a coupling constant follow from the requirement that the corresponding term in the action is dimensionless. respectively. 4. respectively. (4. are dimensionless. We set h = c = 1 so that the ¯ action integral must be dimensionless. again with restriction to 1PI diagrams D = 4L − 2IB − IF (4.4.36) reads D = 4L − N = 4L − 2I (4.

40) where [m]−n is the dimension of the dn x integration measure. One is that in any theory with coupling constants all of positive dimension in the units of mass (λ 3 for example) the number of 1PI diagrams with D ≥ 0 is finite. A theory like this is called super-renormalizable.42) (also see Problem 4. Here (a).160 4 Introduction to the theory of renormalization examples are λ λ 3 4 : : [m]−n [λ][m]3(n−2)/2 = 1. As all contributions to a given Green’s function must be of the same dimension it then follows that with dimensionless couplings. The exception is the possible lowest order no-loop contribution. = K for K fermion mass insertions or cubic scalar couplings) D = 4− B − 3F − 2 (4.4). The canonical dimension of a 1PI diagram with B external boson lines and F external fermion lines is equal to the dimension of a coupling constant corresponding to the vertex of this structure. respectively. as 1/k 2 and 1/k. In particular. (d) is a constant ‘tadpole’ which contributes to the vacuum expectation value of the field and can be disregarded in the = 0 case (diagrams of this kind will be important in theories with spontaneous symmetry breaking) and (e) produces an infinite mass renormalization counterterm which is the only divergent counterterm in λ 3 theory with = 0. there are no D ≥ 0 1PI . the superficial degree of divergence of a diagram is determined solely by the external lines.41) As explained.e. i. the superficial degree of divergence coincides with the dimension of the diagram in the unit of mass.3. In four dimensions / the quartic (λ 4 ) and QED (g ¯ A ) coupling constants are dimensionless while the cubic (λ 3 ) coupling constant has the dimension of mass. Then. for any 1PI Feynman diagram built up of vertices with dimensionless coupling constants only. / g ¯ A : [m]−n [g][m]2(n−1)/2+(n−2)/2  [λ] = [m]3−n/2   [λ] = [m]4−n   = 1. [g] = [m]2−n/2 (4. Assume that at large momenta the Bose and Fermi propagators behave. (b) and (c) are irrelevant vacuum diagrams: they cancel with the normalization factor of our Green’s functions. 4. This formula has some important consequences. which is [m]4−B−3F/2 in four dimensions. In the case of the λ 3 theory the diagrams with D ≥ 0 are as shown in Fig. this applies to 1PI diagrams with dimensionless couplings only. Consequently D = 4− B − 3F 2 (4. If the diagram also contains vertices with dimensional coupling constants of total dimension in the units of mass (for example. the single elementary vertex for which the superficial degree of divergence is undefined. [m]−n [λ][m]4(n−2)/2 = 1.

Let us now return to theories with dimensionless coupling constants. any coupling constant of negative dimension. among others. so that (4. For the λ 4 theory and for QED we have the D ≥ 0 connected proper vertex functions shown in Fig. 4. 4.3. Consider a 1PI Feynman diagram with D ≥ 0. Let { f } D denote the sum of first (D + 1) terms of a Taylor expansion of the Feynman integrand f in powers of the external momenta with D equal to the superficial degree of divergence of the diagram considered.4. An important example is Fermi’s theory of weak interactions. like λ 4 in more than four space-time dimensions or G( ¯ )2 in four-dimensional space. diagrams with three external lines and consequently no divergent coupling constant renormalization counterterm. This can be achieved by making a subtraction in the Feynman integrand. Each 1PI Green’s function requires new counterterms and new renormalization conditions: consequently. the theory is non-renormalizable.2 Classification of the divergent diagrams 161 Fig. Necessary counterterms To make the divergent diagrams finite we must find a way to lower the associated degrees of divergence. On the other hand in a theory which involves. the perturbation expansion of any 1PI Green’s function must include D ≥ 0 diagrams when the order is high enough. If we replace the integrand of this diagram by a subtracted expression f − { f }D (4.41) applies: all diagrams contributing to a specific 1PI Green’s function have the same degree of divergence.43) .4 (the triple photon proper vertex also has D = 1 but it vanishes because the photon has odd charge conjugation).

4. We conclude that the subtractions necessary to lower the degree of divergence of the D ≥ 0 Green’s functions are . 4. A similar procedure should be applied to divergent subdiagrams. however. We then have two D ≥ 0 Green’s functions: ˜ (2) ( p) with D = 2 and ˜ (4) ( p1 . . proportional to pµ . This produces two counterterms Ap 2 + B The third one. With some procedure for the regularization of divergent integrals we can calculate the Feynman integrals of the two terms in (4. p4 ) with D = 0. vanishes because of relativistic invariance: all counterterms must be scalars. or a combination of counterterms. . from Weinberg’s theorem. Consider the case of the λ 4 scalar field theory. As the second term is a polynomial in the external momenta. The four-point function has D = 0 and only one (constant) counterterm which corresponds to the coupling constant counterterm. Obviously the counterterms will diverge when the intermediate regularization is removed. after the integration it will produce a polynomial ‘counterterm’. . In a renormalizable theory we should be able to understand these subtractions as due to the inclusion of counterterms. the superficial degree of divergence will become negative. . We recognize the counterparts of the wave-function renormalization constant ( p 2 is equivalent to −∂ 2 ) and the mass counterterm.162 4 Introduction to the theory of renormalization Fig. The twopoint function has D = 2 and consequently requires subtraction of the first three terms of the Taylor expansion of the integrand.43) separately.43) to ultimately give a finite result provided that we have dealt with all divergent subdiagrams. they must combine with the other part of (4.

Can these counterterms.2 Classification of the divergent diagrams 163 Fig. In each of these we make the appropriate subtractions.7. where . We can envisage the following procedure. also deal with all divergences due to the D ≥ 0 subdiagrams? Again (in the λ 4 case) these are the subdiagrams with two (D = 2) or four (D = 0) external lines. We first identify all lowest order (one-loop) divergent subdiagrams.2 overlap each other. This is because we make the subtraction in the indicated subdiagrams. this argument runs into difficulties: one problem is that we cannot always make subtractions independently in different subdiagrams. 4. The effect of these subtractions should be equivalent to adding to the original diagram all diagrams obtained by replacing some of the considered lowest order subdiagrams by the corresponding lowest order counterterms.6. and not in complete one-loop vertex functions which are shown in Fig. Unfortunately. subdiagrams (b) and (c) in Fig. so that no new types of counterterm are apparently required. Another example of such ‘overlapping divergences’ is the QED diagram shown in Fig.5. Note that in this case the dot is not the complete lowest order coupling constant counterterm.4.6. Consider a diagram with no counterterm vertices which includes some D ≥ 0 subdiagrams.5. if included in the interaction lagrangian. This procedure should then be repeated with respect to higher order D ≥ 0 subdiagrams and counterterms until all divergences are removed. The boxes indicate the divergent subdiagrams we are considering. 4. 4. 4. One example is shown in Fig. Fig. 4. For instance. equivalent to including the counterterms of structure as implied by renormalization of the original lagrangian. 4.

the BPHZ construction (Bogoliubov– Parasiuk–Hepp and Zimmermann).1 since this method of renormalization is particularly convenient in applications of the renormalization group equation: this is the so-called mass-independent renormalization. The problem of overlapping divergences requires more detailed treatment and we shall not discuss it here.28).3 λ 4 : low order renormalization Feynman rules including counterterms In this section we introduce in some detail the dimensional regularization method (’t Hooft & Veltman 1972.164 4 Introduction to the theory of renormalization Fig. It also shows diagrams corresponding to the counterterms present in lagrangian (4. 4. Bollini & Giambiagi 1972) and the minimal subtraction renormalization scheme using the λ 4 theory as an example.1. Renormalization constants Z i will be determined by the renormalization conditions in each order of the perturbation theory Zi = 1 + n (Z i − 1)(n) . We must also note another approach. In this case the subtractions are performed systematically by means of an ‘R-operation’ which is based on a somewhat different identification of subdiagrams. the D = 0 divergent subdiagram can be identified in two distinct ways. All counterterms are treated as interaction terms since we want to use counterterms order-by-order in the perturbation theory. The Feynman rules for the λ 4 theory are collected in Table 4. namely the renormalization constants Z i do not depend on mass parameters.7. We choose the minimal subtraction scheme to illustrate our general consideration of Section 4. the ultimate answer is positive: the counterterms in the renormalized lagrangian do indeed cancel all UV divergences. 4. However.

p + q + r + s = 0 symmetry factors S S= S= S= 1 2 1 6 1 2 S= 1 4 vertex counterterm −iλ(Z 1 − 1)(n) mass counterterm wave-function renormalization counterterm −i(Z 0 − 1)(n) m 2 +i(Z 3 − 1)(n) p 2 where (Z i − 1)(n) = f i(n) λn It is clear from the previous section that the main task for the lowest order renormalization is to calculate the proper (1PI) Green’s functions corresponding to the three divergent diagrams of Fig. with Fig.8. The calculation will be based on the dimensional regularization method and we introduce it now briefly. propagator loop integration p2 i − m 2 + iε d4 k (2π )4 vertex −iλ. We begin with vertex renormalization.4.e. In four space-time dimensions this .1.3 λ 4: low order renormalization 165 Table 4. i.8(b). 4. 4.

the dimensionally regularized theory (4. The kinematics are defined in Fig. The first diagram of Fig.9 gives the following expression for the amplitude: A1 = 1 2 i dn k i (−iλµε )2 n k 2 − m 2 + iε (k − P)2 − m 2 + iε (2π) (4. In n = 4 − ε dimensions the coupling constant λ is dimensionful. Fig.166 4 Introduction to the theory of renormalization Fig. is logarithmically divergent ∼ d4 k/k 4 . Calculation of Fig. cancelling all terms divergent when ε → 0 by properly adjusting the renormalization constants Z i (renormalization) and continuing the result analytically to four dimensions. Thus.9. 4. 4. where µ is an arbitrary mass scale and λ is dimensionless.8(b) We first calculate the diagram shown in Fig. so it is convenient to write it as µε λ.1) is defined by the action S= dn x 1 (∂ φ)2 2 µ − 1 m2φ2 − 2 λµε 4 φ + counterterms 4! where n = 4 − ε. 4. This suggests the possibility of considering the analytic continuation of the amplitude to 4 − ε dimensions (regularization). However. 4.9. 4. The Feynman rules in Table 4.8. if we do the integration in 4 − ε dimensions the result is finite.44) .1 are changed accordingly. 4.8(b) in 4 − ε dimensions and then discuss some subtle points of the method.

We can change the contour to integrate over the imaginary axis. since k0 = ±[k2 − sx(1 − x) + m 2 − iε]1/2 . We introduce the standard Feynman parameters 1 = ab 1 0 dx [ax + b(1 − x)]2 (4. We note that the poles of the integrand in the k0 -plane are located as shown in Fig. Changing variables ik0 → k0 dk0 → −i dk0 2 2 k0 − k2 → −k0 − k2 −i∞ +i∞ dk0 → +i ∞ −∞ dk0 one gets the following expression A1 = 1 λ2 µ2ε 2 1 dx 0 dn k (+i) (2π )n [k 2 − sx(1 − x) + m 2 − iε]2 (4.47) The integration over dk0 dn−1 k can be easily done after the Wick rotation.48) .10. We then get A1 = 1 λ2 µ2ε 2 1 (4. 4.45) and change the variables as follows k − Px → k which is legitimate for a convergent integral.46) dx 0 dn k 1 n [k 2 + sx(1 − x) − m 2 + iε]2 (2π ) (4. 4.3 λ 4: low order renormalization 167 Fig.10.4.

The remaining contribution is finite and reads. First.51) The presence of the pole term 1/ε reflects the divergence of the integral in four dimensions. Note also the presence in (4.577 is Euler’s constant.50) To proceed we require the expression (x) −→ 1/x − γ + O(x) x→0 where γ = limn→∞ (1 + 1 + · · · + 1/n − ln n) = 0.52) The inclusion of the crossed diagrams gives the final result A1 = i(λµε )λ 3 1 + A1 (s) + A1 (t) + A1 (u) 16π 2 ε ε (4.49) dx 0 m 2 − sx(1 − x) − iε µ2 −ε/2 (4. after integration. consider the space with one time-like .168 4 Introduction to the theory of renormalization where k is a vector in n-dimensional Euclidean space. and we 2 also use the expansion a ε = exp(ε ln a) = 1 + ε ln a + O(ε 2 ) In the limit ε → 0 we get m 2 − sx(1 − x) + γ − ln 4π + O(ε) µ2 0 (4.50) and (4.51) of the factor λµε (not expanded in ε) since this is the coupling constant in n dimensions and it enters into the definition of the counterterm −iλµε (Z 1 − 1). A1 = iλµε 1 λ 1 λ − iλµε 16π 2 ε 2(4π )2 dx ln A1 (s) = −i m2 (λµε )λ γ − ln 4π − 2 + ln 2 2(4π)2 µ 1 4 +2 − m /s 2 1/2 ln 1 2 1 2 + − 1 4 1 4 − m 2 /s − m 2 /s 1/2 1/2 (4. The next step is to use the formula (in Euclidean space) dn k 1 1 bn/2−α (α − 1 n) 2 = (2π)n (k 2 + b)α (4π )n/2 (α) which leads us to an almost final result A1 = 1 (λµε )i 2 ( 1 ε) λ 2 (4π )n/2 (2) 1 (4.53) Comments on analytic continuation to n = 4 dimensions Before proceeding further a comment on the explored analytic continuation to n = 4 dimensions should be made.

55) In = − 1 I n−4 n where In exists for 1 < n < 5. 2. In addition we would like to be able to continue In to values n > 4. 6. . The above procedure may be repeated to show that In is of the form In = 1 (4 − n) I˜n (4. . .54) can be used to define an analytic function In in the range 1 < n < 4 which coincides with the original In for n = 2.56) 2 where I˜n is well behaved for arbitrarily large n. 0 π sinn−3 θn−2 dθn−2 = 2π (n−1)/2 1 (n − 1) 2 Integral (4. 8. by partially integrating N times over dr 2 one can find explicit expressions for the analytic continuation of In to (1 − 2N ) < n < 4 dimensions. For 1 < n < 4 the surface terms are zero and one gets In ∼ ∞ −∞ ∞ dk0 0 ∞ −∞ dr r n−2 ∞ 2 (−2k0 + 2r 2 ) (k 2 + b)3 1 + 2 so that dk0 0 dr r n−2 (n − 2) = In + 1 (n − 2)In 2 (k 2 + b)2 (4. . 3 (the solid angle for a non-integer n is the analytic continuation of n−1 . . This can be achieved using the identity I = d 1 d k0 + r 2 dk0 dr and by integrating each term by parts over k0 and r . The basic integral In = dn k 1 (k 2 + b)2 makes sense for n = 1. Furthermore. respectively. So one has constructed an analytic function In with simple poles at n = 4. Dimensional regularization assumes the physical theory to be given by In for n → 4. 2 k 2 = k0 − r 2 (4.4.54) where d n−1 is an element of the solid angle in (n − 1) dimensions = 0 2π π n−1 dθ1 0 sin θ2 dθ2 sin2 θ3 dθ3 .3 λ 4: low order renormalization 169 dimension and n − 1 = 3 space-like dimensions. . . n integer). 3 and can be written as In = 1 2 ∞ −∞ ∞ dk0 0 dr 2 (r 2 )(n−3)/2 1 (k 2 + b)2 π 0 dn−1 .

The amplitude of Fig.53) can be cancelled by the vertex counterterm in Fig. Lowest order renormalization Coming back to the result (4. 4. The other two diagrams of Fig.58) The divergence is cancelled by the mass counterterm with ≡ (−i)(Z 0 − 1)(1) m 2 1 λ (4.8(c) requires both mass and wave-function renormalization. i.57) ε 16π 2 and the renormalized (finite) amplitude of Fig.e. by properly adjusting Z 1 .8 require mass and wave-function renormalization.11.8(a) reads 1 (−iλ)µε 2 dn k i (2π)n k 2 − m 2 + iε 1 (4π )ε/2 (m 2 )1−ε/2 1 ε − 1 = 1 (−iλ)µε 2 2 2 (4π) 2 2 2λm 1 1 λm m2 − i = 1i −1 + γ − ln 4π + ln 2 2 (4π)2 ε 2 (4π )2 µ (4.11.53).170 4 Introduction to the theory of renormalization Fig.8(b) is given by A1 (s) + A1 (t) + A1 (u) of (4. 4. The minimal subtraction (MS) prescription assumes that Z i have only (multiple) pole terms in ε and the residua are fixed to cancel all the divergences for ε → 0.53) one can now renormalize the amplitude by including counterterms. The so-called MS scheme in which (Z 1 − 1)(1) = (Z 1 − 1)(1) = 1 3λ 2 − γ + ln 4π 2 16π 2 ε is also often used. 4.59) (Z 0 − 1)(1) = 2 16π ε in the minimal subtraction renormalization scheme. 4. The amplitude of Fig.12. It is of order λ2 and therefore in order-by-order renormalization one must include simultaneously the other diagrams of order λ2 given in Fig. . With such a prescription one then gets to order λ 1 3λ (4. 4. It is clear from the structure of the counterterm that the divergences of (4. 4. 4.

(iii) both k and l large.12(b): iλ2 Diagram 4.12(a)–(c) should be then cancelled by new λ2 terms of the mass and wave-function renormalization counterterms ≡ −i(Z 0 − 1)(2) m 2 ≡ +i(Z 3 − 1)(2) p 2 (notice that in λ 4 theory (Z 3 − 1)(1) = 0).12(a): iλ2 Diagram 4. 4. The calculation of amplitude (4.12(a)–(c).59) it is easy to get the following (we write down explicitly only pole terms): Diagram 4. We tabulate first the results for Fig. (ii) k fixed. The structure of pole terms turns out to be the following I = 1 λ2 µ2ε i 6 I =i λ2 1 m2 1 2 m2 + − 2 + 2 m 2 ln p + (γ − 3 )m 2 2 (16π 2 )2 ε ε 4π µ2 12 .8(c) and those of Fig. 4.12(c): (−i)λ2 m2 2 1 2γ m2 1 2 − ln + − ε 4πµ2 ε ε (16π 2 )2 2 ε2 + O(ε0 ) (4.60a) The only non-trivial contribution is due to Fig.61) which is divergent for (i) l fixed. k large. l large.60c) 1 γ m2 1 2 m2 − ln − 2 )2 2 ε 2 ε 4πµ2 ε (16π + O(ε0 ) (4.12.60b) m2 1 1 γ m2 3 2 − ln + − 2 )2 2 ε 2 2 ε 4πµ ε ε (16π + O(ε0 ) (4. 4.61) is lengthy (Collins 1974). The amplitude reads 1 1 dn k dn l 1 n (2π)n k 2 − m 2 + iε l 2 − m 2 + iε ( p + l − k)2 − m 2 + iε (2π) (4. 4. Using (4. 4.4. The divergences which remain after adding the amplitude of Fig. This is what is called an overlapping divergence because several overlapping loop subintegrations lead to a divergent result.57) and (4.8(c).3 λ 4: low order renormalization 171 Fig.

i. they cancel in the final result for constants Z i and we see explicitly that at least in that order of perturbation theory the minimal subtraction scheme is indeed a mass-independent scheme.13.59) with (4.60) one gets i so that (Z 0 − 1) (Z 3 − 1) (2) 1 1 2 11 2 1 λ2 p − m 2m 2 + 2 )2 ε 2 (16π ε 12 ε2 λ2 11 2 = − 2 )2 ε 2 (16π 2ε λ2 1 =− 12(16π 2 )2 ε        (4. Consider the diagrams of Fig. 4.63) (4. 4.66) 2 i. it should be remembered that one-particle reducible diagrams do not require new counterterms.62) one gets the following results for Z 0 and Z 3 (to order λ2 ) Z0 = 1 + λ2 λ 1 11 2 − + 16π 2 ε (16π 2 )2 ε 2 2 ε 1 λ2 Z3 = 1 − 2 )2 ε 12(16π (4. We observe that .e. Z m = Z 0 /Z 3 B R λB = Z λ λR .e.65) Notice that there are logarithms of mass in the residua of the poles in each diagram. Firstly. Adding (4. However.172 4 Introduction to the theory of renormalization + (1) + (1) + (1) (1) Fig. We see in particular that to order λ2 Zm = 1 + 1 2λ2 λ λ2 1 5 + 2 − ε 16π 2 12 (16π 2 )2 ε (16π 2 )2 (4. We end this section with two comments. Z λ = Z 1 /Z 3 (4.13.64) In the literature one can also often find the constants Z m and Z λ defined by the relations m 2 = Z m m 2 . Let us illustrate that fact with one example.62) (2) Combining (4.

The number of free parameters is related to the number of so-called counterterms in the lagrangian. for the two-loop diagram alone we have ≡ (1/ε 2 ) f 12 + f 22 + ε2 f 32 + 2(1/ε) f 1 f 2 + 2 f 1 f 3 + 2 f 2 f 3 ε Discarding poles in the last result one obtains f 22 + 2 f 1 f 3 = f 22 which is wrong because we insist on unitarity being satisfied order-by-order in perturbation theory and therefore the finite part of the two-loop diagram should read f 22 as can be seen by considering the imaginary part of the amplitude. Diagrams of Fig. mass loop diagrams ≡ (1/ε) f 1 + f 2 + ε f 3 (1) ≡ −(1/ε) f 1 2 so that in the second order + (1) ≡ f 22 However. Consider. The second remark is devoted to the fact that the finite parts of Feynman integrals are not. It is very important to realize that this number may be finite . we are still able to predict results of measurements in terms of other measurable quantities. this problem is also present in classical field theories. typical for local field theories. In fact. obtained by simply discarding poles and multiple poles at n = 4 but have to be worked out by order-by-order renormalization procedure.70).4 Effective field theories The basic tool for calculations in quantum field theory is perturbation theory. in general. The predictive power of a theory is retained due to the procedure of renormalization: divergent contributions are absorbed into free parameters of the theory. which are taken from experiment. Thus. for example. In perturbative calculations we face the problem of divergent results.e. i.13 = + (1) 2 which illustrates our statement.4 Effective field theories 173 the following relation holds. 4. to absorb all infinities.4. see (3. 4. which are required for the procedure of renormalization.

e. an effective pion nucleon theory which respects the chiral symmetry of QCD but is not derivable perturbatively from it. In this case. For a theory to be renormalizable in this traditional sense. however. then physics at m is described by the ‘renormalizable’ part. the corresponding lagrangian may contain only operators of dimension four. that it provides a tool for easy resummation of large logarithmic corrections O(α n lnn (E/MW )) (Chapter 12) to weak processes. the virtue of the effective theory is not only that it offers a quickly truncated expansion in powers of E/MW but. These are described by the weak effective lagrangian with dimensionful coupling. A different situation occurs when we try to describe strong interactions of hadrons at low momentum transfer. it cannot contain couplings with the dimension of negative powers of mass. We have described here a situation in which the complete renormalizable theory at the scale M is known and the effective theory can be derived from it by perturbative methods. E ∼ 1 GeV. This certainly makes such a theory very predictive: all results are obtained in terms of a finite (usually small) number of measurable parameters. As we shall see. quantum electrodynamics (QED) considered as the effective low energy part of the electroweak theory (Chapter 12). one obtains an effective theory whose lagrangian may consist of operators of dimension four or less (the ‘renormalizable’ part) and higher dimension operators suppressed by inverse powers of large mass M. if we want to calculate QED and QCD corrections to low energy. too. We encounter the latter case. Moreover. Suppose that physics at scale M m is described by a renormalizable theory. for instance. the physical cut-off for that theory is of similar . be interested in the O(m/M) corrections. If we neglect all physical effects of those corrections. A wellknown example is. Since E/MW 1. valid in a limited energy range. We know that such theories are often very good approximations to the real world in some energy range and therefore very interesting to study. we also know now that they should be regarded as only effective theories. At the energy scale m. It is the content of the Appelquist–Carrazone decoupling theorem (Appelquist & Carrazone 1976) that all effects depending on positive powers of M or logarithmic in M can be absorbed into the redefinition of the bare parameters of the effective renormalizable theory. Perturbative QCD is not applicable in that region but we may construct. or less. Effective field theories appear naturally in the physical situation characterized by the presence of two hierarchical mass scales M m. we may.174 4 Introduction to the theory of renormalization or infinite! Traditionally. i. However. weak processes (Chapter 12). with no need for higher order perturbation in it. first of all. for instance. after integrating out heavy degrees of freedom (see Chapters 7 and 12). it is sufficient to use it only in the Born approximation. a theory is called renormalizable only if a finite number of counterterms is sufficient to renormalize it. for example.

The latter still has a lot of predictive power (based on such principles as analyticity. Finally. φ → − .Problems 175 order of magnitude as the maximal energy of its applicability so that the E/ expansion cannot be so quickly truncated and higher order iteration of the effective lagrangian is necessary. we cannot learn about that by studying electromagnetic processes at low energy. although a renormalizable theory can formally be used for calculations at any energy scale (since the dependence on larger scales is totally absent). we may consider the physical situation when we know about (or expect) the existence of a large physical scale (for example. At the scale M Z it must be embedded into the electroweak theory. without the (unrealistic) quest for the theory of everything. The above comments on effective field theories anticipate a lot of knowledge which the reader is going to acquire in the following chapters. Problems 4. (b) Study spontaneous breaking of this discrete symmetry at the classical level. such that m M. although the number of free parameters increases with the order of perturbation theory. However. symmetries). Planck scale) but we do not know the correct theory at that scale. it must be stressed that there is no fundamental difference between a theory which is renormalizable with a finite or infinite number of counterterms (but only those consistent with the symmetry of the bare action). It is then a very natural hypothesis that physics at lower scales is described by an effective theory which contains all higher dimension operators consistent with symmetries of the low energy theory and suppressed by powers of the large scale. E M Z . It is of fundamental importance that Nature can be described by field theories that obey the decoupling theorem. it makes it difficult to learn about new physics at higher scales. with the masses m and M. Therefore. calculate the one-loop effective ˜ potential by integrating the tadpole amplitudes. unitarity. However. (a) Construct the most general renormalizable lagrangian invariant under the symmetry φ → −φ. as such effects are always strongly suppressed. unless we have reached the precision of sensitivity to O(E/M Z ) corrections. this does not mean that it is correct at any scale. (c) Assuming that the discrete symmetry is spontaneously broken calculate the vacuum expectation value v at one-loop level.1 Discuss a model with two scalar fields φ and respectively. On the other hand. This makes it possible to understand fundamental laws of physics step by step. Hint: expand φ = φ − v where . It should be emphasized that. . they may be a useful guide to further reading. A good example is again QED.

the vacuum expectation ( value v is fixed by the equation ( + = 0).4 Calculate the superficial degree of divergence in n dimensions of a Feynman diagram with I f (E f ) internal (external) lines of type f and Ni vertices of type i. 4. Show that D = f I f (2s f − 2 + n) + i Ni (di − n) + n. with the masses M and m. with di derivatives. If all i ≥ 0 then D ≤ Dmax and the theory is renormalizable with a finite number of counterterms. pay attention to the structure of quantum corrections from the exchange of the field . (b) Calculate one-loop self-energy diagrams for the field ψ and compare the corrections to its mass with the analogous corrections to the mass of the field in Problem 4. where s f = 0 (1/2) for spin 0 (1/2) particles.176 4 Introduction to the theory of renormalization φ = v is to be found from the minimum of the effective potential. where n i f is the number of lines of type f in the vertex of type i. V1 /dv = i (d) Single out from the effective potential the corrections to the mass m and to the self-coupling of the field φ which depend on the heavy field and compare them with the result of the direct diagrammatical calculation.2 Discuss a model with a scalar field φ and a fermion field ψ. Using 2I F + E f = i n i f Ni . Only the last term in the equation for D depends on the internal structure of diagrams. such that m < M. where p 4 is the ‘time’ component of the momentum four-vector in the ˆ Euclidean space. For the 4 theory we then have the generating functional W [J ] = where SE = d4 x E 1 1 (∂ )2 + m 2 2 2 2 [d ] exp −S E − d4 x J (x) (x) E λ 24 (4. calculate the one-loop contribution V1 to the effective potential by integrating the equation and identifying φ ≡ v. (e) Formulate the effective theory at the scale µ M. (a) Construct the most general renormalizable lagrangian. 4. rewrite the ˜ ˜ lagrangian in terms of the φ. the presence of the φ 3 coupling generates tree level ) and one-loop ( ) tadpole diagrams.67) + 4 (4. Instead of performing Wick’s rotation in the Feynman integrals it is more convenient to begin with the theory formulated in the Euclidean space. respectively.3 It is convenient to perform multi-loop calculations in Euclidean space. .1. one gets D = n − f E f (s f − 1 + n/2) − i Ni i with i = n −di − f n i f (s f −1−n/2). 4.68) Check that the n-point Green’s functions in the Minkowski space are obtained from those in the Euclidean space by analytic continuation of their momentum variables: ˆ pi4 → −i pi0 .

the results of 177 . The Green’s functions are not. The ’t Hooft–Veltman dimensional regularization defined for perturbative expansion (we can think. Thus. The QED lagrangian is based on the minimal coupling ansatz and in terms of the bare fields and parameters it reads µν B / L = − 1 Fµν + FB + ¯ B (i/ − qeB AB ) ∂ 4 B − mB ¯ B B + LG (5. about perturbative expansion for the bare Green’s function) belongs to this category. Nevertheless. are gauge-independent. for instance. In terms of the Green’s functions these are exactly the conditions which protect the physical equivalence of different gauges. however. As we know from Chapter 3. which is necessary for a consistent theory. so that LG = − 1 (∂µ Aµ )2 B 2aB To begin. crucial for which is the renormalization programme described in Chapter 4 which is usually formulated in terms of the Green’s functions. to derive the Feynman rules for a gauge theory it is necessary to fix the gauge by adding some gauge-fixing conditions. It is important that there exist regularization procedures which preserve the gauge symmetry of the theory. physical objects and they are gauge-dependent. regularized (finite) Green’s functions. All physical quantities. the gauge symmetry of the theory leads to certain constraints on the Green’s functions: there exist relations between them known as the Ward–Takahashi identities.1) where eB > 0 and LG is a gauge-fixing term. So far most of the physical applications of gauge theories rely on perturbation theory. We will work in the class of covariant gauges. This fact. we consider the bare. has to be checked in each case. and the S-matrix elements in particular.5 Quantum electrodynamics We begin our discussion of gauge theories with the simplest and the best-known one: QED.

in most popular renormalization schemes Z 1 = Z 2 including finite terms.178 5 Quantum electrodynamics our formal manipulations should always be understood as valid order-by-order in perturbation theory. compare with QCD.1) we can derive the Ward–Takahashi identities for the bare. Our discussion in Chapter 4 of the renormalization programme applies to the present case.3) are finite (make this argument precise by considering BRS invariance and taking into account that the ghost field is not renormalized. does not undergo renormalization. Apart from the gauge-fixing term. The relations between bare and renormalized quantities are as follows AB = Z 3 Aµ .1). using these Ward identities one can also prove a posteriori that the renormalized Green’s functions are finite). the ghost field. .3) on renormalized fields.2) is invariant under the gauge transformations (1. From the lagrangian (5. Actually.1) written in terms of the renormalized quantities reads† / ∂ L = − 1 Z 3 Fµν F µν +Z 2 ¯ [i/ −qe(Z 1 /Z 2 )A] −Z 0 m ¯ 4 −(1/2a)(∂µ Aµ )2 (5. for example. Leaving aside these formal considerations (consult. † Strictly speaking one should include the scalar ghost field and refer to the BRS invariance (see Problem 3.3). allowing all parameters and fields to become renormalized. regularized Green’s functions of our theory. The gauge-invariant structure of the QED lagrangian is not destroyed when each gauge-invariant piece of it is multiplied by a constant. eB = (Z 1 /Z 2 Z 3 )e aB = Z 3 a 1/2 m B = (Z 0 /Z 2 )m = (m − δm)/Z 2 . Slavnov & Faddeev (1980). Therefore Z 1 = Z 2 up to finite terms. there is no need for a gauge-fixing counterterm). being free.94)   (x) = exp[−iq (x)] (x) (5. Collins (1984)) we simply rewrite the lagrangian (5.91) and (1.2) (as we shall see shortly. The renormalization constants Z i must be fixed by imposing certain renormalization conditions. Using these identities one can then prove. This result will be derived formally using the Ward– Takahashi identities. µ 1/2 B = Z2 1/2 . that the theory is renormalizable without destroying its gauge invariance. Section 8. Thus the most general gauge-invariant form for the lagrangian (5. In addition the parameter e can be arbitrarily changed.1) in terms of the renormalized quantities assuming the most general form consistent with gauge invariance after renormalization and we derive Ward identities for the renormalized Green’s functions (actually. Bogoliubov & Shirkov (1959). lagrangian (5. In addition it is clear that Z 1 /Z 2 must be finite as all other quantities in (5.3) 1 Aµ (x) = Aµ (x) + (Z 2 /Z 1 )∂µ (x)  e If a free scalar ghost field is included the full lagrangian is invariant under BRS symmetry transformation (see Problem 3.

for instance.4) and the lagrangian L(Aµ . The renormalization conditions are then imposed so as to ensure this interpretation of the renormalized quantities in any order of perturbation theory. In addition. However. the so-called on-shell renormalization scheme is the one most commonly used in QED calculations. α] = ¯ where Seff = d4 x [L(Aµ .2). In this scheme we choose to interpret the coupling constant e. Generally speaking.5) D(Aµ ¯ ) exp(iSeff ) (5. this can be achieved by demanding that in the on-shell limit. ) is given by (5. 5. Under the functional integral one can perform an infinitesimal gauge transformation  = [1 − iq ]  (5. in each order of perturbation theory there exist definite relations between sets of parameters corresponding to different renormalization schemes. the mass parameter m and the renormalized fields which appear in the lagrangian (5. an infinite number of other renormalization prescriptions is possible.1 Ward–Takahashi identities 179 Mainly for historical reasons. The generating functional for the full Green’s functions is the following W [J.86)).6) 1 A µ = A µ + ∂µ  e ¯ where e = eZ 1 /Z 2 .7) . when all the four-momenta approach their on-shell values. The renormalized quantities lose their previous interpretation but the theory can be formulated in terms of them in a totally equivalent way. in any order of perturbation expansion.4) cannot change the value of the integral.1 Ward–Takahashi identities General derivation by the functional technique We derive the Ward–Takahashi identities in QED using the functional technique. ) + Jµ Aµ + α ¯ + ¯ α] (5.2) as the electric charge measured.5. the fermion and the photon propagator as well as the fermion–fermion–photon vertex function approach their zeroth order form. α. Therefore we conclude that δW/δ (y)|θ =0 = 0 (5. respectively. a change of the integration variables in (5. The integration measure is invariant under infinitesimal gauge ¯ transformation and the only gauge-dependent terms in the integrand are source terms and the gauge-fixing term. the physical mass of the fermion and the ‘physical’ fields in the sense of Chapter 4. On the other hand. in the Thomson scattering (see (12.

= Aµ . ] = i ¯ × d4 x D(Aµ ¯ ) exp(iSeff ) 1 Jµ (x)∂ µ (x) − iq (x)α(x) (x) + iq (x) ¯ (x)α(x) ¯ e ¯ (5. α α] − ¯  δZ  δZ δZ ¯ =−  . in a class of covariant gauges.8) 1 (x) − (1/a)∂µ Aµ (x)∂ 2 (x) e ¯ Integrating expression (5. α] for the connected Green’s functions (W = exp(iZ )) ¯ δZ 1 2 δZ δZ ∂ ∂µ + ∂µ J µ (y) + iq eα(y) + iq e ¯ α(y) = 0 ¯¯ a δ Jµ (y) δ α(y) ¯ δα(y) (5. The generating functional for the latter has been defined in Section 2. .9) (remember that ( δ /δα)W = −W ( δ /δα) or in terms of the generating functional Z [J.13) where (left derivatives only) . α = . In the present case its arguments are the ‘classical’ boson field Aµ (x) and the ‘classical’ fermion fields (x) and ¯ (x) (it is convenient to use the same notation for the arguments of as for the fields in the lagrangian. α.10) is the general form of the Ward–Takahashi identities in QED. (x). neglecting the surface terms at infinity and remembering that functional integrals of derivatives of fields are derivatives of the Green’s functions (see Section 2. ¯ (x)] d4 x [Jµ (x)Aµ (x) + α(x) (x) + ¯ (x)α(x)] ¯ (5.12)  δ δ δ   Jµ = − µ .11) = Z [J. see Section 2. ¯ ] the identity (5. = δ Jµ δα ¯ δα  (5. α. (5.6.180 5 Quantum electrodynamics where δW [J.10) can be rewritten as follows: 1 2 (y) µ δ δ δ (y) ∂(y) ∂µ A (y) − ∂µ + iq e ¯ (y) + iq e ¯ (y) ¯ =0 a δ Aµ (y) δ (y) δ ¯ (y) (5.8) by parts.3) we get δW 1 2 1 δW δW ∂ ∂µ + ∂µ J µ (y)W + q eα(y) + qe ¯ α(y) = 0 ¯¯ a i δ Jµ (y) δ α(y) ¯ δα(y) → ← (5. α.6) so that [Aµ (x).10) Eq. α= ¯ δA δ δ¯ In terms of the functional [Aµ . It is also very useful to have the analogous relation for the 1PI Green’s functions. for the connected Green’s functions.

15) reads ˜ ˜ µν G µν (k) = −ia(kµ kν /k 4 ) + G T (k) where the transverse projection ˜ µν G T (k) = (gµν − kµ kν /k 2 ) f (k 2 ) (5.16). We observe that the longitudinal (gauge-dependent) part is not altered by interactions (compare with (3. The Ward identity (5.10) with respect to Jν (z) and putting J = α = α = 0 one gets ¯ 1 2 µ δ2 Z − ∂(y) ∂(y) µ a δ J (y)δ Jν (z) J =0 ν = ∂(y) δ(y − z) (5.17) (5.1 Ward–Takahashi identities 181 In the following we shall illustrate the general relations (5. Differentiating the general identity (5. This means.15) ˜ Here G µν (k) is the Fourier transform of the connected part of the photon propagator G µν (x1 − x2 ).5.13) with several examples. one gets ¯ δ 1 2 (y) δ2 ∂(y) ∂µ Z a δ α(x)δα(z) δ Jµ (Y ) ¯ δ2 δ2 Z = iq e ¯ δ(y − x) + iq e ¯ δ(y − x) δα(z)δ α(y) ¯ δ α(x)δα(y) ¯ Since 0|T (x) ¯ (y)|0 = i δ2 Z δ α(x)δα(y) ¯ J =α=α=0 ¯ (5.10) with respect to δ δ δ δ =− δ α(x) δα(z) ¯ δα(z) δ α(z) ¯ and setting J = α = α = 0. Our next example is the well-known relation between the vertex and the fermion propagators.10) and (5.15) constrains the full renormalized propagator to the form (5.134) and Fourier transforming into momentum space we obtain the following relation: ˜ i(1/a)k 2 k µ G µν (k) = kν (5. among other things.32)). Differentiating relation (5. The solution to (5.14) Using (2.18) .16) gives zero contribution to the relation (5. that the renormalization programme does not require any counterterms for the gauge-fixing term. Examples As the first example we consider the Ward identity for the longitudinal part of the connected photon propagator.15).

18) can be rewritten as follows µ 2 −(1/a)∂(y) ∂(y) 0|T Aµ (y) (x) ¯ (z)|0 = q e 0|T (y) ¯ (z)|0 δ(y − z) − q e 0|T (x) ¯ (y)|0 δ(y − z) ¯ ¯ (5.21) in momentum space by Fig.2.19) This is the Ward identity in configuration space. 5. Notice that the form (5. q) (5.1 and relation (5. 5.182 5 Quantum electrodynamics Fig. .1.21) in momentum space can be immediately deduced from the graphical representation of relation (5. q) = q eS( p + q) − q eS( p) ¯ ¯ (5. 5. Using translational invariance and Fourier transforming into momentum space  0|T Aµ (y) (x) ¯ (z)|0 = 0|T Aµ (y − z) (x − z) ¯ (0)|0      4 4  d p dq   exp[−i p(x − z)]  =  4 (2π )4  (2π )     × exp[−iq(y − z)]Vµ ( p. relation (5.20)   d4 k   exp[−ik(y − z)]iS(k)  0|T (y − z) ¯ (0)|0 =   (2π )4     4  dk  ¯ (0)|0 =  exp[−ik(x − y)]iS(k)  0|T (x − y) 4 (2π ) then inserting (5.20) into (5. we obtain the Ward identity in momentum space −(1/a)q 2 q µ Vµ ( p.21) Graphically.19) in configuration space.19) can be represented by Fig. respectively.19) and integrating over dx dy dz exp(i p x) × exp(iq y) exp(ik z) to get rid of the δ-functions. and 0|T Aµ (y) (x) ¯ (z)|0 = δ3 Z δ Jµ (y)δ α(x)δα(z) ¯ J =α=α=0 ¯ relation (5. where the cross denotes −(1/a)∂ 2 ∂ µ or −(1/a)q 2 q µ .

for the free propagator. z)δ(y − x) = 0 (5. In practice.2. Indeed all other quantities. (n) The coefficients f i contain pieces which are divergent in the limit where regularization is absent (for example. i. In addition.13) with respect to δ δ δ ¯ (x) δ (z) set Aµ = ¯ = = 0 and use the definitions (2. z) − iq e ¯ (2) (x. In perturbative calculations.23) where the ˜ (n) s are Fourier transforms of the (n) s. for example. e2 /4π = 1/137.22) reads qµ ˜ (3) ( p.135) generalized to include fermions to get the following: µ ∂(x) (3) µ (y.4. using a regularization procedure which preserves gauge invariance.21) and (5. This point will be discussed in more detail in Section 6. x. when ε → 0 in dimensional regularization) and possibly also finite pieces. the Z i s are (n) n ∞ power series in the coupling constant α = e2 /4π: (Z i − 1) = n=1 f i α .1 Ward–Takahashi identities 183 Fig. y)δ(y − z) + iq e ¯ (2) (y. We recall (2. q) = −iq e ˜ (2) ( p + q) + iq e ˜ (2) ( p) ¯ ¯ (5.e. k As an interesting consequence of the Ward identities (5. 5.5.22) By analogy with (5. in momentum space relation (5. The Ward identity (5. when Z 1 = Z 2 the counterterms form a gauge-invariant set and the Ward identities for . including e. This is convenient (but not necessary) because the parameter e is then universal for all fermions in the theory.19). We take the derivative of (5. It is also worth remembering that only in the on-shell renormalization scheme is e the electric charge. in all the commonly used renormalization schemes Z 1 = Z 2 including the finite terms. Thus the finiteness of the ratio Z 1 /Z 2 implies that order-by-order in perturbation theory Z 1 = Z 2 up to the finite terms.152): ˜ (2) = ˜ i[G (2) ( p)]−1 = +S −1 (k) where.23) we notice that the gauge parameter e = eZ 1 /Z 2 and consequently the ratio Z 1 /Z 2 must ¯ be finite if the theory is renormalizable. S(k) = 1/(/ − m).21) or (5.21) can also be written in terms of the 1PI Green’s functions. in these equations are renormalized quantities.

for example. . in the relation γµ γν + γν γµ = 2gµν 1 l the only space-time-dependent object is the tensor gµν . and define g µ µ = g µν gµν = n This convention is the one most frequently used in calculations but other conventions are also possible (Collins 1984). of course. in supersymmetric theories a dimensional regularization method called dimensional reduction (Siegel 1979) is used which preserves supersymmetry.24) (there is no counterterm to the gauge-fixing term: see Section 5.2 Lowest order QED radiative corrections by the dimensional regularization technique General introduction We recall the QED lagrangian (q = −1. We notice that.e. the electron scattering (or e+ e− pair production) in a static Coulomb potential. The latter can. Dimensional regularization in QED in the presence of γ -matrices requires an extension of previous rules. In n dimensions we can assume Tr[γµ γν ] = 1 2 Tr[γµ γν + γν γµ ] = gµν Tr 1 = 4gµν l (5. e > 0) in a covariant gauge / L = − 1 Fµν F µν + ¯ (i/ + e A) − m ¯ − (1/2a)(∂µ Aµ )2 − (Z 3 − 1) 1 Fµν F µν ∂ 4 4 ¯ i/ − (Z 0 − 1) ¯ m + (Z 1 − 1)e ¯ A / + (Z 2 − 1) ∂ (5. Several other relations and integrals useful in the subsequent calculations are collected in Appendix B.184 5 Quantum electrodynamics the renormalized Green’s functions are identical to those for the bare regularized Green’s functions. The only difference is that one must start with the lagrangian expressed in terms of the bare quantities. An important exercise is to calculate the Thomson limit of the Compton scattering (Problem 5.25) i.1).2) in the on-shell and minimal subtraction renormalization schemes. for example. 5. The Feynman rules are given in Appendix B. We shall calculate now all the elements which are necessary to discuss the lowest order radiative corrections to QED processes such as. we still consider γµ s as 4 × 4 matrices in the fermion and antifermion spin space. In particular. be derived in a similar way to the renormalized Ward identities. This is not the case for the standard dimensional regularization convention used here.

27) where from the Feynman rules we obtain i µν (q) = −(ie)2 dn p i i Tr[γµ ( p + m) / n p 2 − m 2 ( p − q)2 − m 2 (2π) (5.28) × γν ( p − / + m)] + i(Z 3 − 1)(1) (qµ qν − q 2 gµν ) / q Therefore. From consideration of Section 5. 5.26) In the one-loop approximation the transverse part of the photon propagator reads −iPµν (−iPµρ ) ˜ µν G T (q) = + i q2 q2 ˜ µν G T (q) = The scalar function (q 2 ) is given by g µν µν (q) ρσ (q) (−iPσ ν ) −iPµν = [1 + q2 q2 (q 2 )] Summing over all one-particle reducible diagrams one gets q 2 [1 −iPµν − (q 2 )] = (n − 1)q 2 (q 2 ) (5.2 Lowest order QED radiative corrections 185 Fig. On general grounds one can write it as follows: i µν (q) = iP µν = i g µν (q 2 ) q µq ν − 2 q 2 (q 2 ) q T (5.29) + counterterm .1 we know that only the transverse part of the photon propagator has higher order corrections.5. This is a oneloop correction to the free photon propagator. after taking the trace Tr[γµ γα γν γβ ] = 4(gµα gνβ + gµβ gνα − gµν gαβ ) and using the Feynman parametrization 1/ab = gets (n − 1)q 2 (q 2 ) = ie2 dn p (2π)n 1 1 0 dx (1/[xa + (1 − x)b]2 ) one dx 0 4(2 − n)( p 2 − p · q) + 4nm 2 {x( p 2 − m 2 ) + (1 − x)[( p − q)2 − m 2 ]}2 (5.3.3. Vacuum polarization We calculate the amplitude corresponding to the diagrams of Fig. 5.

In the minimal subtraction scheme (Z 3 − 1)(1) = − α 21 π 3ε (5.30) (Z 3 − 1)(1) = − and ON α 21 1 m2 + 3 (ln 4π − γ ) − 1 ln 2 3 π 3ε µ m 2 − x(1 − x)q 2 − iε m2 (5.31) and the finite part of the vacuum polarization. It is useful to notice that in the MS.30) (we have taken n = 4 − ε.30). and α is dimensionless in 4 − ε dimensions: α → αµε .32)  |q 2 | α   ln 2 |q 2 | m2  3π µ The renormalization scheme most frequently used in QED is the on-shell renormalization.33) (q 2 ) = α 2 π 1 0 dx x(1 − x) ln (5.18)–(B.20) we get finally 1 α m 2 − x(1 − x)q 2 − iε −ε/2 (q 2 ) = − (4π)ε/2 1 ε dx 2x(1 − x) 2 π µ2 0 1 α 12 1 m 2 − x(1 − x)q 2 − iε + (ln 4π − γ ) − =− dx 2x(1 − x) ln π ε3 3 µ2 0 + O(ε) − (Z 3 − 1)(1) (5. Hence α ON = α(m 2 ) (see also Chapters 6 and 12).186 5 Quantum electrodynamics Using the integrals (B.34) † The so-called MS scheme in which the counterterms include the constant terms: 2/ε → η = 2/ε − γ + ln 4π is also often used. In the large q 2 limit. (q 2 = 0) = 0 for µ2 = m 2 . α = e2 /4π . from (5. in the MS scheme the sum over all one-loop one particle reducible diagrams gives ˜ µν G T (q) = −iPµν 2 2 1 − α ln |q | q 3π µ2 This is the so-called leading logarithmic approximation for the photon propagator. . where µ is an arbitrary mass parameter). given by (5. It is cancelled by adding the Z 3 counterterm. In this case one demands: ON (q 2 ) = 0 for q 2 = 0 Therefore. can easily be evaluated in the small q 2 and large q 2 limits and reads (up to constant terms)†  m2 q4 α α q2 1  2   q →0 ln +O −  3π µ2 π m 2 15 m4 (5. The pole in ε reflects the UV divergence in four dimensions.

The proper two-point function (2) ( p ) reads / (2) ( p) = p − m − / / ( p .4. Taking into account the symmetry y → −y of the integrand we can write ON (q 2 ) = α1 π2 α1 π2 1 dy 0 ∂ q 2 (1 − y 2 ) (y − 1 y 3 ) ln 1 − 3 ∂y 4m 2 − iε (5.38) which is a so-called once subtracted dispersion relation. 5.2 Lowest order QED radiative corrections 187 Fig.39) The propagator is the inverse of the proper two-point function so we get ˆ G (2) ( p ) = / i p−m− / (5.36) and integrate it by parts to get ON (q 2 ) = − 1 0 dy 2y(y − 1 y 3 ) 3 q2 4m 2 − q 2 (1 − y 2 ) − iε (5. m) / (5. Firstly.4.5. Asymptotically one obtains ON q4 α q2 1 (q ) −→ − +O π m 2 15 m4 q 2 →0 2          (5.37) Another change of variable t = 4m 2 /(1 − y 2 ) gives the result ON α (q 2 ) = − q 2 π ∞ 4m 2 1 1 2m 2 dt 1+ t t − q 2 − iε 3 t 1− 4m 2 t 1/2 (5. Electron self-energy correction We consider the electron self-energy truncated diagram and the necessary counterterms which are shown in Fig. 5.35) ON α (q ) −→ 2 /m 2 →∞ π −q 2 1 3 m2 |q 2 | 5 ln 2 − + O m 9 q2 As a final comment we rewrite ON (q 2 ) in the form of a dispersion integral (the dispersion technique will be discussed in some detail later).34) into y = 1 − 2x.40) . we change variables in the integral (5.

44) m( p 2 − im(Z 0 − 1)(1) + i(Z 2 − 1)(1) p / As before we use dimensional regularization: n = 4 − ε.45) N = γ µ ( p + / + m)γ ν gµν / k we find (after some algebra) Tr N = 4nm Tr[ p N ] = 4[ p 2 (2 − n) + ( p · k)(2 − n)] / Next. the / electron self-energy is gauge-dependent).46) dx 0 1 [k 2 + x p · 2k + x( p 2 − m 2 )]2 (5.47) .43) In this section we shall calculate ( p ) in the Feynman gauge (unlike µν (q).188 5 Quantum electrodynamics The general structure of the electron self-energy correction is obviously the following ( p . −i ( p ) = (ie)2 / dn k µ −igµν i γ 2 γν n (2π ) k + iε p + / − m + iε / k (5. The functions and V ( p 2 ) are given by Tr ( p ) = 4 / m( p 2 ) ) V (p 2 Tr[ p ( p )] = 4 p 2 / / Introducing ) (5. m) + ( p − m F ) / / 1( p = m F .1) appears as a correction to an external line of the on-shell scattering amplitude. we use Feynman parametrization for the denominators 1 1 = k 2 ( p + k)2 m 2 1 (5. m) = / / 0( p = m F . m) + ( p − m F )2 / m2 F / 2 ( p .41) It is often convenient to write it as an expansion ( p .42) the inverse propagator = {[m + m (m F )]/[1 − considered. m) = / V (p 2 )p + / m( p 2 )I (5. m) (5. We recall that 1 (actually 1 1 : see 2 Section 4. The Feynman integral reads. The calculation proceeds as follows. We can then find the one-loop relation between the pole mass m F and the renormalized mass parameter m as well as the correction 1 . It can be calculated as ( p 2 = p 2 ) / where m 2 is the zero of F 2 V (m F )]} to the order 1 = ∂ ∂p / = p=m F ∂ m ∂ p2 p 2 =m F 2m F + 2 2 V (m F ) + 2m F ∂ V (p ∂ p2 2 ) p 2 =m 2 F (5.

48) 1  (1)   dx (1 − x)X − (Z 2 − 1)  V = (2 − n)C 0 where α −1 C≡ 4π 4π −ε/2 1 ε 2 −ε/2 x( p 2 − m 2 ) − x 2 p 2 X= µ2 Using standard tricks to single out the pole terms in ε we immediately find the renormalization constants in the minimal subtraction scheme  2α 1  (1)  (Z 0 − 1) =  π ε (5. Indeed. their derivatives contributing to 1 are not. of course.50) and similarly for V ( p 2 ). We leave it to the reader to check that in the MS scheme at the one-loop level α(µ) m F = m(µ) 1 + π We observe that m and V are IR finite (when p 2 → m 2 ). However.49) α 1   (1)  (Z 0 − 1) = − 2π ε The final (finite) expressions for m and V can then be easily written down. m( p 2 α )=− m π dx ln xm 2 − x(1 − x) p 2 +γ + 4π µ2 1 2 (5.48) with respect to p 2 one obtains the following result ∂ m = nmC − 1 ε 2 ∂ p2 1 dx 0 x(1 − x) 1+2/ε X µ2 (5. UV finite (Z i are constants) and in the limit p 2 → m 2 gives (we can take m 2 = m 2 because we work in the first order in α) F ∂ m ∂ p2 =n ε 1 α 1−γ m 4π 2 m2 4π µ2 −ε/2 p 2 =m 2 (−ε) (2) (2 − ε) (5.2 Lowest order QED radiative corrections 189 and (B. for example.52) .51) which is.20) to perform the integration over momentum k to get the following results:  1  (1)   dx X − (Z 0 − 1) m m = nmC  0 (5. differentiating the regularized expression (5.5.18)–(B.

εUV > 0) and εIR (n = 4 + εIR . Electron self-energy: IR singularities regularized by photon mass The IR singularities encountered above are due to the zero mass of the photon. For the electron self-energy correction we now have −i ( p ) = −e2 / dn k 1 1 γn ( p + / + m)γ µ 2 / k (2π)n k − λ2 ( p + k)2 − m 2 . we get = ∂ ∂p / = p=m 1 α 1 m2 + 3 ln − 1 + 3γ 4 π εIR 4 4π µ2 (5.56) Our final remark is that although V ( p 2 ) does not have IR singularities (when p 2 → m 2 ).54) V (p 2 = m2) = m2 α +γ −2 ln 4π 4π µ2 (5.53) p 2 =m 2 2 V /∂ p ∂ V ∂ p2 In addition p 2 =m 2 =− 1 α 1 m2 − 1 + 1 γ + 1 ln 2 2 2m 2 π εIR 4π µ2 (5. where eventually λ → 0 (actually this is the most standard method). The final result reads (εIR > 0) ∂ m ∂ p2 and for ∂ = 1 1 α m2 + 1 γ − 1 + 1 ln 4 2 2 m 4π εIR 4π µ2 +1 (5.190 5 Quantum electrodynamics The integration over Feynman parameters has introduced a new (IR) singularity which has again been regularized by keeping n = 4 − ε. Instead of using the dimensional method one can regularize the IR singularities by giving the photon a small mass λ. but this time with ε < 0 (changing ε from a negative to a positive value is legitimate since we are already dealing with a UV finite quantity).55) so for 1. In order to keep trace of the origin of the singular terms here and in the following we shall often distinguish between εUV (n = 4 − εUV . εIR > 0). it nevertheless has the so-called mass singularities (when p 2 → m 2 → 0).

π 2m 2 λ (5.5 µ ( p.61) .58) p 2 =m 2 m2 1 α − 2 ln 2 + const.  2 4πµ (5.5.2 Lowest order QED radiative corrections 191 Fig. We get in particular ∂ m ∂ p2 ∂ V ∂ p2 and finally 1 p 2 =m 2 =m = α 1 m2 ln 2 + const. 5. p ) = (ie)2 1 dn k i i γµ γβ (−ig αβ ) 2 (Z 1 − 1)(1) γµ γα (2π)n p + / − m p + / − m / k / k k (5. π 4m λ (5. instead of (5. π2 λ π 4 4π µ2 (5. The derivatives of m and V can now be calculated in an elementary way. where x1 and x2 are the roots of the quadratic expressions. On-shell vertex correction We consider Fig.60) Comparing with the dimensional regularization of IR singularities we observe the correspondence 2/εIR → ln(m 2 /λ2 ).5.57) These integrations can be done by writing the logarithm as ln p 2 (x − x1 )(x − x2 ). dx ln 4πµ2 dx (1 − x) ln      α 2π  x 2 p 2 − x( p 2 − m 2 + λ2 ) + λ2 − iε   + const.50) one gets: α m ( p ) = −m π 2 2 V (p ) = x 2 p 2 − x( p 2 − m 2 + λ2 ) + λ2 − iε + const. Following the previous calculation. 5.59) = m2 α 1 m2 α1 ln 2 + ln + const.

66) IR µ = γµ α α 1 (2q 2 − 4m 2 ) γµ ( p + p )2 I1 + 1 γ I1 + 1 I2 + 2 2 4π εIR 2π α − (5.64) Only the integral with two powers of k in the numerator is UV divergent (but IR finite). so we take n = 4 + εIR there.192 5 Quantum electrodynamics We work in the Feynman gauge and assume that the external electron lines are on-shell p 2 = p 2 = m 2 . We get finally 2 µ = UV µ + IR µ where (in n = 4 − εUV dimensions) −q 2 x(1 − x) + m 2 α 1 1 − (2 + γ ) − 1 dx ln 4 π 2εUV 4 4πµ2 α m + ( p + p )µ I1 + γµ (Z 1 − 1)(1) 2π 2 and (in n = 4 + εIR dimensions) UV µ = γµ (5.63) We now introduce Feynman parametrization for the propagators and then have to perform the following integrations 1 2 0 dx dy y dn k [1. To simplify notation we write the final results in terms of the integrals  1 1 dx dx x    I1 = I3 = −q 2 x(1 − x) + m 2 −q 2 x(1 − x) + m 2  0 0 (5. p ) − (Z 1 − 1) γµ = −ie dn k N n [( p + k)2 − m 2 ][( p + k)2 − m 2 ]k 2 (2π ) (5. kα kβ ] n {k 2 + [ p x y + p(1 − x)y]2k}3 (2π ) (5. Using the identity γ µ / γµ = (2 − n)/ . we can write the a a integral as (1) 2 µ ( p. The other integrals are at most only IR divergent.62) where N = γµ [4 p · p + (n − 2)k 2 + 4( p + p ) · k] k k − 4( p + p )µ/ + kµ [4m − 2(n − 2)/ ] (5.65) 1  −q 2 x(1 − x) + m 2 dx    I2 = ln 2 2 4π µ2 0 −q x(1 − x) + m where q = p − p (we assume q 2 ≤ 0) and µ2 is an arbitrary mass scale.66a) ( p + p )µ m I1 2π . kα . We see that I3 = 1 I1 . We regularize it by taking n = 4 − εUV .

γν ] 2 (5.2 Lowest order QED radiative corrections 193 The above result is often rewritten in terms of the Dirac and Pauli formfactors µ ( p. 2 σµν = 1 i[γµ .72) in this case (the renormalization constants now become IR divergent). In our calculation we have reproduced the famous Schwinger result (Schwinger 1949) for the anomalous magnetic moment of the electron F2 (q 2 = 0) = α/2π (5.2). p=m F1 (q 2 = 0) = 1 the relation Z 1 = Z 2 also holds. One can check that for the on-shell renormalization given by the conditions ∂ ∂p / = 0. p ) = γµ F1 (q 2 ) + 1 (i/m)σµν q ν F2 (q 2 ).68) dx ln −q 2 x(1 − x) + m 2 + m 3 I3 4π µ2 1 + 1 γ I1 + 1 I2 2 εIR 2 (5.67) Using the so-called Gordon decomposition ¯ ¯ u( p )γµ u( p) − 1 (i/m)u( p )σ µν qν u( p) = 1 (1/m)u( p )( p + p )µ u( p) ¯ 2 2 and the relation ( p + p )2 = 4m 2 − q 2 .69) it follows that Z1 = Z2 = 1 − 1 α m2 1 + 3 γ + 1 − 3 ln − 4 4 2 π 2εUV 4π µ εIR (5. Indeed from (5. comparing with (5.73) for the anomalous magnetic moment is finite before renormalization. In the minimal subtraction scheme α 1 (5.5.69) − m2 α 2 m I1 (5. At this point we encourage the reader to calculate the Thomson cross-section in the on-shell renormalization scheme (Problem 5. It depends on the renormalization scheme through the value of α(µ2 ) but that dependence is of the order O(α 2 ). one gets the following: F1 (q 2 ) = 1 1 α 1 − (2 + γ ) − π 2εUV 4 4 + (Z 1 − 1)(1) + + (m 2 − 1 q 2 )I1 2 F2 (q 2 ) α π 1 2 q 2 (5.49). Correction (5. we have Z 1 = Z 2 to this order.71) (Z 1 − 1)(1) = − 2π εUV and.73) is .70) 2π We can now make several observations.73) Notice that the radiative correction (5.56) and (5.

The IR singularities can also be regularized by λ. Here the first two terms are the dimensional regularization form of the famous double logarithms due to the simultaneous presence of IR and mass singularities. The logarithms can also be when |q 2 | ξ 1 obtained simply by writing 0 = 0 + ξ1 and approximating the integrands in both integrals. 5. for instance. . As the next point we discuss the structure of the leading logarithmic terms in our result for the vertex corrections in the limit |−q 2 | m 2 . We have 1 dn k 1 −i ( p ) = −e2 / N 2 (5. A straightforward calculation of the leading double logarithms gives a result which corresponds to the replacement 2/εIR → ln(m 2 /λ2 ). Consider first the self-energy correction.194 5 Quantum electrodynamics finite because the operator ¯ σµν F µν has dimension five and cannot be present in the original renormalizable lagrangian. It is generated by radiative corrections as a non-local term which becomes local in the low energy limit q 2 = 0 (see also Chapter 12). In particular I2 can be written in terms of the Spence functions. in the on-shell renormalization.75) n (2π ) k ( p + k)2 where N = (2 − n)( p + / ) / k † It is easy to see that I1 ≈ I2 ≈ −1 m2 2 ln 2 −q −q 2 −1 −q 2 m2 1 2 ln ln + 2 2 −q 4πµ −q 2 −q 2 ln2 1 −q 2 + 1 π3 O 4 3 m2 q m 2 . We recalculate now in the Feynman gauge the mass and vertex corrections in this approximation. As a final remark we notice that the double logarithm terms do not depend on the renormalization scheme (Z i . the photon mass.3 Massless QED For some high energy reactions it is useful to have results in the m = 0 approximation. In the minimal subtraction scheme we get α −q 2 − ln 4π 4π µ2 as a finite contribution to F1 from α 1 m2 1 2 m2 ln − ln + π εIR −q 2 4 −q 2 1 γ 2 UV and† m2 1 −q 2 m2 + 2 ln ln −q 2 4π µ2 −q 2 (5. contains only single logarithms).74) −1 ln from IR .

after cancelling the UV divergence by Z 2 . Note that for massless QED the finite wavefunction renormalization 1 is given solely by V ( p 2 = 0) and not its derivative. Then the µ2 part cancels the 0 −(Z 2 − 1) written in terms of εIR .20) one gets (n = 4 − ε and α → αµε ) ( p) = / α1 1 p (2 − n) / π4 4π (1) / − (Z 2 − 1) p ≡ −ε/2 − p2 µ2 2 / V (p )p −ε/2 1 ε 2 1 0 x −ε/2 (1 − x)1−ε/2 dx (5. when ε → 0 (to regularize the first integral we work in n = 4 + εIR dimensions. One is to keep external lines off-shell p 2 < 0 to regularize the IR singularity. We have µ = −ie2 dn k N n ( p + k)2 ( p + k)2 k 2 (2π ) (5.3 Massless QED 195 (remember γ µ / γµ = (2 − n)/ ). Then. so we take ε → −εIR ).77) The other way is to take external lines on-shell p 2 = 0 and to use dimensional regularization of the IR singularity. ln π 4 4π µ2 (5.5. Finally.78) dx x −ε/2 (1 − x)1−ε/2 − (Z 2 − 1)(1) p / µ2 ∞ For p 2 = 0 we split the integral into two: 0 + µ2 which are IR and UV divergent.76) We can now proceed in either of two ways. respectively.79) where the pole in εIR reflects the IR singularity. we get† in the minimal subtraction scheme V (p 2 = 0) = − α 2 4π εIR (5.76) in the following form ( p) = / α (2 − n) p / 4π × 0 1 ∞ − p2 q2 dq 2 q 2 4π µ2 −ε/2 1 + 1ε 2 (5. We rewrite (5. µ2 part and the final result is simply .80) where N = γµ [4 p · p + (n − 2)k 2 + 4( p + p ) · k] − ( p + p )µ 4/ − 2(n − 2)kµ/ k k ∞ † This result is immediate if we note that µ2 −(Z 2 − 1) is already UV finite and we can write it in (4 + εIR ) ∞ dimensions by changing ε → −εIR .18)–(B. Using Feynman parametrization and the integrals a a (B. It is also straightforward to repeat the calculation of the vertex correction in the massless electron case. in the minimal subtraction scheme ( p) = p / / − p2 α1 + const.

We note also that. 2 6 . † (1 − ε) ∼ (1) − = (1)ε + 1 2 (1)ε2 = 1 + γ ε + 1 (γ 2 + 1 π 2 )ε2 . Our purpose is mainly a technical one: to show yet another method of calculating Feynman diagrams which in some cases happens to be instructive. As we know. i. in (4 ∓ ε) dimensions It is instructive to repeat the calculation of the electron self-energy and vertex corrections using dispersion relations.66). once (Z 1 − 1)(1) is explicitly determined. finite) in n = 4 + εIR dimensions µ by changing εUV → −εIR . So we can always write for them the dispersion relation without subtraction. has poles in ε and we have to perform next the standard renormalization procedure.81) and† (n = 4 + εIR ) IR µ = γµ − × = γµ α 2π −q 2 4πµ2 εIR /2 1 − 1 εIR 2 dx [x(1 − x)]εIR /2−1 1 + 1 εIR 2 (2 + εIR ) 2 εIR 1+ π2 6 2 1 ε 2 IR dy y εIR −1 (1 − y) α − 2π −q 2 4πµ2 εIR /2 (5.e. Landshoff. Eden. we now get (n = 4 − εUV ) UV µ α = γµ 2π × = γµ 1+ 1 ε 2 UV (2 − εUV ) 1 εUV −1 −q 2 4π µ2 −εUV /2 dx dy y 1−εUV [x(1 − x)]−εUV /2 + γµ (Z 1 − 1)(1) α −q 2 2π 4π µ2 −εUV /2 1 − 1 εUV 1 2 + γµ (Z 1 − 1)(1) (1 − εUV ) εUV (5.196 5 Quantum electrodynamics Repeating the steps which led us to the result (5. 5.82) The double pole structure in IR is due to the simultaneous occurrence of IR µ (massless photons) and mass (massless electrons) singularities. Then the full vertex is written in (4 + εIR ) dimensions.4 Dispersion calculation of O(α) virtual corrections in massless QED. Olive & Polkinghorne 1966). in the minimal subtraction scheme. the calculated amplitude. of course. to add counterterms. we can rewrite UV (which is. in 4 − ε dimensions Feynman amplitudes are UV convergent. The dispersive technique has been extensively used in particle physics. Both have been regularized by working in n = 4 + εIR dimensions. However. We do not attempt its systematic discussion here. referring the interested reader to other books (Bjorken & Drell 1965. of course. for example.

Self-energy calculation The dispersion relation for the function V (q 2 V (p 2 ) defined by (5.85) is the two-body phase space element in n dimensions.86) We now introduce polar coordinates for the Euclidean vector r in n − 1 dimensions dn−1 r = |r|n−2 d|r| d n−1 = |r|n−2 d|r| dθ (sin θ)n−3 d n−2 (5.83) The imaginary part of a Feynman diagram amplitude is obtained by the replacement p2 1 → −2πiδ( p 2 − m 2 ) ( p0 ) ≡ −2πiδ+ ( p 2 − m 2 ) − m 2 + iε in the Feynman integral (Landau–Cutkosky rule). In the (k.84) dn k dn l 2πδ+ (k 2 ) 2π δ+ (l 2 )(2π)n δ (n) (k + l − p) (2π)n (2π )n dn k = 2πδ+ (k 2 )2π δ+ (( p − k)2 ) (2π)n (5. Dispersion calculation of O(α) virtual corrections 197 Fig. Therefore for the diagram shown in Fig. 5.6 we can write Im where d 2 = V (p 2 ) = − 1 e2 (2 − n) ( p 2 ) 2 d 2 (1 − p · k/ p 2 ) (5.87) . l) centre of mass system it can be written as follows in terms of momenta p = k +l and r = 1 (k −l): 2 d 2 1 dn r δ( 1 p 2 + pr + r 2 )δ( 1 p 2 − pr + r 2 ) 4 4 (2π)n−2 1 dn r 1 δ( 1 p 2 + r 2 )δ( pr ) = 2 4 n−2 (2π) 1 1 = dn−1 r δ( 1 p 2 − r2 ) n−2 (2π) 2 p0 4 = (5.41) reads V (p 2 = 0) = 1 π d p2 Im p2 ) (5.

91) which is equivalent to (5.7 we have Im (q 2 ) = −e2 d 2 1+ 4k · ( p − p ) 1 2q 2 k2 (5. 5. we define dimensionless α by 2 α → αµε ) V (q 2 = 0) = − α (2 − n) 1 4π (1 − 1 ε) 2 2 ∞ 0 d p2 p2 p 2 4π µ2 −ε/2 0 1 dy [y(1 − y)]−ε/2 (5. We shall limit ourselves to the UV finite part IR to understand better the origin of the IR and mass µ singularities. Vertex calculation Calculation of the vertex correction proceeds similarly.92) and using the kinematical variables defined in Fig.93) . Finally one can write dθ (sin θ)n−3 = dz (1 − z 2 )−ε/2 = dy 2[4y(1 − y)]−ε/2 where z = cos θ and y = 1 (1 + z) to get 2 d 2 = 1 p2 8π 4π −ε/2 1 dy [y(1 − y)]−ε/2 (1 − 1 ε) 2 (5.88) For the phase space element integrated over angles one then gets d 2 = π 1−ε/2 1 (2π)2−ε (1 − 1 ε) 2 p2 4 −ε/2 (sin θ)n−3 dθ (5.78).90) We are now ready to calculate Im V ( p 2 ) and V (q 2 = 0). Since in cms p · k = 1 2 p we obtain (as usual. in n = 4 − ε dimensions. Writing IR µ = γµ 2q 2 (q 2 ) (5.198 5 Quantum electrodynamics where 0 ≤ θ ≤ π and due to an azimuthal symmetry of the problem we integrate over all but one angle θ to get dn−1 r = (r2 )(n−3)/2 dr2 (sin θ)n−3 dθ π (n−2)/2 ( 1 (n − 2)) 2 (5.89) 2 (we specify n = 4 − ε and remember that r0 = 0 and p0 = ( p2 )1/2 in cms).

96) The integral over dy gives [ (1 + 1 ε) ( 1 ε)/ (1 + 1 ε) − 2 (1 + 1 ε)/ (2 + ε)] 2 2 2 2 and it is divergent for ε → 0.4 Dispersion calculation of O(α) virtual corrections 199 Fig. where d 2 = dn k 2πδ+ (( p + k)2 )2π δ+ (( p + k)2 ) (2π)n dn l dn l = 2πδ+ (l 2 ) 2π δ+ (l 2 )(2π )n δ (n) (q + l + l ) (2π)n (2π )n 2 1 + 2 2 k q (5.97) .e. We then get the final result ∞ 0 dq 2 (q 2 )ε/2−1 = (−Q 2 )ε/2−1 (1 − 1 ε) ( 1 ε) 2 2 q 2 − Q2 (5. l ) cms system (l + l = p + p = 0) r = l. The divergence comes from the region y ≈ 1 and therefore z ≈ 1.5. the exchanged photon has to have zero momentum. In our specific reference frame z = 1 corresponds to l p. i.94) Therefore Im (q 2 ) = −e2 d 2 (5. (the term proportional to / when integrated must give A p + B p .95) The next step is to use the expression (5. 5. so it does not k / / contribute to the on-shell limit).90) for d 2 . Taking the z axis 2 in the direction p we have k 2 = −2l · p = −2l0 p0 + 2l · p = −2( 1 q0 )2 + 2( 1 q0 )z = − 1 q 2 (1 − z) 2 2 2 where z = cos θ and finally (n = 4 + ε) (Q 2 ) = α 1 1 2 4π 4πµ (1 + 1 ε) 2 × 0 1 ∞ 0 dq 2 q2 q 2 − Q 2 4πµ2 ε/2−1 dy 1 − 2 [y(1 − y)]ε/2 1−y (5.7. To integrate over q 2 (for Q 2 < 0) we change variables twice: q 2 = −Q 2 x and u = 1/(1 + x). In the present case r = 1 (l − l ) and in the (l.

5.5 Coulomb scattering and the IR problem Corrections of order α In the Born approximation the scattering of an electron from a static Coulomb potential A0 (x) = Z e/4π|x|. The pole in ε is due to the lower limit of integration over q 2 . Halzen & Martin (1984)): flux−1 = (2π)4 δ(P + p − P − p ) 1 2E p 2E P |v − V| → 1 1 2E p |v| 2|p| d3 P d3 p d3 p → 2π δ(E − E) 2E P (2π )3 2E p (2π )3 2E p (2π )3 . using (5.98) which can be seen to coincide with (5.200 5 Quantum electrodynamics Fig. We use the static approximation to the general formulae (see. for example.92) we get (n = 4 + ε) IR µ = γµ − α 2π −Q 2 4πµ2 ε/2 2 (1 − 1 ε) 2 ε (1 + 1 ε) ( 1 ε) 2 2 − (1 + ε) 2 (1 + 1 ε) 2 (2 + ε) (5.8. due to a massless electron–positron intermediate state (mass singularity). Finally.82).e.99) |M0 |2 2π δ(E − E) 2|p| (2π )3 2E where (0) |M0 |2 = 1 2 |u( p )γ 0 u( p)|2 ¯ pol Z 2 e4 |q|4 (5. Ai (x) = 0 is described by the diagram in Fig.8.100) and Z e2 /|q|2 is the Fourier transform of the A0 (x). i. 5. 5. According to the general rules the cross section reads 1 d3 p (0) dσ0 = (5.

3 2 1 =2 m2 α 1 3 + ln − 1 + 3γ 4 π εIR 4 4π µ2 (5. As usual 1 2 pol |u( p )γ 0 u( p)|2 = 1 ( p + m)αβ (γ 0 )βδ ( p + m)δγ (γ 0 )γ α = ¯ / / 2 1 2 Tr[· · ·] (5. p )](1 + 1 2 1 )u( p)[1 + (q 2 )] (5. The first term in F1 (q ) comes from the dimensional regularization of the UV divergences.103) we have neglected the terms O(α 2 ) and the corrections which are regular in the IR limit: (q 2 ) and the Pauli formfactor 2 F2 (q ).5 Coulomb scattering and the IR problem 201 for P = P = 0 (remember.106) The above expressions are valid for |q 2 | m 2 and in the minimal subtraction 2 scheme.) Including virtual corrections to order α we obtain therefore (1) (0) dσ0 = dσ0 [1 + 2 1 + 2F1 (q 2 )] (5.101) Performing the necessary calculations we find (0) dσ0 = 1 Z 2 α 2 4 1 |p|4 sin4 ( 1 θ)/E 2 2 [1 − β 2 sin2 ( 1 θ)] d dE δ(E − E ) 2 (5. The amplitude now reads (0) M0 = u( p )(1 + ¯ 1 2 1 )[γ0 + 1 0 ( p. vacuum polarization and the vertex. Virtual corrections of order α are easily included using our results for the electron self-energy.103) ≈ u( p )γ0 u( p)[1 + ¯ + F1 (q 2 )] In the final expression (5. We shall now study real and virtual corrections of order α to the Born cross section.105) (one-quarter (one-half) of the ln(m 2 /4π µ2 ) comes from the dimensional regularization of the UV (IR) divergence) and 2F1 (q 2 ) = −2 m2 −q 2 α1 α 1 −q 2 1 −2 ln 2 + ln2 ln π 4 4π µ2 π εIR m 4 −q 2 −q 2 −q 2 1 −q 2 + ( 1 γ − 1) ln 2 + ln ln 2 2 m 2 4π µ2 m (5. we use the relativistic normalization of 2E particles in a unit volume V = 1 – see Appendix A). (In the amplitude for bremsstrahlung of a real photon we shall only be interested in the limit of a soft photon. The remaining µ2 dependence is due only to our method of regularization of the IR singularity and should cancel out in every cross section free of such singularities.5.104) where using the results of Section 5. Together with the analogous term in 1 it gives (α/4π) ln(−q 2 /m 2 ). .102) where β = |p|/E and p · p = 2|p|2 cos θ.

202 5 Quantum electrodynamics Fig. Using the standard rules we get the following: (1) (1) dσ1γ = |M1γ |2 1 d3 p d3 k 2π δ(E − E − k0 ) |v| (2π)3 2k0 (2π )3 (5.100) and the sum over photon polarizations has been taken. E = E − E is the energy of the undetected real photon. Therefore the previously defined cross section dσ0 is not a physically meaningful (measurable) quantity. The first term comes from the interference of the two diagrams .9. E and E are.109) the amplitude |M0 |2 is defined by (5.104) and dσ1γ which is the cross section for a single photon emission given by the Feynman diagrams shown in Fig. or in other words in the long range nature of the electromagnetic interactions. In consequence the system has a high degree of degeneracy: a final state consisting of a single electron state is not distinguishable by any measurement from an e + n soft (k → 0) γ s state.107) where E is the energy resolution of the detection equipment. (1) The origin of the IR singularity encountered in our cross section dσ0 is in the masslessness of the photon. Later we shall discuss this problem in more general terms.108) where in the soft photon approximation (momentum k neglected in the numerators of the Feynman expressions) (1) (0) |M1γ |2 = |M0 |2 e2 2p · p m2 m2 − − k · pk · p (k · p)2 (k · p )2 (5. and dσ1γ is the differential cross section with a photon in the final state. Following Bloch and Nordsieck and working to order α we introduce a ‘measurable’ cross section σ as follows E σ = 0 dE d (1) (1) (σ + σ1γ ) dE 0 (5.9. as before. One feels that a summation over experimentally indistinguishable final states is necessary. We shall explicitly check that the cross section σ is free of the IR singularity. 5.109) (0) In (5. (1) (1) Working to order α we have to include dσ0 given by (5. 5. We observe that the corrections due to the soft photon emission factorize. the energies of the initial and final electrons in the laboratory system.

We replace d3 k dn−1 k → 2|k|(2π )3 2|k|(2π )n−1 2 2 where |k| = (k1 + · · · + kn−1 )1/2 and introduce spherical coordinates in n dimensions (see (5. Our last task is to calculate the corrections Ii . . We now proceed as follows E dE 0 (1) dσ1γ E dE = 0 dE E d3 k E |p |2 d (1) 2π δ(E − E − k0 ) |M1γ |2 2k0 (2π )3 |p| (2π)3 ∼ = (0) |k|< E dσ0 d3 k e2 dE 0 2|k|(2π )3 0 2p · p m2 m2 × − − k · pk · p (k · p)2 (k · p )2 dE (5.9. .112) and I2 and I3 are self-evident. Let us consider I2 first. 5. σ . At this point we get for the cross section. . the result dσ (0) dσ = 0 [1 + 2 d d where.5 Coulomb scattering and the IR problem 203 shown in Fig. the next two are just the respective amplitudes squared. I1 = e 2 0 |k|< E 1 + 2F1 (q 2 ) + I1 + I2 + I3 ] (5.113) where z = cos θ1 and the integration over the remaining angles has been done in the last step (assuming the integrand depends only on θ1 ). We see also that the dominant contribution to I2 comes from that region of phase space where k p.88)): dn−1 k = =2 d|k| |k|n−2 sinn−3 θ1 sinn−4 θ2 . So.5. sin θn−3 dθ1 . Writing (k · p)2 = |k|2 (E − |p|z)2 we can easily perform the necessary integrations. They are IR divergent and for consistency with the calculation of virtual corrections we must regularize them by working in 4 + εIR dimensions.110) where the last step is legitimate in the soft photon limit (we denote |k| = |k| = k0 ).111) d3 k 2p · p 2|k|(2π )3 k · p k · p (5. dθn−2 π n/2−1 ( 1 n − 1) 2 d|k| |k|n−2 1 −1 dz (1 − z 2 )n/2−2 (5. We see that the probability of a soft photon emission into an interval d|k| is proportional to d|k|/|k| which is easy to understand on dimensional grounds (soft emission does not depend on the electron momentum p). . for example. that part of the radiation tends to be collimated around the direction of motion .

The limit |q 2 | m 2 can also be easily obtained. So working to order α we can neglect the difference. π εIR 2 4π µ2 2 m (5. where α ON = 1/137. The results presented here are for |q 2 | m 2 . The transition probability decreases with decreasing E. In the Breit system the 2 integration over the angle can be performed with the help of Spence’s functions and we get α 1 −q 2 1 ( E)2 −q 2 1 2 m 2 −q 2 I1 ∼ 2 ln 2 + ln ln 2 + ln + 1 γ ln 2 + f (θ) = π εIR m 2 4π µ2 m 4 −q 2 2 m (5. It should be noted that although the final result (5. Otherwise the correction becomes large and we must include higher order contributions which. This can best be formulated if we represent . Adding all the contributions we obtain the final result (0) dσ0 α −q 2 dσ ( E)2 −q 2 3 −q 2 + ln ln 2 + ln 2 = 1+ ln d d π ( E)2 −q 2 m 2 m + f (θ) + const. as we shall see later. exponentiate.116) (remember that −q 2 ∼ 4E 2 ).204 5 Quantum electrodynamics of the initial electron. where θ is the scattering 2 angle in the laboratory (indeed pi · pf = p · p = 2E 2 sin2 ( 1 θ) in the laboratory 2 ˜ ˜ and pi · pf = 2 E 2 in the Breit frame. The coupling constant α is defined by the minimal subtraction renormalization scheme and α = α ON [1 + O(α ON )].115) for |q 2 | m 2 . Our calculation is valid for |(α/π) ln[−q 2 /( E)2 ]| 1. The cross section dσ/d is indeed free of = the IR singularity and depends on the experimental resolution E. To calculate the term I1 it is convenient to choose the reference frame in which p + p = 0 (the Breit frame). the splitting into real and virtual contributions does. depend on the method of regularization of the IR singularity. of course. (5.114) For I3 we get the same result with E changed into E . The final result reads I2 = − α 1 1 ( E)2 1 4E 2 + ln − ln 2 + const.116) does not. so E = E sin( 1 θ)). In this frame ˜ ˜ ˜ p · k = |k|( E − |p| cos θ) ˜ E + |p| cos θ) ˜ ˜ p · k = |k|( where the sign ˜ denotes quantities in the Breit frame and the condition |k| < E ˜ in the laboratory system translates into |k| < E sin( 1 θ). Finally we observe certain systematics in the cancellation of the IR singularities between virtual and real corrections.

(0) respectively. 5.5. An important fact is that in QED the IR divergent terms due to virtual corrections and to real soft emissions exponentiate and cancel each other in the physically meaningful (observable) cross section.10.1 and the factor 2 takes account of 2 two identical interference terms. Let us consider a process in which there are some electrons and non-soft photons in the initial and final states and we represent their momenta by p and p . p ) (n) and let M0 ( p. IR problem to all orders in α The previous results can be generalized to all orders in the electromagnetic coupling constant α (Yennie.5 Coulomb scattering and the IR problem 205 Fig. where the factor 1 follows from the rule of Section 4. Frautschi & Suura 1961). The cancellation occurs between contributions to the cross section represented by the diagrams in each column. p ) = ∞ n=0 (n) M0 ( p.10. separately. 5. the cross section for real emissions by cut diagrams as shown in Fig. Let the lowest order matrix element for our process be M0 ( p.117) . The complete matrix element is then M0 ( p. p ) (5. p ) be the contribution corresponding to all diagrams in which there are in addition n virtual photons.

104)–(5.106). this is to be expected on physical grounds since long wave photons should not see charge distribution at short distance.120) is called the on-shell electron formfactor in the double logarithmic approximation.120) we have changed the dimensional IR regularization into a cut-off by a small photon mass λ: 1/εIR → 1 ln(m 2 /λ2 ) (see Section 5.120) ˆ where M is free of IR singularities.119) and (5.119) and the IR divergent exponent can be read off from our lowest order calculation.    n   (n) i   M0 = m n−i (α B) /i! i=0 (0) where m j is an IR finite function (independent of n) of order α j relative to M0 and each factor α B contains the IR contribution from one virtual photon.2) in order to follow the traditional notation and in the 2 exponent we have only retained terms quadratic in large logarithms. The exponent in (5. Actually. in the socalled double logarithmic approximation. We observe that due to the IR divergences encountered in order-by-order calculation the complete amplitude for purely elastic scattering is zero in the limit λ → 0.   . An important intermediate step in proving (5. ln(−q 2 /m 2 ) and ln(−q 2 /λ2 ). 1961) that the M (n) s have the following structure  (0) M0 = m 0     (1)   M0 = m 0 α B + m 1    (2) 2 2  M0 = m 0 (α B) /2 + m 1 α B + m (5. To write (5. Mel = exp − α −q 2 −q 2 ˆ ln 2 ln 2 Mel π m λ (5. (5. This is just a reflection of the fact that electron scattering is always accompanied by soft photon radiation and the cross section for elastic scattering is not a physical observable. The differential cross section . where E is the energy resolution of the detectors.106).206 5 Quantum electrodynamics It has been shown (Yennie et al.118) . Let us now consider an inclusive cross section summed over any number of undetected real photons in the final state. with total energy E such that E ≤ E.  . For the Coulomb elastic scattering we get from (5.119) is to convince oneself that IR divergences originate from only those virtual photons which have both ends terminating on external lines. Thus we get M0 = exp(α B) ∞ j=0 ˆ m j = exp(α B) M0 (5.

k1 .5. . km ) = σ0 p(k1 ) p(km ) ˆ (5. . p . Writing (Yennie et al.124) where p(k1 ) is given by the expression in the square brackets in (5. each σm is zero due to the IR divergences. .116)): ˜ 2α(B + B) = α −q 2 E2 −q 2 ln 2 ln 2 + ln π E −q 2 m (5. .121) where p and p represent. and they cancel with virtual IR singularities in B giving together (see (5. physical quantity.122) turns out to be free of IR singularities and thus it is a finite.123) using a generalization of (5.128) . 1961) δ E− i ki0 = 1 2π ∞ −∞ dy exp iyE − iy i ki0 (5.5 Coulomb scattering and the IR problem 207 for emission of m undetected real photons with total energy E reads 1 dσm = exp(2α B) dE m! m i=1 d3 ki δ E− 2ki0 ki0 ρm ( p. an inclusive cross section defined as ∞ dσ dσm = dE dE m=0 (5.109) and. km ) i (5.121) and. However.109) ρm ( p.126) d3 ki p(ki )[exp(−iyki0 ) − 1] 2ki0 (5. p . Virtual corrections to all orders in α are included in (5. as before.127) Thus all IR singularities due to real emissions are collected in the exponential factor ˜ ˜ exp(2α B).125) d3 ki p(ki ) 2ki0 (5. momenta of all particles in the initial state and of the detected particles in the final state. as mentioned before. k1 . where B is given by our order α calculation. Soft real photons must terminate exclusively on external lines in order to contribute IR singularities. neglecting for soft photon radiation the energy–momentum conservation constraints one gets the following result: dσ ˜ ˆ 1 = exp[2α(B + B)]σ0 dE 2π where ˜ 2α B = and D= |k|<E |k|<E ∞ −∞ dy exp[iyE + D(y)] (5. as in our order α calculation.

in general. For the forward Compton amplitude and in the limit of the photon momentum k → 0 (the Thomson limit) express the amplitude in terms of the electron propagator.4 Calculate the off-shell electron formfactor (Sudakov formfactor) in the double logarithmic approximation µ ( p1 .3). Hint: calculate the difference P( p 2 + i ) − P( p 2 − i ).1 Derive the Ward identities (5. terms like ln(q 2 /µ2 ). Problems 5. 0). 5.5 Calculate the on-shell (see (5. 5. | p2 | m2 5. .21) using the invariance of the QED Green’s functions under the BRS transformations (Problem 3. The coupling constant α in (5.2 Derive the Ward identity for the truncated connected Green’s function of the Compton amplitude. p2 .26): choose the cms frame p = ( p0 . For a detector resolution E the integration over E up to E is necessary.120)) and the off-shell electron formfactors in the double logarithmic approximation using the light-like gauge n · A = 0 where n 2 = 0. 5. q) = γµ exp − |q 2 | |q 2 | α ln 2 ln 2 2π | p1 | | p2 | |q 2 | 2 2 | p1 |. One should also remember that the IR finite part σ0 contains. Compare with the calculation in the Feynman gauge. study the pole structure of the integrand as a function of k0 and integrate over dk0 first.208 5 Quantum electrodynamics The cross section dσ/dE is finite.3 Study the singularity structure of the photon propagator in the one-loop approximation in the complex p 2 -plane and derive the Landau–Cutkosky rule. Hint: apply the BRS transformation to the Green’s functions 0|T Aµ (x)ω(y)|0 = 0 and 0|T (x) ¯ (y)ω(z)|0 = 0 where ω(x) is the free auxiliary scalar field. where P( p 2 ) = p 2 ( p 2 ) and ( p 2 ) is defined by (5. as expected.15) and (5.121) is α = α(µ2 ). where q 2 ˆ is some characteristic for the process four-momentum squared and µ2 is the renormalization point. Show that in the on-shell renormalization scheme the coupling constant e can be defined by means of the Compton amplitude in the Thomson limit.

The latter are. and also in this section. only subsets of arbitrary renormalization prescription transformations which can be parametrized by a single mass scale parameter µ are discussed. of course. 209 .6 Renormalization group 6. the mass defined as the pole of the propagator is a physical constant) which are directly measurable and therefore renormalization-invariant. For a single-mass-scale-dependent subset of renormalization transformations we derive in the following a differential equation which controls the changes in the renormalized parameters induced by changing µ: the RGE.) Most often. Once the infinities of the theory have been subtracted out by a renormalization prescription R one is still free to perform further finite renormalizations resulting in each case in a different effective renormalization R . there is a residual renormalization scheme dependence in each order of perturbation theory. in each case related differently to physical constants (for example. Actually. the value of the subtraction point in the µ-subtraction schemes or the dimensionful parameter µ in the minimal subtraction prescription. by construction. for example. Each renormalization prescription can be interpreted as a particular reordering of the perturbative expansion and expressing it in terms of the new renormalized constants. all exact physical results renormalization-invariant. The parameter can be. A change in renormalization prescription is compensated by simultaneous changes of the parameters of the theory so as to leave. the set of renormalization prescription transformations {R} does not always have a group structure so the name is partly due to historical reasons.1 Renormalization group equation (RGE) Derivation of the RGE The spirit of the renormalization group approach lies in the observation discussed in Chapter 4 that in a specific theory the renormalized constants such as the couplings or the masses are mathematical parameters which can be varied by arbitrarily changing the renormalization prescription. (In practice.

1) dx 2x(1 − x) ln µ2 x(1 − x) + m 2 2 µ2 x(1 − x) + m 2 1 (6. for arbitrary pairs of Z s: the product Z (Ri . . however. Their relation to the bare Green’s functions B will be R = Z (R) B where Z (R) denotes the appropriate product of renormalization constants defined within the R scheme. namely R = Z (R . R).210 6 Renormalization group Let us assume a particular renormalization scheme R and call R the Rrenormalized 1PI Green’s functions. there exists a relation between both renormalized Green’s functions.3) This representation of Z obeys the composition law (6. R) there is an inverse Z −1 (R .1) but the group multiplication law (6. The relevant quantity will be Z (µ1 . unless R j = Rk . R )Z (R . R) = Z (R. to each element Z (R . R ) and we can define the unit element as Z (R. R = Z (R ) B Obviously. If we choose another renormalization scheme R then. in addition. R) and. µ2 ) which in the one-electron loop approximation reads Z (µ2 . R) R where Z (R . Suppose we work in the µ-subtraction scheme and we want to relate two renormalizations performed at subtraction points µ1 and µ2 . unless m = 0.2) is not obeyed. R) = Z (R . R j )Z (Rk . Among the elements of this set there is the composition law Z (R . R) = 1 The composition law (6.1) is not always defined. Rl ) (6. µ1 ) = 1 + α 2π 1 0 (6. R) for arbitrary R and R . R) = Z (R )/Z (R) Let us now consider the set of all possible Z (R . in general. A simple example of the above considerations is provided by the photon wave-function renormalization constant Z 3 (de Rafael 1979).2) is not. an element of the set Z (R .

4) (for the definition of the renormalization constants Z i see Section 4. of course. −2 λB µ−ε = Z 3 Z 1 λ R Similarly. in 4 − ε dimensions (n) R ( pi . m B . in a trivial way any prescription for a massless theory. The finite limit (for ε → 0) of (6. ε) = Z3 n/2 (n) B ( pi . The RGE will be derived by differentiating relation (6. dimensionful in 4 − ε dimensions. m R . µ. λB . λ R .8) † Similar considerations apply to additive renormalization.6) (6. ε) ∂µ dµ ∂λ R dµ ∂m R d n/2−1 (n) = 1 n Z3 µ Z 3 B ( pi . ε) 2 dµ 1 d = 1 n µ Z 3 (n) ( pi .e. µ.5) We can define now the following functions:† β(λ R . λ R . In the mass-independent renormalization scheme we have Z i = Z i (λ R .4) and using d µ = 0. λ 4 theory can be derived as follows. m R . λB .1). one obtains dλ R ∂ ∂ dm R ∂ (n) + + R ( pi . i. µ. These include the minimal subtraction prescription and. we have −1 m2 = Z 3 Z 0m2 B R B /dµ Differentiating (6. ε) = 1 µ ln Z 3 −→ γ (λ R ) 2 ε→0 dµ µ dm R d −1 γm (λ R . λB . m B and ε. m B . ε) R 2 Z 3 dµ d d 3 −1 λR = λR µ Z3 Z1 − ελ R −→ β(λ R ) ε→0 dµ dµ d γ (λ R . We begin with the relation between the renormalized and bare 1PI Green’s functions in the theory regularized by the dimensional method.1 Renormalization group equation (RGE) 211 In this section we shall discuss the RGE in renormalization schemes such that renormalization constants Z i are mass-independent (mass-independent renormalization schemes).7) (6. m R . ε) = = 1µ ln(Z 3 Z 0 ) −→ γm (λ R ) 2 ε→0 m R dµ dµ (6. ε) is in an obvious way independent of the renormalization scale µ. . ε) (6.4) with respect to the renormalization scale µ for fixed λB . The (n) bare Green’s function B ( pi . ε) = µ (6. m B . λ R . ε) where λ R is defined as dimensionless. i.6.4) exists since the theory is renormalizable. for example. The RGE in.e.

12) . µ) =0 (6. λ. λ. λ. m) = m exp ¯ ¯ λ(t. m(t.10) where µ = exp(−t)µ0 . λ).5)–(6.6) and (6. µ0 exp(−t)) = 0 (6. Solving the RGE Eq.9) suppose for a moment that γ (λ R ) ≡ 0.9) It is important to remember that only in mass-independent renormalization schemes or for m = 0 do the functions β.λ)   dx/β(x) = t   λ m(t. m(0. This happens. γm and γ depend on λ R only.8) exist since the theory is renormalizable) we obtain the final result µ ∂ ∂ ∂ + β(λ R ) + γm (λ R )m R − nγ (λ R ) ∂µ ∂λ R ∂m R (n) R ( pi . We then have a homogeneous equation µ(d/dµ) (n) )( pi . In general they also can depend on the ratio m R /µ.11) as differential equations ¯ ¯ ∂ λ(t.8) simply express the µ-independence of the bare parameters λB and m B .5) (finite limits of relations (6. λ) = λ. λ)/∂t = β(λ) ¯ ¯ ∂ m(t. with the boundary conditions ¯ λ(0. m) = m ¯ (6. (6. Taking the limit ε → 0 in (6. m)/∂t = γm (λ)m ¯ in accord with (6. in the µ-subtraction scheme. µ) = 0 which R can be written in the form (subscript R in λ R and m R will be omitted from now on): − ∂ ∂ ∂ + β(λ) + γm (λ)m ∂t ∂λ ∂m (n) R ( pi .11) Taking the partial derivative ∂/∂t one can rewrite (6.λ) λ dx (γm (x)/β(x)) = m exp 0 1 ¯ γm (λ(t )) dt     (6. m R . As we shall see the meaning of this equation is that the dependence of the function (n) on t and hence on µ can be R ¯ expressed solely through the functions λ(t. λ R . m. and µ0 is some fixed scale. To solve (6.8). m) defined by the equations ¯  ¯ λ(t. for instance.6) and (6. λ R . We recall also that the parameters λ and m are renormalized at µ.9) describes the change of renormalized parameters and multiplicative scaling factors for the renormalized Green’s functions which compensates a change of the renormalization scale µ. λ. λ.212 6 Renormalization group The differential equations (6. m R .

m(t). where λ(t) and m(t) can be understood as the renormalized parameters corresponding to the scale µ exp(t). λ(t). t.10) can be written as follows (n) R ( pi . λ). λ. µ0 ) = f (λ(λ.11) with respect to λ one gets ∂ ∂ ¯ − + β(λ) λ(t. λ. µ) = 0 dµ ∂µ ∂λ ∂m (6. λ(t).16) as can be checked by applying to it the differential operator of (6. m. λ.1 Renormalization group equation (RGE) 213 We also observe that (6. t). ¯ differentiating λ(t. Eq. we can write a Laurent expansion for F ( p 2 . µ exp(t)). m.18) : has a pole at p 2 = m 2 . µ0 ) = 0 (6. m(t). m. m. m. µ0 ) µ exp(t) (6. for instance. λ) given by the first equation of (6.9) reads (n) R ( pi .13) ∂t ∂λ ¯ Using (6. µ).6.16) relates a Green’s function (n) ( pi .9). m. The solution to the inhomogeneous (6. m(t). The renormalized parameters are renormalization-scale dependent. λ. with the renormalized parameters R equal to λ and m at the renormalization scale µ. λ. that this is indeed the case for the physical mass defined as the position of a pole in a propagator and for renormalized S-matrix elements. Being by definition renormalization-scheme-independent P must satisfy the homogeneous RGE: µ d ∂ ∂ ∂ p= µ + β(λ) + γm (λ)m P(λ. µ) ≡ RP(λ. µ exp(t)) exp −n 0 1 ¯ γ (λ(t )) dt (6. λ(t). m. Let us consider the scalar propagator ( p 2 ) which satisfies the equation [R + 2γ (λ)] ( p 2 . λ. µ) = (n) ¯ ¯ R ( pi . µ) − m2 F (6. µ0 (n) ¯ ¯ R ( pi . µ) = 0 Assuming that R + ˜ ( p 2 .11) can be written in the form of the RGEs.9).13) we can easily check that any function f (λ.17) We shall now check. λ). λ) = 0 (6. Any physical quantity P (renormalization-scheme-independent) can be calculated in terms of them: P = P(m. µ0 ) is µ-independent and satisfies the equation µ ∂ ∂ d f = − + β(λ) dµ ∂t ∂λ exp(−t)) = µ ¯ f (λ(t. using (6. µ) = p2 (6. (6. m.14) Therefore the solution to (6. to the Green’s function (n) ¯ ¯ ¯ ¯ R ( pi .19) .15) ¯ ¯ where λ(t) stands for λ(t. λ. µ).

µ) (6. µ) = 0 R ∂ρ ∂m ∂µ Combining (6. m/µ. λ. µ) = 0 F [R + 2γ (λ)]R(λ. Since (n) satisfies (6.214 6 Renormalization group The position of the pole. m. µ) = ρD (n) R ( pi . This is obtained from the 1PI Green’s function (n) by going to the physical poles pi2 = m i2 and multiplying by R n/2 (see (4.9) and m i2 and R satisfy (6.27)).25) .19) in (6. m.9) and (6. namely to discuss the change in the Green’s functions when all momenta are rescaled pi → ρpi (but for fixed µ). We notice first that for m = 0 and for a non-interacting theory (β ≡ γ ≡ 0) the solution to (6. λ) (6. On dimensional grounds it can be therefore written as follows R (n) R (ρpi . Green’s functions for rescaled momenta The RGE is used most frequently in a somewhat different context. m F (λ.21) One must stress that it is an exact S-matrix element which is renormalizationscheme-independent.24) which is the result we are looking for.22) We observe that (n) . is a homogeneous function of order D of variables m. µ) = 0 (6. m. λ. µ) = µ D f (ρ 2 pi · p j /µ2 .23) +m +µ − D (n) (ρpi . m.20). µ) = 0 (6. Using the basic property of such functions we can write for it the following equation: ∂ ∂ ∂ ρ (6.24) reads (n) R (ρpi . µ) and the residue of the pole R(λ. λ.20) We consider next an arbitrary n-scalar particle S-matrix element. λ. m. In perturbation theory truncated at order N there is always a residual renormalization scheme dependence O(λ N +1 ). we have R lim R n/2 2 pi2 →m i (n) = lim RR n/2 2 pi2 →m i (n) = lim [nγ (λ) − nγ (λ)]R n/2 2 pi2 →m i (n) =0 (6. ρ and R µ. m.23) one gets (t = ln ρ) − ∂ ∂ ∂ + β(λ) + (γm − 1)m − nγ (λ) + D ∂t ∂λ ∂m (n) R (ρpi . λ. m. λ. Let us take a Green’s function of dimension D : (n) ∼ [M] D . µ) themselves satisfy RGEs which can be derived by using (6.18) and equating the residua of the poles at p 2 = m 2 F Rm 2 (m.

m. The RGE . and λ. One can say. m. γ (λ) and γm (λ) which can only be calculated in perturbation theory. 2 pi < 0. This solution can also be obtained directly by the ¯ following steps (n) R (exp(t) pi .s. m (t).26) applies in the deep Euclidean region. To use these results in physical applications one must know the functions β(λ). exp(t)µ) × exp −n = ρD γ dt −n γ dt (n) ¯ ¯ R ( pi . therefore. that they are Green’s functions of different theories. away from possible IR problems and new thresholds associated with timelike momenta. Result (6.9) and it reads (n) R (exp(t) pi .26) gives us Green’s functions for the rescaled momenta exp(t) pi (off¯ mass shell) in terms of the ‘effective’ parameters λ(t) and m(t) which are solutions ¯ of (6.h. m(t).26) where ¯ m (t) = m(t)/ρ ¯ ¯ and λ(t) and m(t) satisfy (6. λ. m ¯ ¯ in the l. m (t) in the r.16) and the last step is a change of the mass scale by a factor ρ. µ) = = (n) ¯ ¯ R (exp(t) pi . µ) = (n) ¯ ¯ R ( pi .24) can be written using arguments similar to those used to solve (6.h.1 Renormalization group equation (RGE) 215 i. exp(t) exp(−t)m(t).11). the Green’s functions exhibit canonical scaling (see Chapter 7). λ. relates Green’s functions which both have the same renormalization scale µ. λ(t). λ(t).6. λ(t).26). m (t).16). µ) exp t Dt − n 0 ¯ γ (λ(t )) dt (6. Strictly speaking the RGE solution (6. The general solution to (6.e. RGE in QED We choose to work in the set of mass-independent renormalization schemes and in a covariant gauge defined by the gauge-fixing term: −(1/a)(∂µ Aµ )2 .s. λ(t). µ) exp The first step corresponds to a change of the renormalization point according to (6. exp(t)µ) exp −n γ (t ) dt (n) ¯ ¯ R (exp(t) pi . but different values of the renormalized couplings: λ. unlike (6. We notice that (6.11).

α) = a ¯ (6. a) ∂µ ∂α ∂m ∂a − n F γF (α. see Chapter 5).n F ) ( pi .16) with the running parameters given by (6.27) is obtained by the obvious modification of (6. α.11) and.28) −1 Derivatives are taken for fixed αB and ε. a)∂t = δ(α. ∂ a(t. a)m + δ(α. a) = lim µ (6. γm We address ourselves to the problem of calculating functions. α.29) The solution to (6. a)∂/∂a with δ(α. m. a) where in analogy to (6. a) − n γ γγ (α. m B = Z m m. in addition.27) d α dµ d ln Z m ε→0 dµ d ln Z 3 γγ (α. γ . a). αB = µε Z 3 α in 1/2 1/2 −1 n = 4 − ε dimensions. for example. ε) = lim µ ε→0 ε→0 (n γ .6)–(6. a.6)–(6. 6. γ (λ) and γm (λ) in a mass-independent renormalization scheme. a) = lim µ ε→0 d da = lim −aµ ln Z 3 dµ ε→0 dµ (6. a) = lim 1 µ 2 ε→0 dµ d γF (α. The only new element is the −1 µ-dependence of the gauge-fixing parameter α = Z 3 aB which is responsible for the term δ(α. a) = lim 1 µ ln Z 2 2 ε→0 dµ γm (α. We recall the definitions (6. ε) = lim µ ε→0 ε→0 d λ dµ (6.8).216 6 Renormalization group can then be derived following the steps outlined previously and we get µ ∂ ∂ ∂ ∂ + β(α.2 Calculation of the renormalization group functions β.30) Using our previous experience we can also derive the RGE analogous to (6. β(λ) = lim β(λ. ¯ ¯ ¯ a(0. β(λ). a) = lim β(α.26) for QED. AµB = Z 3 Aµ . µ. B = Z 2 (for definitions of the renormalization constants. where α = e2 /4π.31) fixed λB . a) = 0                            (6. a) + γm (α.ε .8) β(α.

35) dλ    d daν+1  = β(λ) (λaν )  λ2 dλ dλ We arrive. ε) = −ελ + β(λ)      2 da1 β(λ) = λ (6. am (λ) = am1 λ + am2 λ2 + · · · + amn λn Using (6. the simple pole in ε.6) we get −1 β(λ.33) in the following form β(λ.34) Substituting expansion (6. for instance.32) in (6. for ε → 0.e.32) εν ν −2 where. therefore. (ii) The residua of the higher order poles of Z λ are totally determined by a1 (λ). This reflects an important aspect of renormalization theory. of course. ε) = −ελ − λZ λ µ d Zλ dµ (6. up to 1/ε N in the N th order but their residua are determined by lower order perturbation expansion.6. In each successive order of perturbation theory only one new divergence arises. . at two important conclusions (i) β(λ) is totally determined by the residua of the simple poles in ε of Z λ . ε). Equating the residua of the poles in ε we obtain the following final result  β(λ. ε) dµ dµ dλ dλ we can rewrite (6. One can easily check in particular that amn = 0 for m < n. ε) in the form of a series β(λ. There will. be multiple poles. ε)λ dZ λ /dλ = 0 (6. Z λ = Z 3 Z 1 for the λ the series in λ 4 theory. .34) we look for the solution for β(λ.2 Calculation of renormalization group functions 217 In perturbation theory we can calculate the renormalization constants Z i which relate λ to λB aν (λ) λB = µε Z λ λ = µε 1 + λ (6. exists). ε) = ν βν εν which is regular at ε = 0 (since our theory is renormalizable a finite limit of β(λ. ε)Z λ + ελZ λ + β(λ. i. The coefficients aν (λ) are a1 (λ) = a11 λ + a12 λ2 + · · · + a1n λn . .33) Since µ dZ λ dλ dZ λ d Zλ = µ = β(λ.

32) is replaced by αB = µε Z α α with Z α = Z 3 . (6.37) one obtains the following renormalization group functions in the λ 4 theory:  3λ2  3   β(λ) = + O(λ )   16π 2     2 λ 1 λ 5 3 − + O(λ ) γm (λ) = (6.3 for the renormalization constants Z i in the λ 4 theory  3λ   Z1 = 1 +   16π 2 ε     2 2 λ 2λ λ (6. Using the results of Section 4.37) −1 (Z m = Z 0 Z 3 in the λ 4 theory).35). B Zm = 1 + ν (ν) Zm εν (1) dZ m dλ (6. Using (5.36) and (6.36) and γm (λ) = 1 λ 2 where m2 = Z mm2.39)  2 16π 2 12 (16π 2 )2    2   1 λ  3  γ (λ) = + O(λ )  2 12 16π −1 For QED relation (6.218 6 Renormalization group In a quite analogous way we also find the results for γ (λ) and γm (λ): γ (λ) = − 1 λ 2 where Z3 = 1 + ν (ν) Z3 εν d (1) Z (λ) dλ 3 (6.31) we get Zα = 1 + α 21 + π 3ε α π 2 b2 + · · · .38) Z0 = 1 + − + 16π 2 ε 2(16π 2 )ε (16π 2 )2 ε 2      λ2    Z3 = 1 − 12(16π 2 )ε and with help of (6.

40) Though not calculated explicitly. effective coupling constant 219 and therefore. Note also that β(e) defined as β(e) = µ(d/dµ)e is given by β(e) = 2πβ(α)/e. if t→−∞ ¯ ˜ lim λ(t. for example. or small. λ) = λ ≡ λ− (6. A zero of β(x) ˜ is called a fixed point. ¯ The solution λ(t.41) 6. If t→∞ ¯ ˜ lim λ(t. Eq. as follows ¯ from (6.3 Fixed points. We shall discuss here only theories with one coupling constant.6.11) must for t → ∞ and t → −∞ approach the zero of β(x) nearest to λ or go to infinity if there is no zero to approach. the second order contribution (α/π)2 b2 has been singled out for the sake of the discussion later. momentum behaviour of a quantum field theory which is determined. (6. Otherwise the renormalization scale µ could not vary arbitrarily and the theory would need some natural cut-off . ˜ In both cases one says that λ is in the domain of attraction of the fixed point λ. A .3 Fixed points. The latter is determined by ¯ λ(t. λ) = λ ≡ λ+ (6. from (6.11) to be valid in the whole range −∞ < t < ∞.26) by the behaviour of the ‘effective’ coupling constant λ(t) as t → ±∞.35) β(α) = α α2 + π3 α π 2 b2 + · · · (6.11) where β(λ) can be calculated perturbatively from (6.λ) λ dx/β(x) = t (6. First we assume (6. A renormalizable theory with perturbative expansion valid only in a certain energy range cut-off by a scale parameter (if.42) it is called a UV stable fixed point. Let λ be the fixed point nearest to λ. effective coupling constant Fixed points The main use of the renormalization group is in discussing the large. new interactions switch on at a mass scale ) is not necessarily uninteresting physically and we shall come back to this point later.11) provides a basis for a physically very important classification of different theories.43) it is called an IR stable fixed point. Therefore β(e) = e3 4 e5 8b2 + · · · + 16π 2 3 (16π 2 )2 (6.35). λ) of (6.

1. For theories with several couplings the origin may be UV stable for some couplings and IR stable for others. for a given theory. strictly speaking. In an asymptotically free theory the effective coupling vanishes for infinite momenta. Consequently. The λ 4 theory (with λ ≥ 0) and QED are IR stable (β(λ) is positive for small λ) whereas QCD is asymptotically free (β(λ) is negative for small λ). in general. has two different theories (or two different phases of one theory). then the λ is a UV ˜ β (λ)|λ > 0. domain is a region which lies between two zeros of β. by the sign of the derivative β (λ) at ˜ λ = λ. for two different values of the coupling λ lying in two different domains one. a speculation. λ) for t → ∞ in λ 4 and QED. λ is an IR stable fixed point.11) and Fig. (If λ ˜ ˜ M < 0 (> 0). if it is a simple zero.1 it is clear that for ˜ β (λ)|λ < 0. of odd order M. 6. the only place that we can compute β(λ) in perturbation theory is near λ = 0 and its behaviour for increasing λ is essentially a speculation.2 we show some examples of what β(λ) might look like.220 6 Renormalization group Fig. the behaviour of the ¯ coupling constant λ(t. 6. All single coupling theories are either asymptotically free or IR stable. Of course. Of course. From (6. In Fig.11) the effective coupling ¯ ˜ ˜ ˜ λ(t. ˜ ˜ ˜ In other words. both for λ < λ and λ > λ. and for t → −∞ in QCD also remains.) (IR) stable fixed point for β(λ)/(λ − λ) It is obvious but important to notice that for all single coupling theories the value λ = 0 is a fixed point (β(0) = 0 in perturbation theory). . λ is a UV stable fixed point. 6. λ) → λ for t → ∞ if β (λ)|λ < 0. to satisfy (6. A theory is said to be asymptotically free if λ = 0 is a UV stable fixed point and to be IR stable if it is an IR stable fixed point. ˜ ˜ β (λ)|λ > 0. The nature of the fixed point is determined. and λ(t. λ) → λ for t → −∞ if ˜ ˜ is a multiple zero of β(λ).

2(c)) and that it diverges at infinity in such a way that ∞ λ dx/β(x) = tc < ∞ 4 It is likely that this situation occurs in the λ theory (see (6. If λ is in region I.2(a) we have distinguished two domains of attraction. It is generally believed (but not proved) that IR slavery in QCD is responsible for confinement. For such a .2(c) shows another alternative for QED and Fig. 6.3 Fixed points. λ) must diverge when t → −∞ and β(x) must be such that λ dx/β(x) is divergent.11) ¯ λ(−∞.2. Fig. instead λ(t. If the effective coupling becomes infinite in the IR region one often refers to this situation as IR slavery. 6. if λ is in the region II then the theory is ¯ not asymptotically free. In Fig.2(b) shows what might happen in QED: IR stability and fixed point λ+ for t → ∞. 6. effective coupling constant 221 Fig. On the other hand. suppose that the β-function is always positive (as in Fig.2(d) there is no IR stable fixed point. λ) → λ+ when t → ∞. 6.2(d) shows what we hope happens in QCD. Finally. To satisfy (6. 6. Fig. In Fig. the theory is asymptotically free and for t → −∞ (low momentum) λ approaches the fixed point λ− . 6.39)).6.λ) λ dx/β(x) = −∞ ∞ ¯ λ(t. 6.

where α = λ for λ 4 and α = e2 /4π for QED and QCD. an (t) = (b1 t/π)n−1 + O(t n−2 ) so that α(t) = α 1 + n=1 (6. With β(α). can be transformed into an upper bound for the number of elementary particles in the theory or their masses (see Problem 6.48) Comparing the coefficients at the same powers of α and integrating over t one gets a2 (t) = b1 t/π a3 (t) = (b1 t/π)2 + b2 t/π . .47) Therefore dam (t) m α = dt m=2 n=1 bn am (t)α m α+ πn m=2 n+1 (6. Effective coupling constant We will now study the explicit behaviour of the effective coupling constant in perturbation theory. µ0 ) (6. For the theory to be valid up to c the following equation must be then satisfied ¯ λ( c .λ) λ(µ0 ) dx/β(x) = ln( c .46) and β(α(t)) = n=1 α n+1 (t)bn /π n (6. . a1 = 1 (6.222 6 Renormalization group theory the renormalized perturbative expansion cannot be consistent for all energies (unless λ(µ0 ) = 0 for any µ0 ). λ) = ∞) which.45) am (t)α m . in turn.50) .4). given by an expansion β(α) = α[(α/π)b1 + (α/π)2 b2 + (α/π)3 b3 + · · ·] we can calculate α(t) integrating (6.49) (b1 tα/π )n + O(t n−1 α n ) (6.44) For a given c the equation gives upper bounds for the low energy coupling ¯ constants (assuming λ( c .12) order-by-order in α. We expand α(t) = m=1 (6. The theory must be modified by new interactions which switch on at some mass scale c which provides a natural cut-off or must undergo a phase transition at c .

bn = 0 for n > 2) we get the following α(t) = α1 (t) 1 − α1 (t) b2 α ln 1 − b1 t π b1 π (6.51). Taking 3 2 the limit t → −∞ in (6. In the second approximation (b1 = 0. . α = 0 is a UV stable fixed point. 0 0 1 B1 π Cn π α − + + B1 ln α α(t) α(t) n=1 n α(t) π n − α π n (6.52) Cn = (−1)n+1 Truncating 1/β(x) at a given order in x gives us the corresponding approximate solution for α(t). α(t) = − t→−∞ π b2 1 1 π + 3 2 ln + O 2 b1 t t t b1 t (6. λ) ¯ m(t. m. . depending on whether we use 0 2 2 (6.26).λ) λ [γm (x)/β(x)] dx (6.51) 1 − (α/π)b1 t The same result can be obtained by expanding 1/β(x) in the integral (6. .2) and b2 = 1 (de Rafael & Rosner 1974). In the first approximation (b1 = 0. b2 = 0.53) where a1 (t) is called the leading-logarithm-result (6. effective coupling constant 223 2 For large t = 1 ln(µ2 /µ2 ) or t = 1 ln( p 2 / p0 ). In this way one gets the series b1 t = where Bn = bn+1 /b1 and B1 1 0 . µ.54) we see explicitly that the point α = 0 is an IR stable fixed point. therefore..55) . In this approximation the effective coupling constant reads α α(t) = .51). λ) = m exp ¯ ¯ λ(t.16) or (6.. we can consider the so-called leading logarithm approximation in which we neglect terms O(t n−1 α n ) as compared to the terms O(t n α n ). .6. . Another renormalization group parameter is the running mass m(t. .53). bn = 0 for n > 1) we recover the leading-logarithm-result (6.3 Fixed points.11) in powers of x and directly integrating over x. . α = α(t = 0) (6. . In QED b1 = + 2 (Section 5. In QCD b1 < 0 and b2 < 0. 0 B2 · · · Bn+1 B1 · · · Bn 1 B1 · · · Bn−1 .

for QCD. in a theory with the UV fixed point λ+ ¯ one gets exp(−t)m(t) = m exp{−t[1 − γm (λ+ )]} exp − ¯ ¯ λ(t) λ [dx/β(x)][γm (λ+ ) − γm (x)] (6. In particular.58) where Fi (λ) = 1 + O(λ2 ) (for the sake of definiteness we restrict our discussion to the case of the trilinear coupling theories (QED. another one will give a coupling λ λ = F(λ) = λ + O(λ2 ) for λ 4 = λ + O(λ3 ) for trilinear coupling theories (6. and. If λ is the coupling in one scheme. with p > 0. γm and γ .56) Whether in the high energy limit masses can be neglected or not depends on γm (λ+ ) and on the integral in the exponent. For small enough λ we can assume that dF/dλ = 0 and Fi = 0. for ¯ example. For asymptotically free theories λ+ = 0 (and γm (0) = 0) so only the integral is responsible for the large t limit of m(t). The parameters of one renormalization scheme can be determined in terms of those in another scheme. modifications for λ 4 are trivial).224 6 Renormalization group The behaviour of the Green’s functions for rescaled momenta exp(+t) pi is controlled by exp(−t)m(t). 6. We . mass corrections are suppressed by exp(−t)t p as t → ∞.57) In the tree approximation the coupling is scheme-independent. Z 3 (λ ) = Z 3 (λ)F3 (λ) (6.4 Renormalization scheme and gauge dependence of the RGE parameters Renormalization scheme dependence The last problem we would like to discuss in this chapter is the renormalizationscheme dependence within the class of mass-independent renormalization procedures and the gauge dependence of the perturbative results for the functions β. The wave-function renormalization Z 3 and mass renormalization Z m will be given in the new scheme by Z m (λ ) = Z m (λ)Fm (λ). QCD).

IR stability) is scheme-independent. renormalization-scale-independent.ε γm (λ ) = γm (λ) + 1 d ln Fm β(λ) 2 dλ 1 d ln F3 β(λ)  2 dλ         ˜ ˜ ˜ Using (6. (iii) ˜ ˜ γ (λ ) = γ (λ).6. and have physical significance. In mass-dependent schemes the renormalization constants Z i and consequently the coefficients bi . .e. γi depend on the ratio m/µ and obviously cannot be equal to those in the mass-independent schemes.59) the following results can easily be obtained (let λ and λ = F(λ) be fixed points in both schemes) (i) dβ dλ (ii) If β(λ) = −2b0 λ3 − 2b1 λ5 + · · · then β (λ ) = −2b0 λ 3 − 2b1 λ 5 + · · · (6. In particular within this class of renormalization schemes and in the leading logarithm approximation. respectively (scale factors for the field and the mass).60) i. the first two coefficients of the expansion for β are for mass-independent schemes.11). 0 This is again important since γ0 and γm determine the leading behaviour for t → ±∞ of the exponents in (6.ε d dµ λB.59) λB. ˜ ˜ γm (λ ) = γm (λ) This again must be so since the values of γ and γm at a fixed point determine the scaling behaviour and masses. (iv) If γ (λ) = γ0 λ2 + O(λ4 ) then γ (λ ) = γ0 λ 2 + O(λ 4 ) and similarly for γm (λ).4 Renormalization scheme and gauge dependence 225 get therefore d β (λ ) = µ dµ γ (λ ) = 1 µ 2 λ = (dF/dλ)β(λ) ln Z 3 (λ ) = γ (λ) +           (6. We stress that these properties hold only in the class of mass-independent renormalization schemes. Properties (i) and (ii) are important to ensure that the nature of the fixed points (asymptotic freedom. α(q 2 ) is scheme-independent.61) ˜ λ = dβ dλ ˜ λ (6.26) and (6.

226 6 Renormalization group Effective α in QED It is of some interest to find a link between the effective coupling constant in QED and the ‘on-shell’ a ON = 1/137 defined by (12. where α(m 2 ) is defined in the MS scheme. α(µ)) is renormalizationscheme-independent we can change µ2 → q 2 to get αinv (q 2 ) = α(q 2 )d(1.11) We recall first that α(q 2 ) = 0 q 2 →0 (it is an IR fixed point). we see that in QED the product αB dB (q 2 ) is finite.32) for αinv (q 2 ) = α(q 2 ) q 2 →∞ (6. An important observation is that in QED the quantity αinv (q 2 ) = αd(q 2 /µ2 . Turning again to QED we observe that α ON = lim αinv (q 2 ) q 2 →0 (6. α) µν (q) q µ q ν αd(q 2 /µ2 .26). α) q µq ν − αai 2 2 q2 q2 (q ) (6. αd(q 2 . Relation (6. Using (6.63) we arrive at the result (6. µ2 . in the on-shell renormalization scheme ON (q 2 = 0) = 0 and given the scheme independence of αinv defined by (6.64). α(µ2 )) in the first order in α and with µ2 = q 2 one gets (6. α(q 2 )) Using (5. Thus. µ2 .64) Indeed. α) = 1− 1 (q 2 /µ2 .66) .86).63) does not hold in QCD because then the Ward identities do not imply Z 1 = Z 2 . To this end we consider the photon propagator renormalized in some mass-independent renormalization scheme α D µν (q) = −i g µν − where α = α(µ) and d(q 2 /µ2 . α) = αB dB (q 2 ) (6. α) and αB dB (q 2 ) involves uncancelled renormalization constants. in QCD the relation between αd(q 2 . Since the product α(µ)d(q 2 /µ2 .62) The function is the scalar factor of the vacuum polarization tensor defined in (5. α) = (α/Z 3 )Z 3 d(q 2 /µ2 .32) we also see that α ON = α(m 2 ). It remains to find the relation between the αinv (q 2 ) and the running α(q 2 ) defined as the solution to (6.63) and (5. Incidentally. α) is then renormalization-scheme-dependent and αB dB (q 2 ) is not finite.63) is renormalization-scheme-independent because the bare quantities are.65) (q 2 /µ2 .

in general.1) and β(α) is. in other words. we can expand α (µ) = α(µ) + O(α 2 ) (6. Nevertheless one can prove that: (i) in any renormalization scheme the coefficient b1 in the expansion α α β(α) = α b1 + b2 π π 2 + ··· (6. However.70) α = α in the zeroth order but the coefficients of the higher order terms might depend. the transverse part of the propagator. Now we consider the change of both . This follows immediately from the fact that the renormalization constant Z 3 is gauge-independent. In non-abelian gauge theories the renormalization programme is more complicated (see Section 8. gauge-dependent. for example. α(µ2 )) = A1 α(µ2 ) ln(|q 2 |/µ2 ) + n=2 An α n (µ2 ) lnn−1 (|q 2 |/µ2 ) + non-leading terms The value of αinv (q 2 ) in the limit q 2 = 0 can be obtained in terms of α(q 2 ) from knowledge of the first terms of the perturbation expansion again using (5.67) The inverse relation to first order in α ON = αinv (0) reads α(q 2 ) q 2 →0 α ON ln(|q 2 |/m 2 ) 1 = α ON 1 + α ON |q 2 | ln 2 3π m (6.6).5) (q 2 /µ2 . as the value of the three-point function at some value µ2 of the kinematical invariants or according to the minimal subtraction scheme. It renormalizes the gauge-invariant operator Fµν F µν or. both can be defined. which is gauge-invariant (see also Problem 6.32) as the finite limit of the expression αinv (q 2 ) = 0 α(q 2 ) = . on the gauge parameter a.4 Renormalization scheme and gauge dependence 227 This result actually holds to any order in α in the leading approximation because (see Problem 6.68) Gauge dependence of the β-function In QED the β(α)-function is gauge-independent. in general. To prove the first theorem we note that if we change the gauge we might change α(µ) to α (µ).69) is gauge-independent.6. 2 →0 1 − [α(q 2 )/3π] ln(|q 2 |/m 2 ) 0 q (6. (ii) in the minimal subtraction scheme β(α) is gaugeindependent.

3 Consider the QED of n h ‘heavy’ fermions h . Obtain the effective field theory for the light fields Aµ and l . and n l ‘light’ fermions l each with zero mass.75) + Zα + α α +O 2 =0 da ε da da ε (1) The only way this equation can be satisfied is if dα/da = 0 and dZ α /da = 0.ε fixed = Zα (6.73) we get dα 1 dα (1) 1 dZ (1) (6. γ . (See also Gross (1976). a)/ε + Z α (α.72) and using the relation αB = µε Z α α we have 0= d (Z α α) da αB . gauge-independent parameter. Theorem (ii) can be proved as follows: consider the bare coupling constant αB of the theory. By its definition it is a fixed.1 Calculate the renormalization group functions β. Therefore d αB da ε fixed = daB d αB = 0 da daB d d α + α Zα da da (6. Problems 6. a)/ε 2 + · · · (6. in the one-loop approximation by ‘integrating out’ the heavy fermions from the Green’s functions with light field external legs: take the .73) In the minimal subtraction scheme (1) (2) Z α = 1 + Z α (α.228 6 Renormalization group α (µ) and α(µ) under a change of the renormalization point µ to calculate β (α ) and β(α) β (α ) = µ d α dµ αB . γm in QED in the momentumsubtraction renormalization scheme. Thus in the minimal subtraction scheme the renormalized coupling constant α(µ) is gauge-independent and consequently the same is true for the β(α) = µ(d/dµ)α. 6. and study the effect of the meson selfinteraction on the asymptotic freedom of the theory.71) because α(µ + dµ) = α(µ) + O(α 2 ).) 6.74) into (6. a) which would destroy the following argument. Inserting the expansion (6. Write down the system of coupled RGEs for the gauge coupling g and for the meson self-coupling λ. This proves theorem (i).2 Consider a two-coupling theory: a non-abelian gauge theory with a scalar multiplet transforming according to the representation R of the gauge group.aB .ε fixed =µ d [α(µ) + O(α 2 )] = β(α) + O(a 3 ) dµ (6. each with non-vanishing mass M.74) and in particular in this scheme Z α does not contain any constant term C(α. to the lowest order in perturbation theory.

show by taking the derivative with respect to the gauge parameter a that the transverse part of the propagator is gauge-independent. 4 theory in the Euclidean space (Problem 4.4 Consider the standard electroweak theory based on the gauge group SU (2) × U (1) with the gauge symmetry spontaneously broken (see Chapter 12) by one complex scalar doublet.6 Using the functional integral representation for the bare full two-photon Green’s function in QED. (5. 6.44) and using the lowest order result for the β-function estimate the upper bound for the Higgs particle mass (Dasher & Neuberger 1983).8 Consider λ β-function. What symmetry does it possess when two couplings are equal? Discuss the fixed point structure for these couplings.3). Assume that for consistency the relation m H ≤ c must be satisfied. reads (q 2 /µ2 .7 The model discussed in Problem 4. 6. Study the RGE for the Green’s functions of the effective theory. Calculate the two-loop . In the latter case understand the role of a finite renormalization of the coupling constant in absorbing the effect of the heavy mass. Adopting the point of view expressed by (6. 6. 6.1 (taken now for m = M) has two coupling constants. α(µ2 )) = A1 α(µ2 ) ln(|q 2 |/µ2 ) + n=2 An α n (µ2 ) lnn−1 (|q 2 |/µ2 ) + non-leading terms where the coefficients An are determined in terms of the b1 and b2 of the β-function.26).Problems 229 photon propagator as an example and check the validity of the Appelquist–Carazzone decoupling theorem in the on-shell renormalization scheme (Appelquist & Carazzone 1975) and in the minimal subtraction scheme (Ovrut & Schnitzer 1981).5 Using the RGE in a mass-independent renormalization scheme show that the vacuum polarization in QED. 6.

t)} = δ(x − y) and the scale dimension d is then defined so that the commutation relations (7.2). Its lagrangian does not contain dimensionful constants and should be invariant under the transformation (7. If the configuration variables are scaled down as in (7.7 Scale invariance and operator product expansion 7. it is closely related to the famous Bjorken scaling in the deep inelastic lepton–hadron scattering. t).3) where the constant d is called the scale dimension of the field (x). t). Scale invariance is the invariance under the group of scale transformations.1) where ε is a real number. ϕ(y.5) (y.2) where T (ε) is a finite-dimensional representation of the group (for classical fields (x ) = T (ε) (x)). Let us consider first a free massless field theory. Obviously. all such transformations form an abelian group. in general. We assume it to be fully reducible and therefore we can write T (ε) = exp (−d ε) (7. subject to a unitary transformation U (ε) (compare with Section 1. For free fields one has the canonical equal-time commutation relations [ϕ(x. † (7. Scale transformation for the coordinates is defined as follows x → x = exp(ε)x (7. Among other things.4) 230 .1 Scale invariance Scale transformations An important concept in quantum field theory is that of scale invariance.4) (7. t)] = iδ(x − y) ˙ { (x.4) U (ε) (x)U −1 (ε) = T −1 (ε) (eε x) (7.1) local fields (x) (in this section we denote by (x) any scalar field ϕ(x) or spinor field (x)) will be.

5) implies that these so-called canonical scale dimensions are dϕ = 1 for scalar fields and d = 3 for spinor fields. scale transformation is a transformation of dynamical variables (the fields) but not of the dimensionful parameters such as masses since we want to .8) + =ϕ. the following lagrangian density L=i¯/ ∂ + 1 (∂µ ϕ)(∂ µ ϕ) + g ¯ γ5 ϕ − (λ/4!)ϕ 4 2 3 2 (7. d = δL = U LU −1 − L = L(exp(dφ ε)φ(x )) − L(φ(x)) = it is easy to calculate (see (1.9) (one can also integrate explicitly using (7.6) and the values dϕ = 1. of course.7.4) and (7. as usual.29)) ∂L δ ∂ ∂L ∂ δ ∂(∂ /∂ xµ ) ∂ xµ (7.2) one sees that invariance of (7. Multiplying relations (7. one can easily check that a lagrangian which does not contain dimensionful constants is invariant under the scale transformation (7. would.6) For interacting field theories one still has the canonical commutation relations (7.4) and (7.2) (of course. for instance.1 Scale invariance 231 and (7. strictly speaking. In an infinitesimal form the scale transformation (7. the surface terms at infinity).4) and (7.2). Terms like m 2 2 (x).5) by operators U and U −1 and using (7. it is the action integral which is scale-invariant).8) and neglecting. Take. if added to (7. We see that the canonical 2 scale dimensions of fields coincide with their ordinary dimensions defined on purely dimensional grounds (see Section 4.7) Using (7.7). = ε(4 + xµ (∂/∂ xµ ))L(x) where we have used the relation δ ∂ ∂ xν (x) = ∂ ∂ δφ(x) = ε d + 1 + x · (∂/∂ xν ) (x) ∂ xν ∂x This result is expected on general grounds since in a lagrangian which does not contain dimensionful constants each term has a canonical dimension equal to its ordinary dimension which is four.5) for bare fields. Therefore the corresponding action integral is invariant under scale transformations δI = d4 x δL(x) = 0 (7. Defining scale dimensions for bare fields as canonical ones.2) is as follows: U (ε) (x)U −1 (ε) = (x) + ε(d + xµ (∂/∂ xµ )) (x) + O(ε2 ) (7. Notice the difference between scale dimension and ordinary dimension which is four for the term m 2 2 .5) remain invariant under the scale transformation. break the scale invariance of the lagrangian as their scale dimension is two.

12) From (7.13) (7. one gets U (ω)U (ε) = U (ε)U (ω) and [D. in general.232 7 Scale invariance and OPE stay in the framework of the same physical theory. Turning back to an exactly scale-invariant world we can introduce the generator D of the scale transformation U (ε) = exp(iε D) (7.14) . Pµ ] = Pµ Similarly. However. Mµν ] = 0 (7.6) we get the following differential representation of the generator D i[D. Terms like m 2 2 (x) break scale invariance explicitly.10) and study its commutation relations with generators of the Poincar´ group (for e conventions see Problem 7.1)  U (a) = exp(ia µ Pµ )  (7.11) and  1 µν U (ω) = exp(− 2 iω Mµν ) By comparing two sequences of dilatation and translation (x) → exp(d ε) (exp(ε)x) → exp(d ε) (exp(ε)x + aµ ) ε aµ and (x) → (x + aµ ) → exp(d ε) (exp(ε)(x + aµ )) aµ ε we conclude that U (ε)U (a) = U (exp(ε)a)U (ε) or i[D. (x)] = (d + xµ (∂/∂ xµ )) (x) (7. This is due to the necessity of renormalization which introduces a dimensionful parameter. also broken. it is already worth recognizing at this point that in quantum field theory described by a scale-invariant lagrangian the scale invariance is.

be defined even in the absence of scale invariance. The dilatation current can be written in a more compact way if we introduce a modified energy–momentum tensor µν . ∂µ µν =0 and Pµ = Mµν = d3 x can 0µ = 0ν d3 x − xν 0µ 0µ ) d3 x (xµ . The current is conserved in the presence of the symmetry. but it may.17) The new tensor has the following properties µν = νµ .15) where µ is the conjugate momentum. The conserved dilatation current can be deduced by comparing an infinitesimal change. under the scale transformation of the lagrangian density of a scale-invariant theory (7. obtained without the use of the Euler–Lagrange equations of motion. The Noether dilatation current is therefore the following Dµ (x) = or Dµ (x) = where can µν µd µ (d + xν (∂/∂ xν )) (x) − xµ L (x) + x ν can µν (7. ) (7. of course.16) = µ ∂ ∂xν (x) − gµν L is the canonical energy–momentum tensor.8) δL = ε 4 + xµ ∂ ∂ (xµ L) L=ε ∂ xµ ∂ xµ with an alternate expression for δL which does use Euler–Lagrange equations δL = (∂/∂ xµ )( µδ ) = (ϕ. Let us consider first a scalar field theory and define µν as follows: µν = can µν − 1 (∂µ ∂ν − gµν )ϕ 2 6 (7.7.1 Scale invariance 233 Dilatation current The dilatation generator D can be written in terms of the Noether dilatation current Dµ (x).

17) can be generalized to include fields with non-zero spin (Callan et al.20a) in the context of searching for an energy–momentum tensor whose matrix elements between physical states are finite.46) to the µ µ from the anomalous scale invariance breaking. Coleman & Jackiw (1970) when looking for an object with the ‘soft’ divergences (7. must be separately verified (see. Section 10. µν The new tensor was invented by Callan.22) . see Adler.234 7 Scale invariance and OPE i. Coleman & Jackiw of the finiteness of matrix elements of the µν is incomplete because it ignores the non-soft contribution (7. Actually. Collins & Duncan (1977) for the complete discussion. Eq. µ is a legitimate energy–momentum tensor of our theory.20) for a scale-invariant scalar field theory (and µ µ = m2ϕ2 (7.16) (d = 1) reads µν Dµ = x ν µν − 1 ∂ν (xµ ∂ ν − x ν ∂µ )ϕ 2 6 The last term is the divergence of an antisymmetric tensor. Using the relation = ∂µ ϕ it is easy to check that in terms of µν the Dµ of (7.19) An important property of the new tensor is that its trace vanishes µ µ =0 (7. for instance.e. the proof of Callan.20a) if a mass term is added) while the trace of can contains singular derivative terms.18) (7. In a renormalizable field theory the finiteness of matrix elements of the T-ordered products of local currents is an additional requirement.21) where Tµν = µ ∂ν − gµν L + 1 ∂ λ ( 2 λ µν − µ λν − ν λµ ) (7. which.1). 1970): µν = Tµν − 1 (∂µ ∂ν − gµν )ϕ 2 6 (7. Thus we redefine Dµ (x) = x ν and ∂ µ Dµ (x) = µ µ µν (x) (7. in general. so it does not contribute to ∂ µ Dµ and to D. (7.

which in addition to translations. In terms of Tµν the generators of the Lorentz transformation are Mµν = d3 x (xµ T0ν − xν T0µ ) The tensor µν defined by (7. for example. where λµ λν Mλµν = x µ Tcan − x ν Tcan + and M µν = we get µν νµ Tcan − Tcan + ∂λ ( λ µν ) = 0 d3 x M0µν Eq.22) follows if we require ∂µ T µν = ∂ν T µν = 0. In particular its trace is a measure of scale invariance breaking. . γν ] for 2 4 spin one-half fields. Lorentz transformations and scale transformation also contains special conformal transformations.e.24) = f (x) gαβ dx α dx β ≡ f (x) ds 2 where f (x) is a certain scalar function. µν = − 1 iσµν = 1 [γµ . A conformal (i. angle-preserving) transformation must leave invariant the ratio dx α dyα |dx| |dy| So it must have the property that ds 2 = gαβ dx α dx β = gαβ (∂ x α /∂ x γ )(∂ x β /∂ x δ ) dx γ dx δ (7.21) has the same properties as the tensor (7. For infinitesimal transformations x µ = x µ + εµ (x) λ µν † It can be constructed by noting that from conservation of the angular momentum current ∂λ Mλµν = 0. (7.17) defined for scalar fields only.1 Scale invariance 235 is the conventional symmetric energy–momentum tensor† and where all fields of the theory are assembled into a vector whereas the scalar fields are denoted by ϕ. The spin matrix µν is defined by the transformation properties of the fields under an infinitesimal Lorentz transformation δ = − 1 (xµ ∂ν 2 − xν ∂µ + µν )ωµν (7. Conformal transformations It turns out that many scale-invariant theories are invariant under a larger symmetry group of transformations. µν = 0 for spin 0 fields. the so-called conformal group.7.23) and reads.

ωαβ . see Problem 7.27) are conserved for any εν = εν (x.25) (7. We see that for n > 2. x µ + bµ x 2 = . In this case the currents ( j ε )µ (x) = µν (x)εν (x) (7. requires a separate discussion. From (7.26) = x µ + aµ. . = µν (ω)xν . It can be checked that such a µν exists in all renormalizable field theories of fields of spin ≤ 1. This equation gives (n − 2)∂α ∂β h(x) = 0 ∂ µ ∂µ εα (x) = 1 (n − 2)∂α h(x) 2 where n is the dimension of the space-time. However. ωµν = −ωνµ It includes (listed are the finite transformations) translations scale transformation Lorentz transformations special conformal transformations x x x x µ µ µ µ (7. a α . consequently. The general solution for εµ (x) satisfying our constraints reads εµ (x) = a µ + εx µ + ωµν xν + (−2b · x x µ + bµ x 2 ).25) (check it) and indeed scale invariance goes together with the special conformal symmetry. bα ) satisfying (7.26) and (7. 1 − 2b · x + b2 x 2 h(x) = 0 h(x) = 2ε h(x) = 0 h(x) = 4b · x which in n dimensions form together the (n + 1 + n(n − 1)/2 + n)-parameter conformal group (see Problem 7. Scale invariance implies conformal invariance if.† h(x) is at most linear in x and.28) For the transformation of fields under special conformal transformations see Problem 7.1). εµ (x) is at most quadratic in x.236 7 Scale invariance and OPE one therefore obtains the following condition ∂ν εµ (x) + ∂µ εν (x) = h(x)gµν where h(x) = f (x) − 1. Conformal invariance implies scale invariance since the commutator of an infinitesimal conformal transformation and an infinitesimal translation contains a scale transformation. interesting for statistical mechanics and string theories.2. = e+ε x µ . and only if.27) we find explicitly the following four generators of the special conformal transformations Kα = d 3 x ( j α )0 = d3 x (x 2 α0 − 2x α xν ν0 ) (7. the opposite is not true and examples of field theories which are scale-invariant but not conformally invariant can be constructed. there exists in the considered theory a symmetric traceless conserved energy–momentum tensor so that ∂ µ Dµ = µ µ = 0. † The case n = 2. ε.1.

2 Broken scale invariance 237 7. all particles would have to be massless or their mass spectra continuous. the state exp(−iε D)| p is an eigenstate of P 2 with eigenvalue exp(2ε) p 2 .2 Broken scale invariance General discussion Scale invariance cannot be an exact symmetry of the real world.31) means that the state exp(−iε D)| p is a quantum of the same field as the state | p but with a rescaled momentum (if the vacuum was not unique then the state exp(−iε D)| p would belong to a different Hilbert space than the state | p and our final conclusion would be avoided) and therefore. P 2 ] = −2iP 2 or. One assumes that there are terms in the lagrangian (e. but there is no way of quantizing the theory so as to preserve that . but the ground state is asymmetric.12) that [D.7. Here the dynamical equations are completely symmetric. The simplest idea for symmetry breaking is that of ‘approximate’ symmetries.e. by (7. exp(iε D)P 2 exp(−iε D) = exp(2ε)P 2 (7. If we assume in addition that the vacuum is unique under scale transformations (i. exponentiating. mass terms in the case of scale invariance) which violate the symmetry. if scale invariance is not spontaneously broken) exp(−iε D)|0 = |0 then we conclude that exp(−iε D)| p = exp(−iε D)a † ( p)|0 = exp(da )a † (eε p)|0 (7. the Noether currents associated with the symmetry are conserved and the Ward identities are still valid. but that they are ‘small’. Result (7. all particles must be massless or the mass spectrum must be continuous. it follows from commutation relation (7.29) Acting on a single particle state | p with four-momentum p(P 2 | p = p 2 | p ) one gets P 2 exp(−iε D)| p = exp(2ε) p 2 exp(−iε D)| p (7.g. Indeed. We believe this idea to be realized for chiral symmetry in strong interactions.30). We will learn in the next chapters that there are several ways to account for the symmetry breaking. If it were.31) where a † ( p) is the creation operator for the considered particle with momentum p and da is its dimension. Another concept is that of spontaneous symmetry breaking.e. Whenever a classical theory possesses a symmetry.30) i. Finally there is a third way: anomalous breaking of symmetries.

In QCD the effects of the anomalous explicit breaking of scale symmetry are dominant. simultaneously. (xn )|0 where δ s are given by (7. . (xn )|0 (7.14). . (xn )|0 = i i 0|T (x1 ) .33) µ = i nd + x1 ∂ µ ∂ µ + · · · + xn µ ∂ x1 ∂ xn . δ (xi ) .s. . δ (xi ) . (xn )|0 = 0|T ∂ yµ −i i µ µ (y) (x1 ) . . . quantum field theoretical calculations must be regularized to avoid infinities and anomalies occur when there does not exist a regularization procedure which respects the classical symmetry and the corresponding Ward identity. Anomalous breaking of scale invariance Ward identities for the dilatation current (trace identities) can be obtained from the general formula (10. we say that there are anomalies in the conservation equations for the symmetry currents. .238 7 Scale invariance and OPE symmetry. a given symmetry may be broken by several of these mechanisms. an unavoidable renormalization procedure necessarily introduces a scale at which the theory is renormalized and this breaks scale invariance. (xn )|0 (7. . . . . . Of course. In the rest of this section we shall study the ‘anomalous’ response of the theory to scale transformations by comparing the naive Ward identities with the renormalization group approach.32) δ(xi − y) 0|T (x1 ) . In quantum field theory.11) ∂ 0|T Dµ (y) (x1 ) . Equivalently. (xn )|0 0|T (x1 ) . . More technically. By integrating with respect to variable y the l. even with a scale-invariant lagrangian. We then obtain d4 y 0|T µ µ (y) (x1 ) . in reality the scale invariance may be spontaneously broken but simultaneously the corresponding Goldstone bosons (dilatons) get masses due to anomalies. . .h. For a discussion of the spontaneous breaking of scale invariance we refer the reader to the literature. Scale invariance requires that there be no dimensionful parameters whereas the regularization of the quantum theory is effected by introducing a dimensionful cut-off or dimensionful coupling constants in the dimensional regularization procedure. integrating with respect to y is equivalent to taking the Fourier transform and letting k → 0. the anomalous scale invariance breaking is quite expected. and there is no remaining consequence of the original classical scale symmetry. . vanishes: it gives surface terms which can be neglected if there are no zero mass particles coupled to µ µ . .

the solution to (7. . . . however. . (xn )|0 (7. we have only n − 1 four-dimensional integrations in (7. . . . .39) . p1 . . (x2 )]|0 ∂ x1 ∂ x2 0 0 0 + x2 δ(x1 − x2 ) 0|[ (x2 ). . pn−1 ) = −iG (n) (0. . exp(t) pi ) (7.38) reads (n) (ρpi ) = ρ D (n) ( pi ) (7. . . . i. . . . Eq.14). . . cancel in pairs. . effectively. dxn exp i i = pi xi 0|T (x1 ) .33) we have brought the derivatives outside the T -product which results in further equal-time commutator terms which.e. . pn−1 ) and (n) (0.35) and analogously for G (n) (k. .36) as follows (t = ln ρ) n−1 − pi [−∂/∂t + D]G (n) (exp(t) p1 . . .34) Defining the momentum-dependent Green’s functions n (2π)4 δ 1 pi G (n) ( p1 . pn−1 )(G (n) ∼ M D ). . We observe that for d = d can the nd − 4n + 4 is just the dimension D of the Green’s function G (n) ( p1 . . pn−1 ) dx1 .37) one can derive analogous equations for the 1PI Green’s functions (n) ( p1 . In addition. (7. .37) and (7. . .35). pn−1 ) as the Fourier transform of the 0|T µ µ (y) (x1 ) . for a scale-invariant lagrangian. if we parametrize all momenta by the scaling factor pi = ρ pi so that i pi ∂/∂ pi = ˆ ρ∂/∂ρ we can rewrite (7.33) the following Ward identity ∂ + n(d − 4) + 4 G (n) ( p1 .2 Broken scale invariance 239 In the last step of (7. for µ µ = 0.38) are the Ward identities following from the Noether relation ∂µ D µ = µ µ and from the canonical field transformation rule (7. (x1 )]|0 (7. For example.36) Summation over n − 1 momenta and the additional term 4 in the square bracket are due to the fact that. . (xn )|0 we get from (7. . exp(t) pn−1 ) = −iG (n) (0. .37) From (7.7. . exp(t) p1 . . p1 . .38) where D = 4 − nd is the dimension of the 1PI Green’s function (n) . pn−1 ) [−∂/∂t + D] (n) (exp(t) pi ) = −i (n) (0. pn−1 ) ∂ pi i=1 (7. exp(t) pn−1 ) (7. . p1 . In particular. . . . . . for n = 2 we have µ 0|T (x1 )x2 ∂ µ ∂ µ (x 2 )|0 + 0|T x 1 µ (x 1 ) (x 2 )|0 ∂ x2 ∂ x1 µ ∂ µ ∂ 0 0 0 = x1 µ + x2 µ 0|T (x1 ) (x2 )|0 + x1 δ(x1 − x2 ) 0|[ (x1 ). .

38) we see that the previously mentioned anomalous breaking of canonical scale invariance occurs in quantum field theory.39) to (7. or both. we have already mentioned that there is no reason to expect these identities to be valid in quantum field theory since the scale invariance is always broken by a dimensionful cut-off.240 7 Scale invariance and OPE In this case the Green’s functions exhibit canonical scaling.e. µ) = exp(Dt) (n) ¯ ( pi . However. For easy reference we recall the solution to (6. this happens when λ+ is a simple zero of β(x) : β(λ+ ) = 0. The response of a quantum field theory to the scale transformation is given by the RGE (6. The exponential factor in (7. then ε(t) = O(1/t) and the theory is asymptotically.24). do we recover the canonical scaling. λ(t).24) with the Ward identity (7.40) where ¯ λ(t) λ dx/β(x) = t (7. Imagine now a theory with a UV fixed point λ+ . or the field transforms anomalously. As mentioned in Section 6. for t → ∞. λ. m(t). Apart from a trivial case when λ is exactly equal to λ+ and the RGE can then be solved directly and gives (7. for a non-interacting theory. Comparing (6.39) with D → D − nγ (λ+ ). scale-invariant.40) can then be written as t exp −n 0 ¯ γ (λ(t )) dt = ρ −nγ (λ+ )+ε(t) where ε(t) = −(1/t) ¯ λ(t) λ dx n[γ (x) − γ (λ+ )]/β(x) (7. β (λ+ ) < 0 (we assume that γ (x) is differentiable). Either the dilatation current has an anomalous divergence. The asymptotic behaviour of .1 only for β(λ) = γ (λ) = 0.41) and t m(t) = (1/ρ)m exp ¯ 0 ¯ γm (λ(t )) dt dx [γm (x)/β(x)] (7.41a) = (1/ρ)m exp ¯ λ(t) λ Consider first a massless m = 0 theory with a conserved Noether dilatation current ∂µ D µ = 0. m.42) If the integral defining ε(t) is convergent.24) (n) (exp(t) pi .38). µ) exp −n ¯ 0 t ¯ γ (λ(t )) dt (7. namely solution (7. i.

to be specific) m ∂ i i i = 2 (−2im 2 ) 2 2 − m2 2 ∂m p p −m p − m2 (7. We see again that only for λ = λ+ = 0 do we recover exact scale invariance. λ+ . The leading term is (n) (exp(t) pi . λ. So the integral (7. We now extend our discussion to theories with scale invariance explicitly broken by mass terms.43) The function γ (λ+ ) is called the anomalous dimension of the field. µ) = ρ D−nγ (λ+ ) exp − × (n) λ+ λ dx [γ (x) − γ (λ+ )]n/β(x) (7. with µ µ = m 2 2 .40) and (7.45) We also note an operator equation for the anomalous divergence of the dilatation current in gauge theories reflecting the anomalous scale invariance breaking µ µ µν = (β(λ)/2λ3 )G a G a µν (7. µ) ∼ ρ D−nγ (λ+ ) (n) ( pi . µ) and of the exponential factor (7. λ. More specifically one gets ( pi .46) Rigorous proof of relation (7.7.44) ( pi . is not a feature of asymptotically free (λ+ = 0) gauge theories. In this case β(x) = −bx 3 + O(x 5 ) and for most of the interesting operators γ (x) − γ (0) = 2cx 2 + O(x 4 ). Collins. The mass effects in general are summarized in the solutions (7.24) and (7. Thus we again see that for β = γm = γ = 0 (6.47) implies that the operation m∂/∂m with external momenta and the coupling constant λ fixed (we assume that there is only one mass and one dimensionless coupling constant) is equivalent to the insertion of a new vertex −im 2 2 at zero-momentum transfer. µ) is obtained by expanding λ(t) about λ+ . λ (n) (ρpi .43) arise from O(λ(t) − λ+ ) terms in the expansion of (n) ¯ (t).42) is logarithmically divergent and it causes logarithmic deviations from the asymptotic ¯ scale invariance (7. The lead¯ ing corrections to (7. Duncan & Joglekar 1977. in particular. The identity (for bosons.42). Collins & Duncan 1977. Nielsen 1977). µ)[1 + O(ρ −|β (λ+ )| )] Asymptotic scale invariance.2 Broken scale invariance (n) 241 ¯ (exp(t) pi .40) is controlled by the asymptotic behaviour of .46) is rather lengthy (Adler.37) are identical. The ρ → ∞ limit of (7. λ+ . µ) (7. λ.41) to the RGE.43) (remember that λ2 (t) ∼ 1/2bt) ¯ λ(t) exp λ 2c dx b x ∼ (2bλ2 t)−c/b (7. possibly with anomalous dimensions.

gB ). for example. y) which depends on two massive variables x. The reason why pure dimensional analysis breaks down in quantum field theory is this: our quantity A actually also depends on the dimensionless free parameter(s) of the theory.50) is. dimensional analysis tells us that. for the mass effects to be asymptotically non-leading. 7. To introduce this concept let us consider any dimensionless observable quantity A(Q) which can depend only on one dimensionful variable Q. If we assume our theory to be described by a scale-invariant lagrangian (no dimensionful parameters) then from dimensional analysis we must conclude that A(Q) = const. A(Q) = Q 2 · σtot . the predictions of the theory cannot be expressed directly in terms of gB because .50) Indeed.49) as the condition for ‘softly’ broken scale invariance (Wilson 1969a). Eq. bare coupling constant(s): A = A(Q. y and which is dimensionless and whose definition does not involve any massive constants must be a function of the ratio x/y only. So if it does not depend on y it must be a constant. As we know from our experience with quantum field theories the conclusion (7.242 7 Scale invariance and OPE ¯ both λ(t) and m(t) (we recall that t = ln ρ) ¯ m(ρ) = mρ γm (λ+ )−1 exp ¯ ρ→∞ ¯ λ λ dx [γm (x) − γm (λ+ )]/β(x) (7.49) is automatically satisfied if the theory is asymptotically free because the relevant anomalous dimension γm vanishes. for example.3 Dimensional transmutation The phenomenon of dimensional transmutation is closely related to scale invariance broken by the renormalization of a quantum field theory. in general. for example. wrong. (7. a function f (x. (7. it is essential that m(ρ) −→ 0 or ¯ ρ→∞ γm (λ+ ) < 1 m Introducing the scale dimension d d we get m m of the mass operator m=m¯ m = 2m2 = 3 + γm (λ+ ) 2 + 2γm (λ+ ) d <4 m (7. Otherwise the mass insertion would effectively correspond to a hard operator and would influence the asymptotic behaviour of the Green’s functions. Moreover.48) −→ mρ γm (λ+ )−1 × {logs of ρ} Now.

53) (7. (7. One-parameter ambiguity for α(Q 2 ) is reflected in the fact that the theory specifies the first derivative of α(Q 2 ) rather than α(Q 2 ) itself dα/dt = β(α) (7. an effective scale Q 0 appears as a constant of integration of the RGE (7. it depends on the definition of the renormalization scheme).51) where Q = exp(t)Q 0 .53) can be also rewritten in the following form /Q = exp{−π/[|b1 |α(Q)]} (7.51) we get ln(Q/Q 0 ) = F(α(Q)) − F(α(Q 0 )) where F(x) = dx/β(x) Thus. 7. defined by the condition F(α( )) = 0. is the confinement scale: α( ) = ∞. be re-established.7. α = g 2 /4π and β(α) = (b1 /π)a 2 + O(α 3 ). Eq.4 Operator product expansion (OPE) 243 the theory requires renormalization. Integrating (7. Dimensionless α(Q) and dimensionful can be traded for each other. An important illustration of these general considerations is provided by the behaviour of the running coupling constant α(Q 2 ) in QCD (we forget here about a subtlety: α(Q 2 ) is not really an observable. The physics is independent of the details of the renormalization prescription but not of the necessity of renormalization which produces an effective scale. at least up to logarithmic corrections.52) . however. In the previous section we have discussed how mass terms and renormalizable interactions break this invariance which may.51). Dimensional transmutation is a process which exploits the above to introduce dimensionful parameters into the predictions of the theory. in the limit of .4 Operator product expansion (OPE) Short distance expansion The OPE technique originates in studies of the relevance of scale invariance in quantum field theory (Wilson 1969a). If we choose Q 0 so that F(α(Q 0 )) = 0 and call it then α(Q) = F −1 [ln(Q/ )] and in the leading logarithm approximation α(Q) = −π/[b1 ln(Q/ )] (7.53a) which tells us that given the strong coupling constant α(Q) at some Q the theory predicts the confinement scale .

54) where {On (x)} is a complete set of hermitean local normal-ordered operators and {C n (x)} is a set of c-number functions. 2 . when sandwiched between physical states.4) and in any order of perturbation theory for interacting fields (Zimmermann 1970). We see that the most singular contribution to A(x)B(y) † The product also can be written in terms of. A(x)B(y) − n C n (x − y)On ((x + y)/2) = 0 AB (7. AB i. may be written as† A(x)B(y) = n C n (x − y)On ( 1 (x + y)) AB 2 (7.55) and in an exactly scale-invariant theory dn − d A − d B − (x − y) n ∂ C n (x − y)On ( 1 (x + y)) = 0 2 ∂(x − y) AB (7. Indeed. respectively. The expansion is valid in the weak sense.244 7 Scale invariance and OPE momenta going to infinity.2) so relation (7. to vanishing separation for the operators in the matrix elements. The physical importance of studying operator products at short distance will be discussed later.57) C n (x − y) = (dn − d A − d B )C n (x − y) AB ∂(x − y) AB and therefore C n is homogeneous of degree dn − d A − d B in its argument AB C n (x) ∼ x dn −d A −d B AB (7.56) where dn . Relation (7. Assuming scale invariance to be an approximate symmetry at short distance one gets the leading behaviour of the Cs for small values of their arguments. in terms of the scale dimensions of the operators. According to Wilson’s hypothesis a product of two local operators A(x) and B(y).58) Asymptotic scale invariance at short distance implies that relation (7. OPE and scale invariance arguments allow us to determine the short distance singularity structure of operator products. d A and d B are scale dimensions of the operators On . Here we shall introduce the technique of the OPE.e.54) we get D. A and B. Since the operators On can be chosen to be independent we deduce that ∂ (x − y) (7. when xµ is near to yµ .58) holds at least in the limit x → 0. In QCD asymptotic scale invariance is broken by logarithmic corrections (see Section 7.58) is modified but only by logarithmic terms. say. In position space this corresponds to short distances.e. On (y) by expanding On ( 1 (x + y)) in a Taylor series about y.54) is true in free-field theory (see Problem 7. i. commuting the dilatation generator D with (7. hopefully up to at most logarithmic corrections.

i. The Lorentz transformation properties of the Wilson coefficients C n (x) require AB them to be polynomials in the components of x times scalar functions of x 2 : C n = C n (x. We have learned in Section 7. To do this let us refer to free scalar field theory.4) T J (x)J (0) = −2 x →0 ˙ 2 2 F (x.59) where 1 1 d4 k exp(−ikx) 2 4 (2π ) k − m 2 + iε i 1 = − 2 2 (7. d B and dn are not.60) 4π x − iε x 2 →0 is the free particle propagator in position space. for several interesting operators like conserved currents or currents of softly broken symmetries.4 Operator product expansion (OPE) 245 as x → y is given by the operator On having the lowest dimension. x 2 − AB AB iεx0 ). Since F = −i 0|T (x) (0)|0 = T A(x)B(0) = A(x)B(0) (x0 ) + B(0)A(x) (−x0 ) and −x 2 1 1 1 (x0 ) + (−x0 ) = − 2 2 − iεx + iεx0 −x x − iε 0 (7.43)) can be calculated in perturbation theory. m )I 2 2 F (x. broken by operators with scale dimensions d < 4. (0)]|0 = ε(x0 )2 Im(i F ) = −(1/2π )ε(x0 )δ(x 2 ) (7. equal to the canonical dimensions of the corresponding operators. For instance. Anomalous dimensions (see (7. B(0)] we then get (1/(x ±iε) = P(1/x)∓iπ δ(x))† x2 1 1 1 − 2 = 2π iδ(x 2 )ε(x0 ) = ε(x0 )2i Im 2 − iεx0 x + iεx0 x − iε = −i 0|[ (x). For commutators [A(x). simple products.61) for simple products we have the following prescription C n (x. in general. Notice that : (x) (0): may be expanded in a Taylor series since it has no singularity. x 2 ).e.7. for the T -product of two currents J (x) = : (x) (x): in free scalar field theory we have (see Problem 7.1). the anomalous dimensions vanish (see Section 10. m 2 + 4i + 4i F (x. We have to determine prescriptions for handling light-cone AB AB singularities of the coefficients C n for different types of products: time-ordered AB products. m )I F (x.62) † Note in particular that: . m F (x. Thus for T -products we conclude that scalar factors in Cs depend on x 2 − iε. x 2 ) ≡ C n (x.2 that the scale dimensions d A . commutators etc. m 2 2 ) : (x) (0): + : )J (0) 2 (x) 2 (0): = −2 + 4i )xµ : (0)(∂ µ )(0): + · · · (7. However.

63) is (x 2 )− p with some non-integer power p. (7. To get (7. x 2 − iεx0 ) − C n (x. B(0)] = n ˜ AB C n On (0) (7. by placing the cut along the positive real axis. the difference (−x 2 + iεx0 )− p − (−x 2 − iεx0 )− p appearing in a commutator [A(x).65) one considers an integral ˜ AB d3 x C n (x0 . x 2 + iεx0 ) AB AB and letting x0 → 0 one can study the equal-time commutators.246 7 Scale invariance and OPE and by differentiating both sides with respect to x 2 1 2 + iεx −x 0 where ε(x) = 1 −1 for x > 0 for x < 0 n − 1 2 − iεx −x 0 n =− 2πi δ (n−1) (x 2 )ε(x0 ) (n − 1)! (7.66) where ρ(x) is a differentiable trial function and the Taylor expansion is justified ˜ AB for small x0 because C n vanishes for |x| ≥ |x0 |.66) and from the definitions of the δ-function and its derivatives. x)x etc.65) where n 0 FAB (x 0 ) n 1 F AB (x 0 ) = =− ˜ AB d3 x C n (x0 . This is interesting because the appearance in the commutator of the so-called Schwinger terms (proportional to derivatives of the δ-functions) is linked to the singularity structure of operator products at short distances. B(0)] vanishes for space-like separation x 2 < 0: the prescription x 2 → x 2 − iεx0 .65) follows immediately from (7. Following Wilson (1969a) we can see that ˜ AB the coefficients C n are equivalent to a sum of δ-functions n ˜ AB C n = 0 FAB (x0 )δ (3) (x) + 1 Fn (x0 ) · ∇ δ (3) (x) + · · · AB (7. Eq.64) where ˜ AB C n = C n (x.63) In a general case in which the light-cone singularity (7. x)ρ(x) = ˜ AB d3 x C n (x0 . The . x) ˜ AB d3 x C n (x0 . is consistent with causality. x)(ρ(0) + x · ∇ρ(0) + · · ·) (7. Using the expansion of the form [A(x).

Ji (0)] (but not for two J0 s): [J0 (x. (7. two currents in free scalar field theory. however. As an example let us take the product of two electromagnetic currents Jµ (x) = : ¯ (x)γµ (x): In free fermion field theory one has (see Problem 7.54) which are finite in the limit xµ → 0 may be singular in the limit x 2 → 0 with xµ finite. It is obvious. in contrast to the situation at x = 0 where the leading behaviour is given by one term. the Lorentz tensor operators (∂µ )(0).4 Operator product expansion (OPE) 247 n dependence of the functions i FAB (x0 ) on x0 is determined by dimensional analysis to be n i FAB (x 0 ) ∼ x0 −2 pn +3+i (7. Therefore we must expect an infinity of terms to contribute to the leading singularity on the light-cone. We will see that the vacuum expectation value of this term is related to the e+ e− annihilation cross-section calculated under the assumption that the leading short distance singularities are those of the free-field theory. Some terms may vanish due to rotational invariance.67) when ˜ AB C n ∼ x −2 pn The equal-time commutator is obtained in the limit x0 → 0.5 the coefficient of the δ-function is infinite and for p > 2 derivatives of the δ-functions appear in the commutator. Ji (0)] ∼ Ci j ∂ j δ (3) (x) (7.4) Jµ (x)Jv (0) ∼ x 2 gµν − 2xµ xν 1 + O(x −3 ) l π 4 (x 2 − iεx0 )4 (7.7.70) where Ci j is quadratically divergent. As mentioned before. for instance. Such terms are usually not revealed by direct calculation of the commutator of two operators based on equal-time canonical commutation relations between fields and their conjugate momenta. (∂µ ∂ν )(0) etc. that terms in (7.68) The first term gives the familiar Schwinger term in the commutator [J0 (x). 0). cannot be neglected on the light-cone and the general structure of the expansion is as .69) (7. Let us look in more detail at the expansion of a product of two Lorentz scalar operators. Thus for p > 1. Light-cone expansion The OPE can be extended to the light-cone region as well.59).

74) into (7. Nevertheless. This should not lead to any confusion. they are irreducible tensors of spin j. As an exercise let us expand ∗ − x 2 x 2 = 1 ← ← x µ1 x µ2 x µ1 + ∂ µ1 ∂ µ2 − ··· 2 2! 2 2 x ν1 → 1 x ν 1 x ν2 → → × 1+ ∂ ν1 + ∂ν ∂ν + · · · (0) 2 2! 2 2 1 2 1 x µ1 x µ2 x µn ∗ ↔ ↔ ↔ = (0)∂ µ1 ∂ µ2 · · · ∂ µn (0) ··· 2 2 n n! 2 ∗ (0) 1 − ∂ µ1 ← (7.73) The difference (dimension of an operator – its spin) is called twist. ∂ µ j When dimensions are modified by the interactions it is not obvious that they are correlated with the spin in this way. the C n (x 2 ) in the AB neighbourhood of x = 0. short distance behaviour and light-cone behaviour are connected C n (x 2 ) = (x 2 )(dn − jn −d A −d B )/2 + higher orders in x 2 AB so that the degree of singularity on the light-cone is determined by τn = dn − jn (7. .59) we get the leading terms in the light-cone expansion of the product of two scalar currents in free-field theory. . We see that the behaviour of the expansion for small x 2 is determined by scalar functions of x 2 . .72) (7.µ j to be symmetric traceless tensors with j Lorentz indices. one usually takes the free-field theory classification of operators as a guide in writing down the OPE. . . In a similar way we can obtain the light-cone expansion for two electromagnetic currents.. x µ j Oµ1 .74) Inserting (7.. The scalar field and the fermion field as well as their derivatives have twist one. since operating on a field with a derivative raises the spin and the dimension by one unit simultaneously. The leading singularity on the light-cone comes from operators with the lowest twist. . In free-field theory there exist infinite series of operators with fixed twist. Examples of twist two operators are ∗ ∂ µ1 ∂ µ2 . .µ j (0) AB n follows: A( 1 x)B(− 1 x) = 2 2 (7.248 7 Scale invariance and OPE n C n (x 2 )x µ1 .. (Note that we use the same notation for complete Wilson coefficients in (7.. ∂ µ j ↔ ↔ ↔ and ↔ ↔ ¯ γµ ∂ µ1 .71).71) n We have taken the bases Oµ1 .) As long as there are no extra divergences introduced by the summation over n.54) and for their scalar part in (7.

76) by the term −(2π)4 δ(q + pn ) 0|Jν (0)|n n|Jµ (0)|0 .79) differs from (7. (7. We focus only on the total cross section which is of the order of a few tens of nanobarns and is comparable to the cross section for the e+ e− → µ+ µ− . 7. Thus the total cross section is determined by the tensor Wµν (q) = n (2π)4 δ(q − pn ) 0|Jµ (0)|n n|Jν (0)|0 d4 x exp(iq x) 0|Jµ (x)Jν (0)|0 (7. Consider first the process of the electron-positron annihilation into hadrons.7.76) = Jµ is conserved and therefore one may write Wµν (q) = ρ(q 2 )(q 2 gµν − qµ qν ) (7.1) the amplitude is Tn = e2 v( p+ )γµ u( p− )(1/q 2 ) n|Jµ (0)|0 ¯ (7.78) To see what happens when q 2 → ∞.75) where Jµ is the hadronic electromagnetic current.5 The relevance of the light-cone 249 Fig.79) Eq.1.5 The relevance of the light-cone Electron–positron annihilation We briefly discuss now the relevance of the light-cone for high energy phenomena. In the one-photon exchange approximation (see Fig.77) It can be checked that the total annihilation cross section for unpolarized initial leptons reads σ (q 2 ) = 8π 2 α 2 ρ(q 2 )/q 2 (7. 7. 7. we observe first that Wµν (q) can be written in terms of the current commutator Wµν (q) = d4 x exp(iq x) 0|[Jµ (x). Jν (0)]|0 (7. It is a process widely investigated experimentally and interesting in many respects as a laboratory for important discoveries in particle physics.

Jν (0)]|0 (7.63) and the relation d4 x exp(ikx)δ (n) (x 2 )ε(x0 ) = 22−2n 2 2 n−1 (k 2 )ε(k0 ) π i(k ) (n − 1)! (7.81) we easily find (for a single fermion with charge ±1) Wµ µ (q) = (1/2π )q 2 q 2 →∞ (7. E. E and θ are the lepton energies and lepton scattering angle in the laboratory . Since q 2 > 0 we may choose a frame in which q has only a time component and use the fact that the commutator vanishes for x 2 < 0. In a free-field theory the leading singularity of the commutator originates from the term (7. k are lepton four-momenta.69) which is responsible for the Schwinger term (7.70).2.80) Since the dominant contribution to the integral over x0 comes from regions with the least rapid oscillations this Fourier transform is determined by the singularity of the commutator at x = 0. Using (7. 0 which is zero because q0 and pn are positive. 7.2 which also contains some definitions of kinematical variables: k.82) and consequently σ (q 2 ) → 4π α 2 /3q 2 (7. Deep inelastic hadron leptoproduction Let us now consider another famous process: the production of hadrons in lepton–hadron collisions with very large momentum transfer. Then in this frame Wµν (q) = ∞ −∞ dx0 exp[i(q 2 )1/2 x0 ] |x|<x0 d3 x 0|[Jµ (x). 7.83) This result is not surprising: a free-field theory is asymptotically scale-invariant and the cross section must behave like 1/q 2 on dimensional grounds. It is shown in Fig. P is the four-momentum of the hadronic target (usually a proton).250 7 Scale invariance and OPE Fig.

5 The relevance of the light-cone 251 frame. which are important for further discussion. The amplitude Tµν has normal threshold cuts in s = (P + q)2 and u = (P − q)2 for m 2 < s < ∞ and m 2 < u < ∞. As we shall see it is also very useful to define the so-called Bjorken variable xBJ = −q 2 /2P · q For elastic scattering s = m 2 ⇒ 2P · q + q 2 = 0 ⇒ xBJ = 1 In terms of the Bjorken variable. q3 ) . q = (q0 . for fixed q 2 . P · q) = i d4 x exp(iq x) P|T Jµ (x)Jν (0)|P (7. Let us choose a frame in which P = (m. Tµν (q 2 . 0. 0.87) then it follows from the spectral representations (see also (7. q). respectively.88) where Disc Tµν is the discontinuity of the Tµν along its branch cuts. As in case of the e+ e− annihilation into hadrons in the one-photon exchange approximation the cross section for this process is determined by the tensor Wµν (P.7. Jν (0)]|P (7. the physical region for the process is therefore 0 < xBJ < 1 (7. P · q) given by the deep inelastic cross section. If we introduce the forward Compton scattering amplitude for photons with mass q 2 .86) = = where Jµ is the hadron electromagnetic current. They correspond to a cut in xBJ for |xBJ | < 1. The passage from product to commutator can be justified in a way similar to that for the annihilation reaction. 0). We collect together some properties of the tensor Wµν (P. P · q) = 2 Disc Tµν (q 2 . P · q) = 2 Im Tµν (7.62)) that Wµν (q 2 .85) (7. 0. 0.84) To be specific we consider the process of electroproduction. Note that only for 0 < xBJ < 1 is the Wµν (q 2 . q) = n (2π)4 δ(q + P − pn ) P|Jµ (0)|n n|Jν (0)|P d4 x exp(iq x) P|Jµ (x)Jν (0)|P d4 x exp(iq x) P|[Jµ (x). Now we are ready to discuss the relevance of the light-cone in our process.

74). For the realistic case of particles with spin. twist two operators are again the leading ones.252 7 Scale invariance and OPE and introduce the so-called light-cone variables x± = x0 ± x3 . J (0)]|P = 1 2πiδ(x 2 )ε(x0 ) π2 1 xµ . x− → 0 for x+ finite corresponds 2 to x → 0 (but not to xµ → 0) and Wµν is determined by the singularity of the current commutator on the light-cone. x⊥ = (x1 . ν → ∞ and xBJ is fixed (xBJ = −q 2 /2ν). J (0)]|P (7.59)) the leading singularities of the operator products on the light-cone are given by the twist two operators P|[J (x). . .92) where we have used (7. We see that in the limit q+ → ∞ q− fixed (7. . q) = × dx− exp(iq+ x− ) 2 x⊥ ≤x+ x− dx+ exp(iq− x+ ) (7.90) the behaviour of Wµν is determined by the behaviour of the integrand for x− → 0.90) is due to Bjorken (1969) and it has turned out to be one of the most creative ideas in modern elementary particle physics. Jν (0)]|P where the limit on the x⊥ -integration comes from the vanishing of the commutator 2 for x 2 = x+ x− − x⊥ < 0. ∂µn (0)|P + less singular terms . x+ finite. In the limit (7. Apart from the c-number contribution which is irrelevant for the considered matrix element (first term in (7. Spin complications are not essential for general orientation.90) q 2 → −∞. x2 ) ν = q0 m q± = 1 (q0 ± q3 ) = (ν/2m)[1 ± (1 − m 2 q 2 /ν 2 )1/2 ]. . The general form of the matrix n × P| (0)∂µ1 . xµn n! 1 (7. The idea of studying the limit (7.e. 2 Then Wµν (P.89) d2 x⊥ P|[Jµ (x). Since the integrand is 2 non-vanishing only for x⊥ < x+ x− (causality). in the region with the least oscillations. For simplicity let us study the scalar current analogue of the electroproduction tensor W (q 2 . i.63) and (7. P · q) = d4 x exp(iq x) P|[J (x).91) in the Bjorken limit.

P · q).62): 0|[ (x). P · q) = 4 dξ δ((ξ P + q)2 )ε(q0 + ξ P0 ) f˜(ξ ) d4 x exp(ikx)δ(x 2 )ε(x0 ) = (2π )2 ε(k0 )δ(k 2 ) (7.92) is P|Oµ1 .97) In the free-field theory with twist two operators giving the leading singularity on † Identity (7.95) dξ exp(iξ x · P) f˜(ξ ) The roots of the argument of the δ-function are ξ± = − ν ν q 2m2 ± 2 1− 2 m2 m ν 1/2 −→ BJ 2ν −q 2 . is non-vanishing for |−q 2 /2ν| < 1. Thus only the ξ+ is relevant and in terms of the Bjorken variable our result reads νW (q 2 . Pµn + δµ1 µ2 Pµ3 . W (q 2 . J (0)]|P = δ(x 2 )ε(x0 ) f (x P) + less singular terms (7. As has already been explained.. So we have 2i P|[J (x).96) where ‘BJ’ means Bjorken limit. .95) follows from the spectral representation for the free-field commutator (Bjorken & Drell 1965) and the relation (7.7. (0)]|0 = − = 1 d4 k exp(ikx)δ(k 2 − m 2 )ε(k0 ) (2π )3 i δ(x 2 )ε(x0 ) 2π . . P · q) = 2 f˜(xBJ ) (7.− 2 2ν m (7.92) they give additional factors x 2 and therefore are less singular. Pµn + · · · (7. .93) Terms having at least one Kronecker delta may be ignored because when contracted with the xµ s in (7. being the imaginary part of the amplitude for forward off-mass-shell Compton scattering.5 The relevance of the light-cone 253 element in (7. .94) π where 1 (x · P)n An f (x P) = n n! Defining the Fourier transform f (x P) = (1/2π ) and using the identity† i we get W (q 2 ..µn |P = An Pµ1 .

Instead of (7.91) is the discontinuity of T . x µn Cn (x 2 ) (7. .92) let us consider T J ( 1 x)J (− 1 x) ≈ 2 2 x 2 →0 µ k Cn (x 2 )xµ1 .92). The standard procedure is as follows. Remembering that it is an analytic function of xBJ in the complex xBJ -plane with a branch cut along the real axis for . An important final observation is that only even spin operators are allowed in the expansion (7. Consider first the forward scattering amplitude for scalar currents T (q 2 . Using (7. Its validity for renormalized operators has been proved in perturbation theory and therefore it can be used to obtain further information about the scaling limit of the structure functions. xµn Ok 1 .88). P · q) if C we integrate T (q 2 . An important step in this direction is to establish some relation between the Wilson coefficients and measurable quantities which is not limited to the canonical scaling case. Therefore ˜k ˜k in the case of canonical scaling Cn (q 2 ) = Cn are constants. beyond that contained in the hypothesis of canonical dimensions. P · q) in the Bjorken limit can be written as T (q 2 . P · q) = i d4 x exp(iq x) P|T J ( 1 x)J (− 1 x)|P 2 2 (7. .k (7.254 7 Scale invariance and OPE the light-cone. see (7. for fixed q 2 . . P · q) = 1 −q 2 −n ˜ k xBJ Cn (q 2 )Ak n n. xBJ ) over xBJ . Wilson coefficients and moments of the structure function Use of the OPE is by no means limited to the free-field theory case.µn n. q µn ˜ k 2 Cn (q ) = i (−q 2 )n+1 k d4 x exp(iq x)x µ1 .. Indeed W is real (it is a cross section) so f˜(ξ ) must be real and therefore f (x P) must be even in x. . The coefficients k ˜ n (q 2 ) can be related to the moments of the structure function W (q 2 .101) ˜k Note that Cn (q 2 ) are defined so that they are dimensionless for twist two operators in the expansion of the product of two currents with dimension d J = 3. . .98) The structure function (7. we obtain the famous Bjorken scaling for the structure function W (q 2 .k (7.93) and defining 2n q µ1 .. P · q).100) the leading contribution to T (q 2 .99) where the sum is taken over all twist two operators Ok and all even spins n.

104) we have written down explicitly the dependence on the renormalization point and the coupling constant.3. This may not be the case for low moments: for fixed q 2 the limit x → 0 corresponds to ν → ∞. g)Ak (µ2 . where ν = P · q ≈ s and s is the cms energy squared in the forward Compton amplitude. x) = −W (q 2 . x) = C 1 −q 2 ˜k Cm (q 2 )Ak m k (7. x) ∼ x −1 . g) (7.104) vanishes). −x) (7.102) and on the other hand 1 2πi Finally we get Mm (q 2 ) = (1/2π) 1 −1 dx x m−1 T (q 2 . We also note that (7. Using the crossing properties of the structure function under q → −q or x → −x W (q 2 . P · q) we have (x ≡ xBJ ) 2 1 2π i dx x m−1 T (q 2 . 7. 0 < x < 1 (for odd m the integral (7.105) we can rewrite (7. x) = k ˜k Cm (q 2 .7. µ2 . Since the discontinuity of the T is just 1 W (q 2 . . 7. |xBJ | < 1 we integrate over the contour shown in Fig. so that W (q 2 . x) = C 1 4π 1 −1 dx x m−1 W (q 2 .103) dx x m νW (q 2 .5 The relevance of the light-cone 255 Fig.3. At high energy the imaginary part of this amplitude is expected to behave as s α . Thus we get a relationship between moments of the deep inelastic structure function and the coefficient functions.104) for even m in terms of the integral over the physical region in the electroproduction process.104) m In (7.104) is meaningful for values of m such that the integral converges. x) (7. α ≈ 1.

7. . . require renormalization. P → ρ P relates Mn s for different q 2 s and m 2 s. (xn )O(x)|0 O or in momentum space ˜ (n) (2π)4 δ( p1 + · · · + pn + p)G O ( p1 .256 7 Scale invariance and OPE An important virtue of the OPE is that it allows us to factorize out the q 2 dependence in quantities like Mm which depend on q 2 and P 2 = m 2 . ] = D exp iS + i d4 x (J + O) (7. pn . .109) generates these new Green’s functions. in general.108) The generating functional formalism and perturbation theory rules can be extended to such cases if we introduce sources (x) coupled to operators O(x). . The functional W [J. . . The counterterms present in the lagrangian for the renormalization of the Green’s functions involving only elementary fields are not sufficient for eliminating divergences in the Green’s . . for example. Connected Green’s functions are obtained from the functional Z and the Legendre transformation performed on sources J (only) generates 1PI Green’s functions with O insertions. As usual one has to take derivatives with respect to and then set = 0. xn . G (n) (x1 . . This dependence can then be studied by means of the RGE. Feynman rules for vertices involving elementary fields and the composite operators O(x) can be derived in exactly the same way as in Chapter 2 or Chapter 3. (7. . As we know. exact Bjorken scaling gives ˜k Cn (q 2 ) = const. x) O (7.107) = d x exp(i px) i=1 4 d4 xi exp(i pi xi )G (n) (x1 .6 Renormalization group and OPE Renormalization of local composite operators We are used by now to working with Green’s functions involving not only elementary fields but also composite operators. .106) Interactions can modify this behaviour. p) n (7. . P 2 = m 2 ) because the scale transformation q → ρq. The new Green’s functions calculated perturbatively. In the next section we shall study the q 2 dependence of the Wilson coefficients using RGE. Similar analysis is not possible directly for Mn (q 2 . x) = 0|T (x1 ) . . . . xn .

7. 0) = 1 (7. .110) then the 1PI npoint Green’s functions with a single O insertion satisfy (n) O. . The divergent Green’s functions with this insertion are (2) and (2) with DO = 0. For instance. p. where the vertex ⊗ is in both cases just one and for simplicity we can take the momenta of the external legs to be equal. For instance. ( 2 B )B = Z 2( 2 B )R =Z 2 Z 3( 2 R )R (7. we may impose (2) 2 (0. p. Additional operator renormalization constants are needed OB (x) = Z O OR (x) (7.R ( p1 . .B ( p1 . 0) = 1 and (2) ¯ (0. new renormalization constants require new renormalization conditions. . λB .2) DO of O differs from D of (n) and reads can DO = D + (dO − 4) (7. it may not eliminate divergences from some Green’s functions involving more then one insertion of O. It is an easy exercise to calculate 2 ¯ Z O in these simple cases in the lowest order of perturbation theory. (7. pn . The Z O can be calculated in perturbation theory by calculating the divergent Green’s functions with operator insertions. Of course.115) Simple multiplicative renormalization. for example. Relations (7. µ.110) If we want to express OR in terms of the renormalized fields the wave-function renormalization constants will explicitly appear. is actually not always sufficient.4. .109). We have taken O(x) to be of the ith order in the elementary fields (x). The relevant Feynman diagrams are shown in Fig. The last term in (7.1).110) and (7.6 Renormalization group and OPE 257 functions with composite operator insertions. λ. −1 pn . A well-known example is the vacuum . . .113) can where dO is the canonical dimension of O. The superficial degree of divergence (n) (see Section 4.112) is the new counterterm.112) in the generating functional. m) = Z O Z 3 n/2 (n) O.111) correspond to a replacement (x)OB ( i B (x)) = OR ( i R) + (Z O Z 3 − 1)OR ( i/2 i R) (7. .7. As an example let us consider λ 4 theory and discuss a single insertion of O(x) = 1 2 (x) (Itzykson & Zuber 1980) 2 or an insertion of O(x) = ¯ (x) (x) in QCD.110). If an operator O is multiplicatively renormalized as in (7. (7.111) Only in exceptional cases like conserved currents or currents of softly broken symmetries does Z O = 1 (see Section 10. m B ) (7.114) in agreement with the lowest order.

24). − where γO = − µ ij (n) On .4. λ. p. Another complication arises for the following reason. µ. 7. . (7. .117) we can derive for Green’s functions with an operator insertion the RGE analogous to (6.110) takes a matrix form OB = (Z O )nk Ok n and correspondingly pn . .R ( p1 . . m) = (Z O )nk Z 3 n/2 (n) Ok . all possible counterterms with dimensions ≤d allowed by the symmetries of the bare lagrangian. m) = 0 (7. expectation value of two electromagnetic currents Jµ (x) which requires a subtraction although Z J = 1 (see Section 10. λ(µ).2 concerning the necessary counterterms in the standard renormalization programme we can convince ourselves that an operator in the lagrangian with dimension d requires.116) ∂ ∂ ∂ ij + β(λ) + γm − nγ + D δi j + γO ∂t ∂λ ∂m (n) t O j (e pi . .B ( p1 .258 7 Scale invariance and OPE Fig. From (7. p. . Thus. in general.117) Note that the matrix Z i j is not symmetric: Z i j = 0 if dim Oi < dim O j . (7. Using the arguments of Section 4. If we limit our discussion to Green’s functions involving only one insertion of an operator O the same is true for the renormalization of such composite operators. . m B ) (7.9) and (6. −1 pn .1). Therefore if there exist several operators with the same quantum numbers as the operator O and with their canonical dimensions lower or equal to the dimension of the operator O. they will mix with O under the renormalization. λB . for instance.118) d −1 (Z O )ik (Z O )k j dµ . . µ.

49). g(µ). µ)Oi (0) i (7.9) and (6. and of the composite operators. . we get. µ) = i ˜ CiAB (q. g(µ).7. We can interpret the operators On as a perturbation of the lagrangian (4. Both can be formulated in the same language.9) and (6. p1 . γm interpreted now as the anomalous dimension of the operator 2 (7. g(µ). According to (2. .119) and therefore for any Green’s function with insertion of the product AB we have (n) AB (q. For instance. the n-particle Green’s function is. Let us recall that in general A( 1 x)B(− 1 x) ≈ 2 2 x 2 →0 CiAB (x. with. defined in Chapter 4 as renormalization of the lagrangian parameters m B and λB . .47)) and we arrive to the same RG equations (6. could as well be defined as the renormalization constants for the operators 2 and 4 . . pn . µ) (n) Oi (0. putting Cn = 1.89). (4. (7. in fact. In the opposite direction.24). . B d4 x Cn OB n RGE for Wilson coefficients We now turn back to the OPE and study the q 2 dependence of the Wilson coefficient functions by means of the RGE. a superposition of Green’s functions with the vertex and mass operator insertions (we can treat the mass as one of the parameters in perturbation theory).1): LB 4 → LB 4 + n n B and we can renormalize the new coupling constants: Cn = Z O Cn . The derivatives m 2 ∂/∂m 2 and λ∂/∂λ count then the number of such insertions (see also (7.1). p1 . the renormalization constants Z 0 and Z 1 .24) and.120) By applying the differential operator D=− ∂ ∂ + β(g) ∂t ∂g . pn . We then derive (n) the RGE for the On by a complete analogy with the derivation of (6. for example. .118). . for example. . At this point it is worthwhile to observe that in perturbation theory there is no fundamental difference between the renormalization of the original lagrangian. the renormalization of local composite operators discussed here can be phrased in the language of Chapter 4. µ) (7.6 Renormalization group and OPE 259 is the anomalous dimension matrix for the operators Ok and D is the dimension of the Green’s function (equal to the canonical dimension of all operators involved) and the mass dependence has been neglected.

124) 2 In the last step we have used (7.118) we get − ∂ ∂ ij ˜ + β(g) + D A + D B − DO + γ A + γ B δi j − γO CiAB (et q. µ) ≈ (q 2 /µ2 )−γi (g+ ) Ci (µ. i. 0.120) with all momenta rescaled by exp(t) and using (7. We have assumed here that operators A and B do not mix with other operators when undergoing renormalization.123) Thus for the moments of the structure function we have (from (7. g(µ). β(x) = (2/16π 2 )b1 x 3 + · · · and γi (x) = (4/16π 2 )γi0 x 2 + · · ·. µ) (7. if there is only one twist two operator in expansion (7. D A + D B − DO = 0. This is in particular trivially true for the electromagnetic currents. g(t). For the no-mixing case ij γO = δ i j γi we then have ˜ Ci (et q. µ)[1 + Ci1 g 2 (t)/4π + · · ·] ¯ ¯ in the leading order in g 2 we get 0 ˜ ˜ ¯ Ci (q. (7. In addition put γ A = γ B = 0 as for conserved currents.26).104) when k = 1.99)) Mi (q 2 ) g 2 (q 2 ) ¯ = 2 2 2 Mi (q0 ) g (q0 ) ¯ −γi0 /b1 ln(q 2 / = 2 ln(q0 / ) 2) 2 +γi0 /b1 (7. g(µ). g(µ).53).260 7 Scale invariance and OPE to (7.122) ˜ ¯ × Ci (q. µ) = Ci (µ. µ).e. µ) ∂t ∂g =0 (7. D B and DO are canonical dimensions of A. µ)[g 2 (t)/g 2 (µ2 )]−γi /b1 (7. Let us write it down ˜ for some specific cases. t = 1 ln(q 2 /µ2 ) and expanding 2 ˜ ˜ Ci (µ. B and O respectively and ij γ A . g(t). In asymptotically free theories the UV fixed point g+ = 0. γ B and γO are their anomalous dimensions. µ) ˜ ¯ To proceed further one has to calculate Ci (q. Firstly take Ci dimensionless. g(t). Eq. µ) = exp[−γi (g+ )t] exp − g g(t) ˜ dx [γi (x) − γi (g+ )]/β(x) (7. γi (g) and β(g) in perturbation theory. Putting q 2 = µ2 . The solution to the RGE for the Wilson coefficient is analogous to (6.e.122) gives a power-like violation of scaling ˜ ˜ Ci (q.125) . i.124) is valid for both q0 and q 2 large as 2 compared to .121) where D A . g(µ). µ) = Ci (µ. For a non-zero UV fixed point g+ = 0 (7. 0. g+ .

Cn  γ = [γi j ] ˆ and the Tg ordering.126)  C1 ˆ  .127) OPE beyond perturbation theory Strictly speaking. that the coefficient functions Cn are calculable in QCD perturbation theory. which is discussed in this section. Nevertheless we assume the validity AB of the OPE and. the matrix elements of the operators taken between hadronic states contain non-perturbative effects responsible for the binding of quarks and gluons into hadrons. Vainshtein & Zakharov (1979) and extensively discussed in the recent years. g(µ). This approach is implicitly based on the parton model assumption discussed in Section 8. µ) ¯ (7. . g(q 2 ).6 Renormalization group and OPE 261 In the case of operator mixing the solution to the RGE can be written in the matrix form (take g+ = 0) ˆ C(q. One can then expect that to a good approximation the coefficient functions are calculable in perturbation theory while the non-perturbative effects are accounted for by the matrix elements of the operators On . when the strong interactions are described by QCD. This is because in the realistic case. moreover.  C =  . pioneered by Shifman. µ) = Tg exp where g(µ2 ) g(q 2 ) ¯ dx γ (x)/β(x) ˆ ˆ C(µ. The same idea underlies the use of the OPE to account for the non-perturbative structure of the QCD vacuum. that the contribution from short distances can be separated from that of large distances. is defined as follows ˆ ˆ g Tg exp dx g ¯ γ (x) ˆ =1+ β(x) 1 + 2! g dx g ¯ g γ (x) ˆ β(x) x dx g ¯ g ¯ dy ˆ γ (x) γ (y) ˆ + ··· β(x) β(y) (7. It is also in agreement with the possibility.7. very suggestive as the physical sense of the OPE formalism. the use of the OPE to study the moments of the structure function in deep inelastic lepton–hadron scattering. . Non-perturbative effects manifest themselves through non-trivial vacuum expectation values of various operators which vanish in perturbation theory 0| :On : |0 = 0 if On = I . γ (x2 )] = 0. necessary since [γ (x1 ). goes beyond perturbation theory.4.

) + i Dϕ D exp i ik k! dx J (x)ϕ(x) dx Lϕ (x) + J (x)ϕ(x) dx L (x) .129) dy1 . (1984) and references therein).e.128) We are interested in calculating amplitudes involving only the light particles ϕ with external momenta much smaller than M. we assume for simplicity that the interaction lagrangian Lϕ is ˜ linear in . Important for this programme is the theorem that the coefficient functions in the OPE manifestly reflect the symmetries of the lagrangian whether or not they are symmetries of the vacuum (Bernard et al. AB The coefficient functions Cn (q) can be found by calculating the appropriate Feynman diagrams (Problem 7.84)) W [J ] = N =N × ∞ k=0 Dϕ D exp iS(ϕ.7 OPE and effective field theories OPE is a convenient tool for discussing effective field theories. . Let us consider a renormalizable action containing some light field(s) ϕ with mass m and some heavy field(s) with mass M (such that m M) S(ϕ.262 7 Scale invariance and OPE The best-known examples are quark and gluon condensates 0| : ¯ : |0 and µν 0| :G a G a : |0 µν the first one is responsible for the spontaneous breaking of chiral symmetry in QCD (Chapter 9). for instance. This use of the OPE has been successful phenomenologically but its theoretical status and possible limitations are still subject to some discussion (see. ) = dx Lϕ (x) + L (x) + Lϕ (x) (7.5). Lϕ = L(ϕ) . 1975). dyk Lϕ (y1 ) . 7. (7. Lϕ (yk ) exp i From now on. . . i. Novikov et al. (The derivatives acting on can be removed . . The Green’s functions for processes involving only light particles are generated by the functional (recall (2.

L(yk )G conn (y1 . . dyk L(y1 ) .7. yk ) We have thus arrived at the effective theory described by a non-local action. . . (yk ) exp i dx L (x) (7.105)–(2. . Now. . .7 OPEs and effective field theories 263 by integration by parts.130a) ˜ ˜ dy1 . we can write W [J ] = N × × =N × ∞ k=0 ∞ k=0 Dϕ exp i ik k! dx Lϕ (x) + J (x)ϕ(x) ˜ ˜ dy1 . Then there exists a set of Wilson coefficients {C k (x)} AB such that† Dϕ A(x)B(y) exp iS[ϕ] + i = k dx J (x)ϕ(x) x+y 2 exp iS(ϕ) + i dx J (x)ϕ(x) (7.130) D Dϕ exp i ik k! dx Lϕ (x) + J (x)ϕ(x) ˜ ˜ dy1 . which is due to the features described below (2. The important point is that the effective region of integration over y involves distances much shorter than the region of integration over x. . L(yk )G (y1 . . the series in the r. . it is perfectly equivalent to the the underlying theory (7. As far as the processes involving only external ϕ particles are concerned.107)). . The OPE hypothesis (7. Moreover. yk ) Comparison with (2. . .131) and (2. if A or B involve derivatives. . .128). . of (7. dyk L(y1 ) . (2.h. Let A(x) and B(x) be some monomials in the fields ϕ(x) and their derivatives.56). . . .131) C k (x − y) AB Dϕ Ok † Strictly speaking. All the local field products here are tacitly assumed to involve counterterms which in the operator language would correspond to their normal ordering (see (2.130). it corresponds only to OPE for T -products. .131) is not in a one-to-one correspondence with that of (7. L(yk ) (y1 ) .s.54) can be formulated in the path integral language as follows. dyk L(y1 ) . . belonging to a complete basis of such monomials {Ok (x)}. .54).137) allows us to ‘exponentiate’ with simultaneous replacement of ordinary Green’s functions by the connected ones W [J ] = N + ∞ k=0 Dϕ exp i ik k! dx Lϕ (x) + J (x)ϕ(x) (7.) In such a case. . . the OPE has to be performed.

one should stress the important difference with respect to the original (true) Wilson coefficients defined by (7. Therefore. .136) The factors M n(l) have been extracted above just to keep Cl dimensionless. . one can show that if A(x)B(y) corresponds k to a series C k Ok and C(x)D(y) corresponds to a series CC D Ok . yk )G conn (y1 . the coefficients Cl are constants. . In a similar way. In this way we obtain W [J ] = N with Seff given by Seff = and Cl = M n(l) k Dϕ exp iSeff + i dx [ Lϕ (x) + J (x)ϕ(x)] (7. . .264 7 Scale invariance and OPE One can prove the validity of a similar expansion for multiple products like A(x)B(y)C(z) by appropriate differentiations of (7. . then AB A(x)B(y)C(z)D(w) corresponds to the product C k (x − y)Ok AB k x+y 2 n CC D (z − w)On n z+w 2 (7.130a) by sums of the form ˜ ˜ L(y1 ) . The coefficients Cl are customarily also called Wilson coefficients. . yk )δ(ymean ) (7. . . . yk )Ol (ymean ) (7. . which follows from translational invariance. In the next step.136) the latter are integrated with the Green’s functions for heavy particles. .132) Finally. one arrives at the conclusion that it is possible to replace the products ˜ ˜ L(y1 ) . i. dyk Cl(k) (y1 . M n(l) n(l) ≥ 1.135) ik k! dy1 . . This latter conclusion would be rather obvious in the operator language. performing only k − 1 integrations for each k-point function. L(yk ) −→ l Cl(k) (y1 . However. we integrate the coefficients Cl(k) with the connected Green’s functions keeping all the ymean s fixed.133) in spite of the fact that they stand under the exponent. L(yn ) in (7.131) to each term on the r. . The results of this integration are independent of ymean . .133): in (7. logarithms and positive powers of large mass M) can be absorbed into Lϕ by . (7. .s. and applying (7.e. It remains to show that Lϕ (which may contain M independent terms.h. Lϕ stands for terms which would have n(l) ≤ 0.131) with respect to J (z) and z (according to the structure of C(z)).134) dx Lϕ (x) + l 1 Cl Ol (x) . . . which can be interpreted as additional independent coupling constants for the effective interaction vertices Ol .

By assumption (i) Lϕ contains only terms allowed by symmetries of S (we assume the absence of anomalies).135) is. in principle. yk ) at points where all the yi almost coincide. . the renormalization programme can be effectively carried out.7. the behaviour of the Cl s at M → ∞ is determined by the leading behaviour of (Euclidean versions of) Cl(k) (y1 . but these divergences are expected to cancel among different integrals when counterterms for the full theory defined by the lagrangian (7.7 OPEs and effective field theories 265 renormalization of its fields and parameters.131).‡ This is equivalent to saying that they behave at most logarithmically. . and that Cl s are convergent (or behaving at most logarithmically) in the limit M → ∞. the leading behaviour of the Cl s at M → ∞ has the same property. and the symmetries defining† Lϕ are also the symmetries of S including the possible gauge fixing terms for . The action is built from all possible terms of dimension ≤ 4 that are invariant under the symmetries. they are UV divergent. This guarantees that Lϕ can be absorbed into Lϕ by appropriate redefinition of fields and parameters in Lϕ . Hence. § This can be easily shown on dimensional grounds. The action Seff in (7. (ii) asymptotically scale-invariant. In fact.136) are not divergent in spite of their proper behaviour at M → ∞. Of course. . Such a situation occurs when Higgs–fermion couplings in the Standard Model are considered in the process of heavy fermion decoupling. Moreover. The parameters (couplings) Cl undergo renormalization and also evolve according to their RGEs which do not depend on the anomalous dimensions of the original † A renormalizable theory can be defined by specifying its field content and its symmetries.135) to make the Cl s dimensionless. We assume that the theory described by Lϕ is (i) renormalizable. up to logarithmic corrections. An important point omitted in the above discussion is checking whether the integrals present in (7. only a finite number of Ol s needs to be considered in practice. We are going to bypass this point here.128). non-renormalizable because it contains terms of arbitrary dimension. not every set of fields and symmetries can give us a renormalizable theory. This requires the introduction of some restrictions on the action S in (7. Assumption (ii) tells us that the latter behaviour is independent of dimensionful parameters present in Lϕ . Checking this would require expressing Cl(k) in terms of the ϕ-particle Green’s functions (and their derivatives) with the help of (7. . However. ‡ Provided the interaction L ϕ does not contain coupling constants growing with M. Determining the behaviour of the Cl s in the limit M → ∞ is based on the fact that the (Euclidean) connected Green’s functions for the particles exponentially vanish when splitting among their arguments becomes much larger than 1/M. as we will see below.128) are properly included. but not for ϕ (see Weinberg (1980)). since the contributions to the amplitudes from higher-dimensional Ol s are suppressed§ by powers of (external momenta)/M. up to logarithmic corrections. because powers of 1/M have been already extracted in (7. Thus. .

. Only in this sense are they dependent on m/M and other parameters present in (7. in principle. F fermionic fields and no more than D derivatives. The renormalization scale µ has then to be set equal to µ ∼ M. if the full theory is known.. Ol j -vertices is equal to D = 4 − B − 3 F + n(l1 ) + · · · + n(l j ). The sum over l in (7. . They can be considered to be new free parameters to be determined from experiment or.128) by (7. be carried on in the theory described by (7.137) l ˜ where the Z l s are products of renormalization constants of the fields present in Ol s. one recovers the renormalization constants in Seff determined by the requirement of finiteness of amplitudes generated by Seff .135) combined with the RGE evolution allows us to extract large logarithms from all orders of the perturbation series. it may require counterterms which 2 are polynomials in external momenta only of order ≤ D. There are four main steps of this procedure: First.2.‡ The third step is deriving and solving the RGEs for the effective action parameters.l j l is proportional to M − p . but not on the coefficients Cl .133). After removing subdivergences.135) is the fact that the perturbative expansion of low energy amplitudes generated by (7. Therefore.135).128) and (7. they can be recovered by requiring the equality of amplitudes generated by (7.128).. which results in expressing these parameters renormalized at the scale µF ∼ † According to (4. the superficial degree of divergence of a diagram with Ol1 . . one finds the coefficients Cl (and. . Replacing (7. also the renormalized parameters in Lϕ ) by requiring equality of amplitudes generated by (7.42). It vanishes for p < 0 which can be justified† using the arguments of Section 4. only a finite number of renormalization constants has to be found when one works up to a given order in 1/M.135).128). Next. It is easy to convince oneself that Z l1 . and Z l1 .. Such Ol s have n(l) ≤ D + B +(3/2)F −4. when expressed in terms of renormalized quantities. The reason for introducing the effective action (7.135).l j l are additional renormalization constants necessary to cancel divergences in diagrams with j Ol -vertices. All the Z s in the above expression depend on m/M and the couplings present in Lϕ .266 7 Scale invariance and OPE operators in the OPE (7.. in order to avoid large logarithms.135). takes the form 1 M n(l) Z l1 l Cl1 + l1 l1 l2 ˜ Z l1 l2 l Cl1 Cl2 + · · · Z l Ol (7. Such counterterms are generated by Ol s involving B bosonic fields. where p = n(l1 ) + · · · + n(l j ) − n(l). This is why the renormalization programme can. .128) is in practice an expansion in powers of (coupling constants)· ln[(external momenta)/M] which may be too large to be expansion parameters. ‡ We restrict ourselves to the µ-dependent renormalization schemes such as MS or the µ-subtraction scheme. in practice.128) and (7. The renormalization constants in Lϕ may be different than in the underlying theory (7. and therefore n(l1 ) + · · · + n(l j ) − n(l) ≥ 0.

135). The RGEs are then derived. Even at the leading order in 1/M.137) needs to be considered and. for example. let us restrict our attention to the leading order in 1/M. Then. Here we consider only those cases in which it is possible to avoid the appearance of dimensionful parameters.128) are expected to appear in the expansion (7. dµ (7. In principle. multiplying them by m.2 are used for determining the µ-derivatives of renormalization constants.7 OPEs and effective field theories 267 (external momenta) in terms of their counterparts renormalized at µI ∼ M. Therefore. As long as one works in the framework of dimensional regularization. interpreting Cl s as new coupling constants. One should also remember that even the coefficients Cl vanishing at µI can acquire non-zero values at µF .2 have to be modified. from the equation d/dµ Seff = 0. the meaning of this statement will be made more precise. Buras & Harlander 1990). As already mentioned. we get µ d Cl (µ) = dµ γkl (g(µ))Ck (µ).137). Sometimes this requires changing the normalization of some of the Ol s. by † It may not always be possible to define a mass-independent renormalization scheme in a theory described by Seff . In the following. it is convenient to solve the RGEs using the variable v = g 2 ln(µ2 /µ2 ) instead of µ itself. Then. their number is usually very large. due to the non-diagonal character of the mixing in (7. The variable v is treated I as independent of g(µI ). and the methods† of Section 6. at the beginning.138) where the anomalous dimension matrix γ is determined by ˆ γkl (g) = − j µ d Z k j (Z −1 ) jl . Let us assume that Lϕ contains only one (dimensionless) coupling constant g. This is done either perturbatively or with use of some non-perturbative methods (Buchalla. Then the methods of Section 6. and all the RGEs are solved perturbatively in g(µI ). up to a given order in g(µI ). . k (7.139) The last step is calculating the desired amplitudes with help of Seff renormalized at µF ∼ (external momenta). The coupling constant g(µ) is therefore determined. for example. all possible terms allowed by symmetries of (7.1) in order to keep them dimensionless. the outlined procedure allows us to extract large logarithms from all orders of the perturbation series.7. elements of all the four steps are performed simultaneously in order to select the set of operators Ol that may contribute to the desired amplitudes at the desired order (even by their influence on the RGE evolution of some directly contributing Ol ). the renormalized Cl s should be rescaled by appropriate powers of µ (as has been done with λR in Section 6. The latter have been already found by ‘matching’ in the previous step and serve as initial conditions for the RGEs. only the first of the sums in the bracket of (7. For simplicity.

There. (7. knowledge of γ and β up to (n + 1)th ˆ 2 order in g is necessary for determining the Cl s up to nth order in g 2 (µI ) and exactly in v. There are two main domains of applications of the described technique. even the coupling constant splitting after the spontaneous symmetry breakdown in GUTs can be most easily understood in terms of the effective theory arising after the heavy gauge boson decoupling (Weinberg 1980). for example.6 and 12. Then the decoupling is done successively and the matching between several ‘effective theories’ is performed (see. Such a result could never be obtained with help of perturbative calculus based on the action (7.128) itself. • the matching at µI (second step) is performed up to nth order in g 2 (µI ). The other domain covers phenomena taking place much above the electroweak scale (see. for example. There.268 7 Scale invariance and OPE ∞ n=0 the series g(µ) = g(µI ) g 2n (µI ) f n (v). . the role of heavy particles is played by the W± and Z0 bosons. for example.138).140) is then inserted into the RGE for the coefficients Cl .140) where f n (v) are certain functions found with help of the equation (see (6. In some processes. (7. (More details of the effective theory approach are presented in Sections 12. • the final amplitude (last step) is calculated up to nth order in g 2 (µF ) and expanded with help of (7. In fact. Thus. It is important to notice that after decoupling the resulting effective theory usually has less symmetries than the original one (as in the examples mentioned above).7). The price we have to pay is the necessity of truncating the series in 1/M.6)) µ d g(µ) = β(g). the top quark and the Higgs bosons. Gilman & Wise (1979)). The coupling which becomes too large after multiplication by the logarithm is the QCD gauge coupling. This is the precise meaning of the statement that all large logarithms are extracted from the perturbation series.141) The coupling g(µ) from (7. the coupling constants are smaller but the evolution is usually longer.140). Springer & Wise (1990)). if • the renormalization constants (first step) are found up to (n + 1)th order in g 2 . Since µd/dµ = −2g 2 (µI )d/dv. Wilczek & Zee (1979)). the b and c quarks are also treated as heavy. we obtain the final answer up to nth order in g 2 (µI ) and exactly in vF = 2g 2 ln(µI /µF ). First of them covers low energy electroweak processes involving hadrons (see. but this is often a much better approximation than truncating the series in g(µI ) which has to be done anyway. Grinstein. dµ (7.

Obtain the . 7. Pν ] = [K µ .14) on the fields as in the scale-invariant theory (use the canonical formula (7. f (x)] = ∂µ f i[M µν . D] = iPµ [K µ . f (x)] = (x 2 gµν − 2xν xµ )∂ ν f where f (x) is a dimensionless scalar function and f (x ) = U f (x)U −1 where U (a) = exp(iaµ P µ ). K ν ] = [D. M µν ] = −i(g αµ M βν − g βµ M αν + g αν M µβ − g βν M µα ) (c) Check the infinitesimal transformation law for a field conformal transformation δµ (x) = (x 2 ∂µ − 2xµ xν ∂ ν − 2xµ d − 2x ν µν ) U (ε) = exp(iε D).1 (a) Check that the set of 15 infinitesimal transformations defined by (7. D] = 0 [Pµ . D] = iK µ [Mµν .26) is closed under commutation. f (x)] = (x µ ∂ ν − x ν ∂ µ ) f i[D. Polyakov & Zamolodchikov 1984).Problems 269 Problems 7. P λ ] = −i(g µλ P ν − g νλ P µ ) [M αβ . U (b) = exp(ibµ K µ ) exp(− 1 iωµν Mµν ). f (x)] = x µ ∂µ f i[K µ . 2 under the special (x) 7. D] = [Pµ . (b) Check the algebra [Mµν .16) for Dµ (x)).2 Study conformal symmetry in two-dimensional quantum field theories (Belavin.3 When the dilatation current is not conserved one may still define a time-dependent dilatation charge by D(t) = d3 x x µ 0µ (x) = t P 0 + d3 x xi 0i (x) Check that D(t) generates the same dilatation transformation (7. K λ ] = i(gµλ K ν − gνλ K µ ) [K µ . Check the differential representation of the generators i[Pµ . Pν ] = 2i(−gµν D + Mµν ) [M µν . U (ω) = for x µ = x µ + εµ (x).

Ellis (1977)). The first term is given by the diagram 0| jµ jν |0 = = CI (0) 0|I |0 . (b) In the free quark model show that a b T Jµ (x)Jν (0) ∼ xµ →0 3δab (gµν x 2 − 2xµ xν ) dabc εµναβ x α Ac (0) I+ (2π)4 (x 2 − iε)4 2π 2 (x 2 − iε)2 ρS σ (0) cabc x µρνσ Jc + 2 + ··· 2π (x 2 − iε)2 dabc εµναβ x α Jc (0) cabc x ρ Sµρνσ Aσ (0) c + + ··· 2π 2 (x 2 − iε)2 2π 2 (x 2 − iε)2 β β T where a Jµ (x)Ab (0) ν xµ →0 ∼ a 1 Jµ (x) = : ¯ c (x)γµ 2 λa c (x): c (x): Aa (x) µ =: ¯ c (x)γµ γ5 1 λa 2 the repeated index c implies a sum over colours.4 (a) Using the Wick expansion and then the Taylor expansion of the well-defined normal-ordered operator products.59) in the free scalar field theory. P α ] = P α − g α0 i[D(t).5 Obtain the expansion d4 x exp(iq x)T jµ (x) jν (0) = (qµ qν − q 2 gµν ) − 1 −q 2 2m ¯ ln 2 + 4 ¯ 4π 2 µ q +··· = 1 (dabc + icabc )λc 2 i where jµ = ¯ γµ . prove (7. 7. M αβ ] = d3 x ∂µ D µ (x) d3 x (g α0 x β − g β0 x α )∂µ D µ (x) 7. λa are SU (3) flavour matrices 1 4 λa λb γµ γρ γν = (Sµνρσ + iεµρνσ γ5 )γ σ Sµρνσ = (gµρ gνσ + gρν gµσ − gµν gρσ ) Also derive the light-cone expansion for commutators of these currents (for example.270 7 Scale invariance and OPE commutators of D(t) with the Poincar´ generators e i[D(t).

Problems the second by 271 ¯ | jµ jν | = = C¯ (0) ¯|¯ | = C¯ (0) · =C¯ (0) Observe that to this order in α taking the appropriate matrix element singles out one particular contribution to the OPE. in addition (1) (1) to CI and C ¯ given by the same matrix elements (now taken to the first order in α) calculate also the two-gluon matrix element g| jµ jν |g = C ¯ g| ¯ |g + C G g|G α G µν |g µν α where C ¯ . (0) (1) . and find C G . Extend the calculation to first order in α. is given by the zeroth order calculation.

so that there are eight gluons. QCD formulation in the axial gauge n 2 < 0 can be given in a similarly precise form to that in covariant gauges (for example.3 and 3. where n is in each case some fixed four-vector. In terms of bare quantities it reads L = − 1 G α G αµν + 4 µν f ¯ f (iD − m f ) / f + gauge-fixing term + Faddeev–Popov term (8. Bassetto et al. As in QED. Its lagrangian density has been discussed in Sections 1. The quarks are assumed to transform according to the fundamental representation: each flavour of quark is a triplet of the colour group SU (3). Let us therefore rewrite lagrangian (8.2. counterterms QCD is a theory of interactions of quarks and gluons. the crucial issue is the gauge invariance or strictly speaking BRS invariance of the theory. (1985)). Let us concentrate on the renormalization programme. In light-like gauges the general proof of renormalizability is still lacking but order-by-order calculations have been consistently performed. Collins (1984)) that the lagrangian is BRS-invariant after renormalization.1) in terms of the renormalized quantities assuming the most general form consistent with BRS 272 . Gauge bosons transform according to the adjoint representation.46) for the class of covariant gauges. The use of the axial gauge n · Aα = 0. n 2 < 0.1) where the sum is over all quark flavours and the last two terms are given in (3. is also often very convenient in perturbative QCD calculations.1 General introduction Renormalization and BRS invariance.8 Quantum chromodynamics 8. It can be proved (for example. n 2 = 0. In this section we discuss the theory formulation in covariant gauges. or light-like gauge.

1) is the following Aα = Z 3 Aα µ Bµ 1/2 B = Z2 1/2 aB = Z 3 a m B = m Z 0 /Z 2 1/2 ˜ 1/2 ηB = Z 2 η ∗ ˜ 1/2 ηB = Z 2 η ∗ (8.8.76) δBRS B = −igB T α α B (∂µ ηB α B ηB α B ηB B ≡ B (s B (s β B) δBRS ¯ B = igB ¯ B T α δBRS Aα = Bµ ≡ ≡ ≡ ¯ B) γ α B (s ABµ ) + gB cαβγ ηB ABµ ) ≡ B α B (sηB ) ∗α B (sηB )                    (8.5) Lagrangian (8. .4) gB = g Z 1YM /Z 3 3/2 = g Z 1 /Z 2 Z 3 ˜ ˜˜ ˜ = g Z 1/ Z 2 Z 3 ˜ 1/2 (8.6) α δBRS ηB = 1 gB cαβγ ηB ηB 2 ∗α δBRS ηB = −(1/a) β γ µα B ∂µ AB † See also the beginning of Chapter 5.3) follows from the requirement of the BRS invariance for the lagrangian (8. As in QED there is no need for a gauge-fixing counterterm.2) where ˜˜ ˜ ˜ ˜ g Z 1YM /Z 3 = g Z 1 /Z 2 = g Z 1 / Z 2 (8.3) implies the corresponding equality for the ratios of the renormalization constants. The correspondence between the renormalized quantities and the bare quantities of lagrangian (8.1 General introduction 273 invariance.3) If we use a renormalization scheme such as the minimal subtraction scheme in which the renormalized couplings satisfy ˜ g=g=g ˜ ˜ then (8.1) is invariant under the BRS transformation (3. Relation (8. Formally it can be derived from the Slavnov–Taylor identities (Ward identities for a non-abelian gauge theory) which themselves follow from the BRS invariance.2).† We get (for one flavour) L = − 1 Z 3 [∂µ Aα − ∂ν Aα − g(Z 1YM /Z 3 )cαβγ Aβ Aγ ]2 ν µ µ ν 4 ¯ iγµ [∂ µ + ig(Z 1 /Z 2 )Aαµ T α ] − m Z 0 ¯ + Z2 ˜ ∗α αβ 2 ˜ Z 1 / Z 2 )cαβγ Aγ ∂ µ ]ηβ − (1/2a)(∂µ Aβµ )2 ˜ + Z 2 η [δ ∂ + g( ˜ ˜ ˜ µ (8.

(8.4) and (8.5) are inserted into (8.2. The complete set which includes the counterterms is given in Appendix C (in the notation of Chapter 12). δ ¯ .2) can be rewritten introducing counterterms (we take g = g = g) ˜ ˜ / L = − 1 G α G µν + ¯ iD 4 µν α −m ¯ − (1/2a)(∂µ Aµ )2 β − m(Z 0 − 1) ¯ + η∗α (δ αβ ∂ 2 + gcαβγ Aµ ∂µ )ηβ γ ¯ i/ − (Z 1 − 1)g ¯ γ µ Aα T α + (Z 2 − 1) ∂ µ − 1 (Z 3 − 1)(∂µ Aα − ∂ν Aα )2 + 1 (Z 1YM − 1)g(∂µ Aα − ∂ν Aα )cαβγ Aµ Aν ν µ ν µ β γ 4 2 −1 2 − 1 g 2 (Z 1YM Z 3 − 1)cαβγ Aβ Aγ cαρσ Aρµ Aσ ν µ ν 4 ˜ ˜ + ( Z 2 − 1)η∗α ∂ αβ ∂ 2 ηβ + ( Z 1 − 1)gη∗α cαβγ Aµ ∂µ ηβ γ (8.7) can be checked explicitly but it also follows immediately from the invariance of bare lagrangian (8. the behaviour of the Green’s .7) Invariance of the lagrangian (8. δ Aα and δηα (Collins 1984). They are given in Appendix B. In this context the behaviour of observable cross sections has been discussed in Chapter 7 in the framework of the OPE which provides a systematic way of factorizing effects at different mass scales. This is actually what is necessary for finiteness of the composite operators δ . Thus.274 8 Quantum chromodynamics and lagrangian (8.6) if relations (8.8) have been derived in Section 3.1) under (8. Asymptotic freedom of QCD The RGE.2) under (8.8) The Feynman rules for the vertices of (8. the notion of the effective coupling constant and its role in determining the behaviour of a quantum field theory at different momentum scales have been extensively discussed in Chapter 6.2) under δBRS ˜ = −ig Z 1 T α ηα γ ˜ ˜ (∂µ ηα Z 2 + g Z 1 cαβγ ηβ Aµ ) ˜ δBRS ¯ = ig Z 1 ¯ T α ηα δBRS Aα = µ ˜ δBRS ηα = 1 g Z 1 cαβγ ηβ ηγ 2 δBRS η∗α = −(1/a) ∂µ Aµα                    (8. QCD is an asymptotically free theory: the coupling constant decreases as the scale at which it is defined is increased.7).3). µ ˜ Lagrangian (8.6) and if is identified as 1/2 ˜ 1/2 = (1/Z 3 Z 2 ) B ˜ We note the presence of the divergent renormalization constants Z i in the BRS transformation (8.

1 General introduction 275 functions when all the momenta are scaled up with a common factor is governed by a theory where g → 0.33) and (6.8). This can explain the success of the parton model in describing the scaling phenomena in the deep inelastic processes.45) β(α) = α α b1 + π α π 2 α = g 2 /4π b2 + α π 3 b3 + · · · (6. 8. Only theories with a non-abelian gauge symmetry can be asymptotically free in four space-time dimensions (Coleman & Gross 1973). and in the diagram calculation one often encounters the following combinations: a a 2 (TR )bc (TR )cd = (TR )bd = CR δbd a.5) one can use one of the relations 2 3 2 2 ˜2 ˜2 Z α = Z 1YM /Z 3 = Z 1 /Z 2 Z 3 = Z 1 / Z 2 Z 3 (8. Tr[λa λb ] = 2δ ab and in the regular (adjoint) eight-dimensional representation (T a )bc = −icabc where cabc s are the SU (3) structure constants.c (8.2) the new features in QCD Feynman diagram computation are the group theory factors present for each diagram. Both appear in the QCD vertices. One can use the most convenient way of calculating Z α . For instance at the one-loop level we have the contributions shown in Fig. The β-function in QCD defined as β(α) = µ(d/dµ)α(µ2 )|αB fixed .45) and the coefficients bi are calculable perturbatively. As compared to QED (Section 5.35): β(α) is totally determined by the residua of the simple UV pole of Z α (beyond the one-loop order we need to calculate a non-leading UV divergence) where according to (8. Tadpole diagrams have been ignored because they do not contribute when dimensional regularization is used.9) Diagrams contributing to different renormalization constants are determined by the structure of counterterms in lagrangian (8.8.1 (sums over flavours and over different permutations are omitted). The basic ingredients are SU (3) generators in the fundamental three-dimensional representation conventionally taken as (T a )bc = 1 (λa )bc 2 where λs are the Gell-Mann matrices. Using dimensional regularization and a mass-independent renormalization scheme we have (6. is given by expansion (6.10) .

and a b a b (TR )dc (TR )cd = Tr[TR TR ] = TR δ ab c.276 8 Quantum chromodynamics Fig.1. 8.11) CR is the eigenvalue of the quadratic Casimir operator in the representation R and .d (8.

respectively. L and M for the composite operators s A. s ¯ . ¯ L. Then one derives the non-abelian analogue of (5. α for the fields A. the coefficient b2 by Caswell (1974) and Jones (1974) and the coefficient b3 by Tarasov. In addition to sources J. α.3.53) in the next-to-leading approximation.12) 79 3 2 −b3 = 2857 CA − 1415 CA TF n + 432 CA TF2 n 2 − 205 CA CF TF n    1728 864 288    44 1 2 + 288 CF TF2 n 2 + 16 CF TF n The QCD effective coupling constant is given by expression (6. and generally by (6. ¯ and .11) (Abers & Lee 1973. This has been discussed in Section 7. . s and sη that appear in the BRS † The coefficient b1 was first calculated by Politzer (1973) and Gross & Wilczek (1973). One can proceed as in Section 5. The Slavnov–Taylor identities As with the Ward identities in QED. one ¯ introduces sources cα and c∗α for the ghost field η∗α and ηα as well as sources K . This form generates all the identities for the connected Green’s functions but. Vladimirov & Zharkov (1980).52). expression (6. It is often convenient to express the result of a QCD calculation in terms of the dimensional parameter instead of the dimensionless α. Marciano & Pagels 1978). their analogue in a non-abelian gauge theory – the Slavnov–Taylor identities for bare regularized or for renormalized Green’s functions – can be derived in several different ways.8.45) have been calculated and in the minimal subtraction scheme they are (for SU (N ) and n flavours)†  −b1 = 11 CA − 2 TF n  6 3     17 2 1 5  −b2 = 12 CA − 2 CF TF n − 6 CA TF n (8.51) in the leading logarithm approximation. ).1: apply gauge transformation to the generating functional for full or connected Green’s functions which is written as a functional of the physical field sources only. because of the presence of ghost fields it is very complicated to obtain the analogous relation for the 1PI generating functional (A.1 General introduction 277 for SU (N ) CF = (N 2 − 1)/2N CA = N The trace TR is TF = 1 2 F ≡ fundamental A ≡ adjoint TA = N So far the first three terms in the expansion (6. ¯ . A more elegant method is based on the BRS transformation (Lee 1976).

η. let us consider the regularized Green’s function 0|T Aα (x)η∗β (y)|0 . K . The general relations whose derivation has just been briefly described contain all the information about BRS invariance and are useful in formal manipulations. L. Llewellyn Smith (1980)).13) ¯ + c∗ η + η∗ c + K s A + s ¯ L + Ls Changing variables in the functional integral according to the BRS transformation and using the invariance δSeff /δ as well as the nilpotent relations s 2 A = s 2η = s 2 = s2 ¯ = 0 =0 one obtains the desired general relation generating the Slavnov–Taylor identities. c∗α = −δ /δηα etc. α.6). One can convince oneself that in the ¯ presence of additional sources K . c∗ . α. For instance. As usual this is achieved by means of the Legendre transform ¯ ¯ [A. This is to obtain identities that are linear in derivatives with respect to the sources. c. η∗ . K . M] = Z [J.15) . α. Using (8. L . ¯ . . Since the ghost fields are treated on equal footing with the physical fields it is now straightforward to obtain similar relation-generating identities for the 1PI Green’s functions. c∗ . L and M the functional still generates 1PI Green’s functions. L.14) where ηα = δ Z /δc∗α . L . M] ¯ = D(A ¯ ηη∗ ) exp iSeff + i d4 x (J A + α ¯ + Msη) + ¯α (8. M] ¯ − d4 x (J A + α ¯ + ¯ α + c∗ η + η∗ c) (8. L . however. α. to get specific identities for the Green’s functions of interest is based on the observation that since BRS invariance is an exact symmetry of the theory all the Green’s functions are BRS invariant (for example. Next one defines the generating functional ¯ W [J. K . L. The simplest way. c.278 8 Quantum chromodynamics transformation (8.6) we have µ δBRS 0|T Aα (x)η∗β (y)|0 = 0|T (∂µ ηα (x) + gcαβγ ηβ (x)Aγ (x))η∗β (y)|0 µ µ − (1/a) 0|T Aα (x)∂ν Aνβ (y)|0 = 0 µ (8. L .

. J·A = α d4 x Jµ Aαµ (8.8. (x)] (8. . µ µ Invariance of the action under the transformation (1. Let us consider first the gauge-invariant part of the action and introduce the background field by the replacement A µ = A µ + Aµ (8.56a) we conclude that µ ν (1/a)∂(x) ∂(y) 0|T Aα (x)Aβ (y)|0 = −iδ(x − y)δ αβ µ ν (8.17) and define. (x)].14)) the longitudinal part of the propagator is unchanged by the higher order corrections.2 The background field method This method is based on the idea of quantizing field theory in the presence of a background field chosen so that certain symmetries of the generating functionals are restored. However.19) Manifest gauge invariance is lost because of the gauge-fixing procedure. as usual. (x) = i α (x)T α and we define Aµ (x) = igAα (x)T α ).18) (8.20) where Aα is the background field and Aα is the quantum field (we recall that µ µ Aµ (x) = ig Aα (x)T α .15) with respect to xµ and using the equation of motion for the Green’s function 0|T ηα (x)η∗β (y)|0 which can be derived analogously to (2. .21) implies invariance under two kinds of transformations on Aµ and Aµ which give the same δ Aµ quantum δAµ = 0 δ Aµ = Dµ (x) + [Aµ . we can restore it by introducing a background field with properly chosen transformation properties.2 The background field method 279 Differentiating (8.22) . Dµ = ∂µ + [Aµ .] (8.16) implies that as in QED (see (5.113) δ Aµ = Dµ (x) = ∂µ (x) + [Aµ (x). We start with the gauge-fixed functional integral (the dependence on other fields and on the ghost sources is not indicated explicitly) W [J ] = D Aµ exp(iSeff + iJ · A ). 8.16) Result (8. Let us consider gauge theories. Z [J ] = −i ln W [J ] and the effective action ¯ ¯ F[ A] = Z [J ] − J · A. α ¯α Aµ = δ Z /δ Jµ (8.

Aµ ]/δ Jµ if ˜µ δ Aα = cαβγ β (8. .25) is manifestly invariant under the background gauge transformations if we complete them with the transformation α δ Jµ = cαβγ β γ Jµ (8.27) (8. . Changing the integration ˜ variables in (8. Aµ ] and for ˜ ˜ ˜ ˜ [ Aµ . The generating functional ˜ W [J. Aµ ] = Z [J. Aµ ] = D Aµ exp{iSeff [Aµ + Aµ ] + iJ · A} (8. (x)] (8. Aµ ))2 = −(1/2a)(Dµ Aµ )2 = −(1/2a)(∂ µ Aα − gcαβγ Aµ Aγ )2 (8. For instance. Aµ ] = W [J ] exp(−iJ · Aµ ) (8. Aµ ] in the presence of the background field.280 8 Quantum chromodynamics background δAµ = Dµ (x).24) α µ β µ The Faddeev–Popov ghost term then takes account of ˜ det Mαβ = det(δ F α /δ β ) ˜ where δ F α corresponds to the quantum gauge transformation δ Aα = (1/g)∂µ µ α + cαβγ β (Aγ + Aγ ) µ µ and it also transforms covariantly under the background gauge transformation. Dµ = ∂µ + [Aµ . A] transforms covariantly under these transformations. .26) The same is true for ˜ ˜ Z [J.28) ˜µ Aγ ¯ The next step is to find the relationship between the original [ A] and the ˜ ˜ [ Aµ .29) . choose ˜ ( F α (Aµ .23) We can maintain the manifest gauge invariance of Seff with respect to the ˜ background transformations if the gauge-fixing function F α [A.] δ Aµ = [Aµ (x). Aµ ]: Aµ → Aµ + Aµ one gets ˜ W [J. Aµ ] − J · A α a ˜ ˜ Aµ = δ Z [J.25) for W [J. Aµ ] = −i ln W [J.

Aµ ] is the usual effective action ¯ ˜ ¯ ¯ µ ] evaluated at Aµ = Aµ + Aµ . the calculation of the β-function in the one-loop approximation simplifies to the calculation of the diagrams in Fig.3). We get [Aµ ] = ˜ [0. Aµ ] (check it).30) ˜ and we conclude that the effective action ˜ [ Aµ . i. Aµ ] must take the form of a divergent constant times (Gα )2 .e. Aµ ] for Aµ = 0. Aµ ] = Z [J ] − J · Aµ − J · Aµ = [Aµ + Aµ ] (8.e. restricting ˜ ˜ µ lines. It should also be remembered that this [ Aµ ] [A α α ˜ corresponds to the gauge-fixing term F [A ] = F [A − A.34) and. A] which may be an unusual gauge in the standard approach.32) A very important fact is that ˜ [0. for instance. The background field method is extensively used in supergravity theories (Gates. Aµ ] (8. Aµ ] is invariant under the transformation δAµ = Dµ (x). diagrams with only internal Aµ lines and external Aµ . Aµ ] given by (8.3) that the renormalization constants (Aµ )B = Z A Aµ gB = Z g g satisfy Zg = ZA −1/2 1/2 (8. ˜ ˜ In particular it is convenient to calculate ˜ [ Aµ .e. it is a gauge-invariant functional of Aµ . i. where µν Gα = ∂µ Aα − ∂ν Aα − gcαβγ Aβ Aγ µν ν µ µ ν It is then straightforward to show (Problem 8. i. It implies. 8.33) (8. Roˇ ek & Siegel 1983). It is recommended that the reader derives the Feynman rules necessary for the calculation of ˜ [0. Aµ ] and completes the evaluation of the β-function in the oneloop approximation (Problem 8. c .8.2 The background field method 281 and correspondingly ˜ Z [J. Aµ ] = Z [J ] − J · Aµ Therefore ˜ ˜ ˜ ˜ [ Aµ .2. These ‘ Aµ -vacuum’ diagrams are ourselves to diagrams with no external A ˜ the 1PI subset of diagrams obtained from W [0. This property is maintained in a perturbative loop-by-loop calculation of ˜ [0. Grisaru. Thus the infinities appearing in ˜ [0. that the divergent terms and the counterterms are manifestly invariant with respect to the background gauge transformation.25). in particular. ghost and other non-gauge field lines.31) (8.

As far as the gauge functions U (r) are concerned the three-dimensional spatial manifold with points at infinity identified is topologically equivalent to S 3 – the surface of a four-dimensional Euclidean sphere labelled by three angles. for example. 8. As we know.8).3 The structure of the vacuum in non-abelian gauge theories Homotopy classes and topological vacua For the moment we shall consider the pure Yang–Mills gauge theory with SU (2) as the gauge group. Since any element M in the SU (2) group can be written as M = a + ib · σ where σ s are Pauli matrices and where real a and b satisfy a 2 + b2 = 1 . that physically admissible gauge transformations satisfy (8. In pure Yang–Mills theory one can argue (Jackiw 1980). the so-called gauge-fixing procedure used in perturbation theory usually leaves some residual gauge freedom.36) which ensures that the non-abelian charge is well defined (see Problem 1. Thus the vacuum defined by G µν = 0 restricts the potentials to ‘pure gauge’ configurations Ai (r) = −U −1 (r)∂i U (r) Further discussion relies on an additional restriction that r →∞ (8.36) where U∞ is a global (position-independent) gauge transformation. For instance we can use the gauge A0 = 0 (for the present discussion this is convenient but not necessary. but not prove.35) lim U (r) = U∞ (8. Coleman (1979)) which obviously leaves room for further gauge transformations U (r) depending on spatial variables only.282 8 Quantum chromodynamics Fig. 8. Equivalently. we assume that all vector potentials fall faster than 1/r at large distances. see. The matrix functions U provide a mapping of S 3 into the manifold of the SU (2) gauge group.2.

The classes of mappings S n → S n with . For this example 1 (U (1)) = 1 (S 1 )=Z where Z denotes an additive group of integers. 0) = f 0 (x) and F(x. A group structure can be defined on the set of homotopy classes. and only if. We can divide all mappings of X into Y into homotopic classes: two mappings are in the same class if they are homotopic. Let us consider a simple example. such that F(x.39) Instead of the unit circle S 1 we can consider the whole real axis −∞ < x < ∞ with the points x = ±∞ identified which is topologically equivalent to S 1 (see Problem 8.37) form a homotopic class for different values of a and a fixed integer N . with θ and θ + 2π identified. Take the mappings from S 1 into G ≡ U (1) represented by a set of unimodular complex numbers z = exp(iu) which is itself topologically equivalent to S 1 .e. f 1 be two continuous mappings from X into Y . Thus. t).3 The structure of the vacuum 283 we conclude that the manifold of the SU (2) group elements is topologically equivalent to S 3 . The function F(x. The group of homotopy classes of mappings from S 1 into the manifold of G is called the first homotopy group of G and is denoted by 1 (G). They can be classified into homotopy classes. if.1). We observe that in our example the winding number N can be written as N = −i 0 2π 1 du dθ 2π u(θ) dθ (8. Hence the integer N is called the winding number and each homotopy class is characterized by its winding number. t) is called the homotopy. i. 1) = f 1 (x).8. This discussion can be generalized by taking X = S n (the n-dimensional sphere) or the topologically equivalent n-dimensional cube with all its boundaries identified and equivalent to some x0 of S n . One can think of u(θ) as a mapping of a circle into another circle such that N points of the first circle are mapped into one point of the second circle (winding N times around the second circle). gauge transformations U (r) provide mappings S 3 → S 3 . Let X and Y be two topological spaces and f 0 . Let S 1 be a unit circle parametrized by the angle θ. We are interested in mappings from S 1 into the manifold of a Lie group G. 0 ≤ t ≤ 1. Thus S 1 → S 1 and the continuous functions u(θ ) = exp[i(N θ + a)] (8.38) The mappings of any winding number can be obtained by taking powers of u (1) (θ) = exp(iθ) (8. They are said to be homotopic if they are continuously deformable into each other. there exists a continuous deformation of maps F(x.

38) and (8.42) where the Ai s are given by (8. mappings S n → S n are classified by the number of times one n-sphere covers the other.36) implies an infinity of topologically distinct vacua for the SU (2) Yang–Mills theory.e. Only if the group is U (1). The same result holds for an arbitrary simple Lie group (for SU (N ) in particular) due to a theorem (Bott 1956) which says that any continuous mapping of S 3 into G can be continuously deformed into a mapping into an SU (2) subgroup of G. We have n (S n )=Z (8.35) and (8. H= a a i ∂0 A i −L (8. It can be shown that analogously to (8. For prototype mapping with N = 1. see Problem 8.44) The hamiltonian H= d3 x H(x). As we have already mentioned the group SU (2) is also equivalent to S 3 .45) .1. is every mapping of S 3 into U (1) continuously deformable into the constant map f (x) = y0 corresponding to N = 0. Physical vacuum The inequivalence of topologically distinct vacua can be understood as a consequence of Gauss’ law Diab G b = 0 0i (8.1 the winding number for gauge transformations providing the mapping S 3 → G is given by N = −(1/24π 2 ) d3 x Tr[εi jk Ai A j Ak ] (8.284 8 Quantum chromodynamics one fixed point f (x0 ) = y0 form a group called the nth homotopy group of S m and designated by n (S m ).40) i.35).168) in Problem 8. Thus 3 (SU (2)) = 3 (S 3 )=Z (8.41) and this together with relations (8.43) To see this let us first recall the canonical formalism. Ordinary space R 3 with all points at ∞ identified is equivalent to S 3 . In the Aa = 0 gauge the 0 canonical variables are the Aia s and their conjugate momenta a i = δL = G a = ∂0 Aia = −E ia 0i δ(∂0 Aia ) (8.

t)] = 0 j j [E ia (x. The generator of these gauge transformations is (see Problem 1. for a gauge transformation in the class N = 1 we have T (1) |N = |N + 1 (8.48) (8.49) (8. We cannot set Diab E ib to zero as this operator does not commute with the canonical variables. t)] = [E ia (x. t). Integrating by parts and noting that for the N = 0 class one gets QG = d3 x [−Diab G a 0i b (x)] (8. that only gauge transformations which belong to the homotopy class N = 0 can be built up by iterating the infinitesimal gauge transformations (see Problem 8.51) Having (8. However. although we obtain a consistent quantum theory. E ia ] = εi jk D ab Bk j (8. it generates infinitesimal space-dependent gauge transformations. t). Bia = − 1 εi jk G a jk 2 (8. it is different from the desired gauge theory.53) we conclude that such transformations leave the vacuum state |N invariant: T (0) |N = |N (8.52) (x) vanishes at x → ∞ ˆ and U = exp(iQ G ). Ab (y.53) Using (8. Ab (y. as the next step. t).51) and (8. Thus.47) When the canonical commutation relations are imposed [Aia (x. t)] = iδ ab δi j δ(x − y) j the hamiltonian equations give ∂0 Aia = i[H.53). Aia ] = −E ia b ∂0 E ia = i[H.55) .2).54) ˆ where T (0) is the unitary transformation corresponding to U = exp(iQ G ) in the class N = 0.8) QG = d3 x G a Diab 0i b (x) a (8.46) (8.3 The structure of the vacuum 285 is the energy H= 1 2 2 d3 x (E a + Ba ) = d3 x 00 2 a E ia = G i0 = − ia . as seen from (8. E b (y.50) but Gauss’ law (8.51) we can return to our main question and recognize.8.43) is not found. Actually. Gauss’ law can be incorporated into the theory by demanding that the physical states are only those which are annihilated by Diab G b 0i Diab G b | 0i =0 (8.

(8. However. one could choose a physical gauge where Aµ is uniquely determined in terms of G µν with no residual gauge freedom at all.58) = ∞ N =−∞ exp(iN )|N (8. the physical | -vacuum is. if O is any gauge-invariant operator ˆ ˆ [ O. In the absence of tunnelling. Dashen & Gross (1976)).286 8 Quantum chromodynamics Since the states |N are functionals of Aµ .57) The physical vacuum is characterized by which is a constant of motion and each | vacuum is the ground state of an independent sector of the Hilbert space. In our gauge.42) −1 −1 N (Un AUn − Un ∂i Un ) = N (A) + n (8. 0 ≤ ≤ 2π. and the physical vacuum is given by | so that T (M) | = exp(−iM )| (8. Gauge invariance led us to the notion of the physical | vacua as superpositions of the topological vacua (see also Callan. the degeneracy between different -vacua is lifted if there exists tunnelling between topological vacua. ˆ ˆ |[ O. This interpretation of the | -vacuum is certainly gauge-dependent (Bernard & Weinberg 1977). Because T is unitary its eigenvalues are exp(−i ).55) follows from the transformation properties of the winding number N (A) given by (8. of course. ˆ Indeed. In pure Yang–Mills theories such tunnelling is due to instantons.56) where Un is in the class N = n and Aµ is pure gauge: A = H −1 ∂µ H . U N ]| = [exp(−iN ) − exp(−iN )] ˆ | O| . As in the quantum mechanical problem. Now the true vacuum must be at least phase-invariant under all gauge transformations because they commute with observables. U N ] = 0 then 0= and ˆ | O| = 0 if = (8. pure gauge configurations for which E and B are zero are analogous to infinitely degenerate zero-energy configurations in a quantum mechanical problem with periodic potential. For instance.59) We want to stress that we have been working in a fixed gauge A0 = 0. when there are massless fermions in the theory. for example. The classical vacuum will be unique. Thus the definitions of the winding number as well as of the topological vacua are gauge-dependent. the -vacua remain degenerate and the angle has no physical meaning.

This will become clear in the next subsection where we discuss the functional integral representation for vacuum-to-vacuum transitions.3 The structure of the vacuum 287 a gauge-invariant concept.64) . We are interested in transitions from one classical vacuum to another so we require that the initial t → −∞ and final t → +∞ field configurations are of the pure gauge form. we must pay some attention to the possible boundary conditions on the fields in the integral (8. Firstly.7) and no surface terms arise from the second term.63) ˜ d4 x Tr[G µν G µν ] (8. Since (see Problem 1. It is stationary with respect to any local variation δ Aµ whether or not the equation of motion is satisfied δQ ∼ = =− 1 2 ˜ d4 x Tr[G µν (D µ δ Aν − D ν δ Aµ )] ˜ d4 x Tr[G µν D µ δ Aν ] ˜ d4 x Tr[Dµ G µν δ Aν ] + ˜ d4 x ∂ µ Tr[G µν δ Aν ] (8.62) The first term vanishes from the Bianchi identities (Problem 1. Let us now consider the integral Q = −(1/16π 2 ) where ˜ G µν = 1 εµνρσ G ρσ 2 Q is gauge-invariant and is called the topological charge or the Pontryagin index. Q is a topological invariant.60). Actually.60) It remains to be understood what one means by ‘vacuum’ for the transition given by (8.6) ˜ −(1/16π 2 ) Tr[G µν G µν ] = ∂µ K µ (8. -vacuum and the functional integral formalism Let us consider a vacuum-to-vacuum transition given by the functional integral 0| exp(−iH t)|0 ∼ D Aµ exp i d4 x (LYM + LG + LFP ) (8.61) For the time being we leave the gauge unspecified.60).8. we shall require that gauge potentials tend to a pure gauge at large distances in all four directions Aµ −→ U −1 ∂µU |xµ |→∞ (8.

The result turns out to have the gauge-invariant form for which we are looking.67) Using the remaining gauge freedom we can set N (−∞) = 0.66) To make contact with the discussion earlier in this section let us take the gauge A0 = 0. We have | exp(−iH t)| = m.67) to be finite) and the functional integral (8. Therefore in this gauge the topological charge Q can be interpreted as the winding number of the pure gauge configuration to which Ai .42) and 8. Then from (8.60).61) Q may be written in terms of U by inserting the asymptotic form of K µ in (8.62). tends (we always also assume (8.65) Q = −(1/24π 2 ) dSµ εµαβγ Tr[U −1 ∂α UU −1 ∂β UU −1 ∂γ U ] (8.66 Q= d3 x K 0 ∞ −∞ = N (+∞) − N (−∞) (8.36) which is necessary for the integral in (8.68) ˜ + (1/16π 2 ) Tr[G µν G µν ]} where we have used (8.n exp[i(n − m) ] exp[in( − ) Q − )] m| exp(−iH t)|n = 2πδ( ∼ Q exp(−iQ ) N + Q| exp(−iH t)|N D A Q exp i d4 x (LYM + LG + LFP ) exp(−iQ ) D Aµ exp i = d4 x {LYM + LG + LFP (8. In the gauge A0 = 0 we are now ready to construct the functional integral corresponding to the | → | transition.62).288 8 Quantum chromodynamics where K µ = −(1/16π 2 )εµαβγ Tr[G αβ Aγ + 2 Aα Aβ Aγ ] 3 we can write Q as an integral over the surface S at infinity Q= dSµ K µ (8. as the transition amplitude between topological vacua |N = 0 and |N = Q or more generally between |N and |N + Q . The gauge-dependent notion of . with the integral restricted to the Q sector defined by (8. As promised the final result is gauge-independent and it defines the theory having a well-defined vacuum state | The | -vacuum is defined by the vacuum expectation values of operators which in turn are determined by the functional integral with a given value of .65) For potentials Aµ satisfying (8.

The instantons are classical E = 0 trajectories in Euclidean space-time. this does not help since C P-violating weak interactions renormalize to a non-zero value. Firstly. Wilczek 1978. With massless fermions changes with chiral rotations and can be rotated away. Dine. hence in Euclidean space A4 = 0) one can cast the instanton solution into a form which tends to pure gauge configurations satisfying (8. Also it does not contribute to the energy–momentum tensor. Fischler & Srednicki 1981). New interesting aspects then appear which cannot be considered in detail before discussion of chiral symmetry and anomalies. However. However. This is no longer true when fermions are massive since then chiral symmetry is also broken explicitly. The new term in the lagrangian is a total divergence so it does not influence the classical equations of motion.† Fermions must be included in a realistic theory. So far no convincing explanation exists for the smallness of (Peccei & Quinn 1977. This mechanism does not imply the existence of the Goldstone boson coupled to the gauge-invariant UA (1) current and it offers a prototype solution to the so-called UA (1) problem.8. Then the -vacua turn out to be degenerate: they are the degenerate vacua of spontaneously broken axial UA (1) global symmetry of a theory with massless fermions. . it modifies quantum dynamics which depends on the action.36) and having different winding numbers as x4 → +∞ and x4 → −∞. We face a situation where on one hand the field configurations with Q = 0 are welcome to solve the UA (1) problem and on the other hand the experimentally allowed value of is surprisingly small as for a strong interaction parameter: from the upper limit on the neutron dipole moment < 10−19 . However. The spontaneous breakdown of the UA (1) is the combined effect of the anomaly in the divergence of the axial current and of the existence of field configurations with a non-zero topological charge Q. take fermions to be massless. It is convenient to work in Euclidean space and to look at Minkowski Green’s functions as the analytic continuation of the Euclidean ones. Weinberg 1978.3 The structure of the vacuum 289 the topological vacua does not appear in this formulation. Thus tunnelling takes place and the -vacua are not degenerate. Finally in the presence of fermions the spontaneous breakdown of the UA (1) and of the chiral † By a gauge transformation which removes A4 (we are in the gauge A0 = 0.68) we can always write the physical vacuum as a superposition | = dν exp(iν )|ν ˜ where |ν is an eigenstate of the operator Tr[G µν G µν ] with the eigenvalue ν. The -term violates P and C P conservation so one could argue that we should set = 0 in the strong interaction theory. using the path integral representation (8. Let us summarize briefly only the most important facts (see also Chapter 13). Explicit solutions for field configurations giving Q = n = 0 are known (instantons).

in which there is a large momentum scale q 2 .5. The first one is to assume the parton picture which provides a connection between QCD and hadrons. Parton picture There are two basic ingredients in applications of perturbative QCD to hadronic processes. the assumption that in processes with large momentum transfer q hadrons can be pictured as ensembles of partons (quarks and gluons) and that there is no interference between the long time and distance physics responsible for the binding of quarks and gluons into hadrons and the short distance effects relevant for the large momentum transfer process under consideration and occurring at the parton level. not necessarily light-cone dominated. The parton picture is implicit in the OPE analysis of the deep inelastic lepton– hadron scattering in Section 7. non-perturbative in nature. for instance. this OPE combined with the renormalization group techniques are applicable to a limited number of problems and a more general approach based on direct summation of Feynman diagrams has been widely developed. However. One then simply sets = 0. 8. The life-time of a virtual state with quasi-free constituents is P/( E)2 and it becomes long compared to the collision interaction time which is O(1/q0 ) (here E is the scale of binding effects: E ∼ O(1/r ) where r is the radius of the hadron).4 Perturbative QCD and hard collisions Historically. the Bjorken limit in the deep inelastic lepton–hadron scattering discussed in Section 7.5 and. it is usually neglected in applications of perturbative QCD to short distance phenomena. the light-cone expansion discussed in Chapter 7 was the original foundation for perturbative QCD applications. in the QCD calculation of the † This point has been particularly emphasized by Crewther (1979). It can be used for any hard scattering process. speaking most generally. However. If in a given process there are several large momenta O(q 2 ) then we assume the ratio of these variables to be fixed as q 2 → ∞. for example. The time scale of the bound state effects is dilated as P rises. No such connection has so far been derived from QCD itself. This is. .† The structure of the QCD vacuum briefly discussed in this section is of fundamental physical importance. The usual argument to support this conjecture is as follows: view the scattering in a frame in which the hadron (or hadrons) is fast: P → ∞.290 8 Quantum chromodynamics SU (n) × SU (n) are phenomena which are connected to each other by the chiral Ward identities based on the spontaneous breaking of the SU (n) × SU (n) and on the anomalous non-conservation of the UA (1) current. By the parton picture one means.

Sachrajda 1983b). A typical example. and use the so-called light-cone parametrization of the four-momenta  P = (P + m 2 /4P. p 2 )σiQCD (z P. it is related to the square of the soft wave-function. In this limit we envisage the proton as a collection of massless.e. p⊥ and p 2 finite (infinite momentum frame) we see that p = z P up to corrections O( p⊥ /z P).2.69) P = (z P + ( p⊥ + p⊥ )/4z P. but not the only one . We assume that in the Bjorken limit dσ had (P. however. p⊥ . In most applications.4 Perturbative QCD and hard collisions 291 total cross section for the e− e+ annihilation into hadrons which is identified with the total cross section e− e+ → quarks and gluons. To be specific let us consider the deep inelastic electron–proton scattering depicted in Fig. The problem is that. which are process-independent. p⊥ . q) = i dz d2 p⊥ d p 2 f i (z. In the limit P → ∞. z P − ( p⊥ + p )/4z P)    d4 p = 1 (dz/|z|) d2 p⊥ d p 2 2 which defines z as z = ( p0 + p3 )/2P and is very convenient for this discussion. We expect that both p⊥ and p 2 are determined by soft binding effects. It is a generalization of the factorization encountered in the OPE approach and applies to the dσ QCD which under the parton model assumption is to be calculated perturbatively as the sum of appropriate quark–gluon Feynman diagrams. p 2 .129) and the discussion following it). q) (8. 1979. i.70) where σiQCD is the cross section for deep inelastic scattering on the parton ‘i’ as the target and f i is the four-momentum distribution of partons of type ‘i’ in the proton. We shall see shortly that in the infinite momentum frame (8.8. with the cross section for the hard subprocess which involves only partons and can be studied perturbatively. p⊥ . z + dz of the total proton momentum. as we expect from our QED examples. P − m 2 /4P)    2 2 2 2 (8. We call P and p the proton and parton momenta. parallelly moving partons and f i (z) dz as the average number of partons of type ‘i’ having momentum fraction z.70) has a very interesting physical interpretation (see (8. the perturbative expression for dσ QCD contains singular contributions coming from certain regions of phase space in which the quark and gluon momenta are small or parallel. We assume that the hard scattering cross section can be written as a convolution of soft hadronic wave functions. this parton model assumption takes a more concrete form. O. 7. respectively. Factorization theorem The second central issue for perturbative QCD calculations is the factorization theorem (Ellis et al.

For instance. In this second case it is usually convenient to let M 2 be the scale q 2 itself and simultaneously be the renormalization scale.73) In the following we explicitly get (8. one can compute the cross section for lepton pair production in p p collisions using ¯ distribution functions defined and measured in the inclusive lepto-production. again for the lepton–parton deep inelastic scattering dσiQCD ( p. Sachrajda 1983a. Then. for example. For more details both technical and phenomenological the reader should consult some of the excellent review articles (e.72) and (8. Dyakonov & Troyan 1980) on this subject. Buras 1980. It is. although we cannot calculate the cross sections dσ had or even dσ QCD from first principles. is large logarithms like α ln(q 2 / p 2 ) coming from collinear parton radiation where q 2 is the large scale of the process and p 2 is the parton mass. This may concern the same process at different momentum transfers. absolutely crucial for any relevance of perturbative QCD to hard hadronic processes that the factorization theorem holds asymptotically to all orders in perturbation theory. Using (8. The idea of factorization can be illustrated by the identity [1 + α ln(q 2 / p 2 ) + · · ·] = [1 + α ln(M 2 / p 2 ) + · · ·][1 + α ln(q 2 /M 2 ) + · · ·] (8. q) = dy Fi (y. Reya 1981. In the following we present a sample of lowest order QCD calculations and then briefly mention possible strategies for all order extensions. The cross sections dσi are finite in ˆ the limit p 2 → 0 and M 2 is an arbitrary mass introduced to factorize out the long distance effects of perturbation theory. the theory has a clear predictive power by relating one cross section to another. for example.73) in the first order perturbation theory. . scaling violation in the deep inelastic lepton–hadron scattering or two different processes that involve the same ˜ distribution functions Fi . q 2 /M 2 ) + O( p 2 /M 2 ) ˆ (8.70) and (8. however.72) where 0 < y < 1 and the Fi are universal process-independent functions containing all the logarithms singular in the limit p 2 → 0 (mass singularities or more generally all long distance singularities).292 8 Quantum chromodynamics (Sachrajda 1983b).72) the hadronic cross section can be written as a convolution dσ had = i f i ∗ Fi dσi = ˆ i ˜ ˆ Fi dσi (8.g. Dokshitzer.71) and the theorem which has been proved to all orders in perturbation theory up to power corrections O( p 2 /q 2 ) states that. p 2 /M 2 ) dσi (yp.

78) The most general Lorentz and parity-invariant form of the tensor Wµν reads Wµν = −(gµν −qµ qν /q 2 )W1 +[Pµ −qµ (P ·q/q 2 )][Pν −qν (P ·q/q 2 )] W2 (8.74) where pn are four-momenta of the final particles and X includes all the necessary spin indices for a given final state (we neglect all particle masses with the exception of the initial nucleon mass).77) d3 pn (2π )4 δ (4) P + q − 2E pn (2π )3 n=1 d4 x exp(iq · x) P|Jµ (x)Jν (0)|P ˜ pn W µν n = pol (8. The cross section in the laboratory frame then reads (Flux = 4(P · k) = 4Em) dσ α 2 16E 2 dσ mEE = 4 [cos2 ( 1 θ )W2 + 2 sin2 ( 1 θ )W1 ] (8.76) comes from the lepton vertex we can rewrite the cross section as follows (q ≡ k − k ): dσ = where we define 4π W µν ≡ N e4 1 d3 k 4π L µν W µν q 4 Flux 2E k (2π )3 N (8. 7. they are Lorentz scalars.e.79) m2 where Wi = Wi (q 2 . Writing 1 4 pol ˜ |M X |4 = L µν W µν (8. i.5 Deep e–n scattering 293 8.2.8. The unpolarized cross section for the process eN → eX has the form dσ = 1 d3 k Flux 2E k (2π)3 d3 pn 2E pn (2π )3 n=1 pn n N N ×(2π)4 δ (4) k + P − k − e4 1 q4 4 |M X |4 pol (8. The kinematical variables are defined in Fig.80) = 2 2 d dE π d|q 2 | dν q 4m .5 Deep inelastic electron–nucleon scattering in first order QCD (Feynman gauge) Structure functions and Born approximation We consider this process in the one-photon exchange approximation.75) where the tensor L µν ≡ 1 2 Tr(/ γµ/ γν ) = 2(kµ kν + kν kµ − k · k gµν ) k k (8. ν).

79) and (8.80) we see that the structure functions W1 and W2 defined by (8.84) is a short-hand notation for: in the case of the proton as a target: ¯ f i (z) dσiQCD = [ 4 (u p + u p ) + 1 (dp + dp ) + 1 (sp + sp )] dσ q + G p (z) dσ G ¯ ¯ 9 9 9 i (8.82) are written for the general case of n space-time dimensions.82) Pµ Pν 1   Wµν = −F1 + F2  2x P ·q 2x where x = −q 2 /2P · q is the Bjorken variable. We also introduce another often used notation F1 = W1 . According to (8.81) region |q 2 | ∼ s are free of kinematical singularities when m → 0.85) .75) and (8.294 8 Quantum chromodynamics This result follows directly from relations (8.79).70) the lepton–hadron scattering is an incoherent mixture of elementary scatterings dσ had = 0 1 dz i f i (z) dσiQCD ( p = z P.78) are real dimensionless functions which are measurable experimentally. Structure functions Fi defined by (8.81) and express the structure functions F1 and F2 in terms of the tensor Wµν as follows:  1  g µν Wµν = −(n − 1)F1 + F2   2x (8. (8. k ) (8. We get finally  1 Pµ Pν µν  F1 = −g + 2x Wµν   n−2 P ·q (8.75). Relations (8. From the expression (8. where p = z P. F2 = P · qW2 /m 2 (8. (8. remember that hadrons consist of quarks with different flavours and of gluons. We assume that partons in a nucleon have some momentum distributions f i (z) dz.84) where dσiQCD is the cross section for the lepton scattering on a parton of type ‘i’. In the actual physical applications of (8.83)  Pµ Pν 1   F2 = 2F1 + 4x Wµν x P ·q We want to calculate the structure functions Fi in perturbative QCD under the parton model assumption (8. Eq. of course.84) we must. k.74)–(8.70). Calculating the structure functions F1 and F2 we also neglect the nucleon mass as we are interested in the kinematical m 2 .

93) (8.89) ¯ f i (z) dσiQCD = [ 4 (u + u) + 1 (d + d) + 1 (s + s )] dσ q + G(z) dσ G ¯ ¯ 9 9 9 i (8.92) . 1 0 1 0 [sp (z) − sp (z)] dz = 0 ¯ (strangeness) (charge) (8. For instance. u ↔ d) gives u p (z) = dn (z) ≡ u(z) dp (z) = u n (z) ≡ d(z) sp (z) = sn (z) ≡ s(z) G p (z) = G n (z) ≡ G(z) and then for protons (8. There are several obvious constraints on the q N (z) determined by the quantum numbers of the nucleon. Once the squared electric charges of quarks are factorized out the dσ q is flavour-independent (and averaged over colours of the initial and summed over colours of the final partons).87) (8. Isospin symmetry (p ↔ n. q N = q N (z) denotes the momentum distribution of quark q in hadron N and G N (z) denotes the gluon distribution in hadron N .8.90) and for neutrons ¯ f i (z) dσiQCD = [ 4 (d + d) + 1 (u + u) + 1 (s + s )] dσ q + G(z) dσ G ¯ ¯ 9 9 9 i (8.5 Deep e–n scattering 295 in the case of the neutron as a target: ¯ f i (z) dσiQCD = [ 4 (u n + u n ) + 1 (dn + dn ) + 1 (sn + sn )] dσ q + G n (z) dσ G ¯ ¯ 9 9 9 i (8.86) where the numerical coefficients are the squared electric charges of quarks.91) It is convenient to decompose the quark distribution functions into valence quark and sea quark distributions q(z) = qv (z) + qs (z) where we assume ¯ u s = u s = ds = ds = ss = ss ≡ S(z) ¯ ¯ (8.88) ¯ { 2 [u p (z) − u p (z)] − 1 [dp (z) − dp (z)]} dz = 1 ¯ 3 3              etc. The dσ G vanishes in the zeroth order in the strong coupling constant α but contributes in higher orders.

77) and remembering about the normalizing factor 1/z P0 in dσ q we can express the q hadron structure functions in terms of F1.79) and (8.99) where F(y. for example.98) is the Bjorken variable defined for electron–quark scattering.84). q 2 ) = dz 1 [u v (z) − dv (z)] dσ q (z P. q 2 ).94) and for neutrons f i (z) dσiQCD = [ 4 dv + 1 u v + 4 S(z)] dσ q + G(z) dσ G 9 9 3 i (8. In the following we calculate dσ q in the zeroth and first orders in the strong coupling α and study flavour non-singlet hadron structure functions. We can define the quark structure functions for the elementary subprocess again by (8.3. In the Born approximation the quark structure functions are determined by elastic electron–quark scattering. 8. dσ ep (x. q 2 ) pµ − q µ 2 p·q q q y = −q 2 /2 p · q = x/z (8. (8.100) We can also use relation (8.105) to calculate the cut diagram shown in Fig. q 2 ) q2 1 p·q p·q q + pν − qν 2 F2 (y. .2 (y.296 8 Quantum chromodynamics Then for protons f i (z) dσiQCD = [ 4 u v + 1 dv + 4 S(z)] dσ q + G(z) dσ G 9 9 3 i (8.81) q Wµν = − gµν − qµ qν q F1 (y. q 2 ) or (1/y)F2 (y. q 2 ) = 0 1 (dz/z) i f i (z)F q (x/z.95) One also often talks about flavour non-singlet cross sections sensitive only to some combinations of the valence quark distributions. q 2 ) (8. Using (8. We have q 4π Wµν = dn−1 p1 (2π )n δ (n) ( p + q − p1 ) 1 Tr[ p γµ p1 γν ] / / 2 0 (2π)n−1 2 p1 (8.96) ¯ and flavour singlet cross sections sensitive to the distributions 1 (qs + qs ) = S(z) 2 and G(z). q 2 ) stands for F1 (y.97) where (8. q 2 ) as follows: F had (x. q 2 ) 3 (8. q 2 ) − dσ en (x.

This will no longer be the case when we include radiative corrections. Indeed. We arrive at a very interesting conclusion: it follows from (8. 8. depend on y only. . In terms of the Bjorken variable x defined for the hadronic process we have 1 q x F x/z 2 z = 2F1 q x z =δ 1− x z (8. in terms of the total nucleon momentum P. which together with q = k − k.5 Deep e–n scattering 297 Fig.104) that in measuring the hadron structure functions as a function of x we effectively study the momentum distribution of the hadron constituents. is the kinematic parameter of the reaction.83) we easily get the quark structure functions Fi in the Born approximation q Wµν = − 1 4π (1/y)F2 = 2F1 = 2 p · qδ(2 pq + q 2 ) = δ(1 − y) q q (8. we know already that y = 1 for elastic scattering. For the hadronic structure functions we get finally had had (1/x)F2 (x.104) The Bjorken variable x in (8. Using (8.3. which in general can depend on y and q 2 . as before. q 2 ) = (dz/z) i f i (z)δ(1 − x/z) = i f i (x) (8.8.102) where y = −q 2 /2 p · q. It is clear from calculation of the elastic cross section (see (8. The Bjorken parameter is therefore determined by the specific kinematic configuration in our experiment.101) n n n (n) (we have used the identity 1 = [(d p1 )/(2π ) ](2π) δ ( p1 − p − q)) which indeed coincides with (8. We encounter here the so-called Bjorken scaling law: structure functions.104) is defined.104) holds up to corrections O( p⊥ /z P).103) where x = −q 2 /2P · q and p = z P.100)) that our result (8. In particular we have learned from experiment this way that half of the nucleon momentum is carried by gluons. q 2 ) = 2F1 (x. Then dn p1 2 1 (2π)n δ (n) ( p1 − p − q)(−2πi)δ+ ( p1 )(−i)2 2 Tr[γµ i p1 γν p ] / / (2π)n (8.100).

5. So the corrections are frame-dependent and the transparent physical interpretation (8. 8.298 8 Quantum chromodynamics Fig.4 and 8. We use the notation of cut diagrams. Fig. Figs. 8.5 show the diagrams for the virtual and real corrections. Deep inelastic quark structure functions in the first order in the strong coupling constant We now calculate the dσ q in the first order in α strong . where p⊥ is the typical transverse momentum of partons in the hadron. .4. p⊥ finite. 8.104) of the hadron structure functions and of the Bjorken variable x (for x = 0) is valid in a so-called infinite momentum frame in which P → ∞.

8.5 Deep e–n scattering


Fig. 8.6.

where Wµν = 1 4π (8.105)

The calculation is performed in the Feynman gauge. The presence for each diagram of the group theoretical factor CF = 3 should be kept in mind. It is omitted in most 4 of the following expressions. All the quantities are for lepton–quark scattering but the superscript ‘q’ will also be omitted. Here and in the following we shall be mainly concerned with the quantity σ =− 1 gµν Wµν 1 + 1ε 2 (n = 4 + ε) (8.106)

The term p µ p ν Wµν in (8.83) is much simpler to calculate. The ladder diagram in Fig. 8.6 gives the following contribution to the cross section σ σL = 1 αµ−ε 1 + 1ε 2 d
1 2 2

Tr[ p γα ( p − /)γµ/ γ µ ( p − /)γ α ] / / l k / l

1 ( p − l)2


(8.107) with d


dn l dn k 2πδ+ (l 2 ) 2π δ+ (k 2 )(2π)n δ (n) (q + p − l − k) (2π)n (2π )n

(contrary to the calculation of Section 5.5 we do not assume the gluon to be soft). The trace gives
1 2

Tr[· · ·] = 4(1 + 1 ε)2 W 2 [−( p − l)2 ] = 4(1 + 1 ε)2 2 2

1−x 2 Q [−( p − l)2 ] x (8.108)

where W = q + p, Q 2 = −q 2 and x = −q 2 /2 p · q is the Bjorken variable for the


8 Quantum chromodynamics

lepton–quark scattering. Using expression (5.90) for the two-body phase space in the ( p, q) cms we get the following: σL = α 1 8π × Q 2 1 − x ε/2 1 (1 + 1 ε)4 2 4π µ2 x (1 + 1 ε) 2 1−x 2 −1 dy [y(1 − y)]ε/2 Q x ( p − l)2


with y = 1 (1 − z) and z = cos θ, where θ is the polar angle between the three2 vectors l and p. We observe that in the ( p, q) cms 2 p · l = 2 p0l0 (1 − z) = p·W (1−z) = (Q 2 /x)y and the denominator in (8.107) gives a singularity at y = 0, i.e. for l p and any value of the momentum |l| allowed by the conservation law q + p = k +l (4|l|2 = −q 2 (1− x)/x in cms). This is the so-called mass singularity. We finally get σL = α 2π Q2 4πµ2

(1 + 1 ε) 2

( 1 ε) (1 + 1 ε) 1−x 2 2 (1 − x) 1 (1 + ε) x (1 + 2 ε)


(8.110) The pole in ε reflects the mass singularity regularized by working in n = 4 + ε dimensions. The behaviour near x = 1 is IR behaviour (|l| → 0 when x → 1). The cut ladder diagram in the Feynman gauge exhibits no IR singularity.† However, we expect that the full cross section for a single photon production is IR divergent. As we know from Section 5.5 in the properly defined physical cross section the IR divergence is cancelled by virtual corrections to the elastic scattering, i.e. Aδ(1− x) dx + σ (x, Q 2 ) dx is finite. In order to demonstrate this cancellation explicitly we write the cross section σ in terms of the so-called regularized distribution. For any function f (x) we define the distribution f + (x) as follows f + (x) = f (x) − δ(1 − x)
0 1

dy f (y)


i.e. we explicitly single out a possibly singular behaviour of the function f (x) as x → 1. We can then write σL = α 2π Q2 4πµ2

1 + 1 + 1 γ δ(1 − x) 2 ε 1−x (1 − x) x

2 + +1+γ ε


The contribution of the longitudinal term pµ pν W µν can also be easily included.
† Note the difference with Section 5.5: now m = 0.

8.5 Deep e–n scattering


Fig. 8.7.

The ladder diagram gives 1 F2 x


α 3 α 3x = [ δ(1 − x) + (3x)+ ] 2π 2π 2


and the contribution of the rest of the diagrams to pµ pν W µν vanishes. We proceed to calculate cut self-energy corrections. The contribution of the diagram in Fig. 8.7 reads σ

1 αµ−ε = 1 1 + 2ε



1 t2

2 1 2

Tr[ p γ µ/γ α / γα /γµ ] / t k t


As before, d 2 is the two-body (l, k) phase space and we work in n = 4 + ε dimensions. The trace gives 2(1 + 1 ε)2 2(2 p · l)2k · l = 2(1 + 1 ε)2 2(t 2 )2 y/(1 − x), 2 2 where y = 1 (1 + cos θ) and θ is the angle between vectors p and k in the (k, l) cm 2 frame. We see that the propagators are cancelled and we get finally σ

=α =

Q2 1 − x 1 8π 4πµ2 x
2 ε/2


4(1 + 1 ε) 2 1−x x

1 (1 + 1 ε) 2

1 0

dy [y(1 − y)]ε/2

y 1−x (8.115)

Q α 2π 4πµ2

(1 + 1 ε) 2

1 1−x

(2 + 1 ε) 2 (3 + 1 ε) 2

In the present case the integration over y is finite: there is no mass singularity. The result is, however, divergent for x → 1 (IR divergence). Using the regularized distribution we get: σ


Q2 α 2π 4πµ2


1 1 1 1 + 2 (γ − 1) δ(1 − x) + ε 2 1−x


Finally we consider the diagrams in Fig. 8.8 and the corresponding cross section σ

= 2αµ−ε

1 1 + 1ε 2


2 2 t (p

1 1 Tr[ p γ µ/γ α / γµ ( p − /)γα ] / t k / l − l)2 2



8 Quantum chromodynamics

Fig. 8.8.

where d 2 is as before. Using γ µ / bcγµ = −2cb/ − ε/ bc and γ µ / bγµ = 4ab + a // a // a/ //a ε/ b we evaluate the trace and get a/
1 2

Tr[· · ·] = −{4(2 p · k)(2 p · t − 2t · l) + 1 ε4(2 p · t)(2k · p − 2k · l) 2 /t k / l + ε Tr[ p//( p − /)] + 1 ε2 Tr[ p// ( p − /)]} /kt / l 2 (8.118)

The last two terms do not contribute (one is exactly zero and the second gives ε2 (−8)( p · l)(k · l), i.e. both propagators are cancelled and there is no divergence to cancel ε 2 ). In the (k, l) cm frame we have 2 p · t = Q 2 /x, 2 p · k = 2 p · (t − l) = Q 2 /x + ( p − l)2 , 2t · l = 2k · l = (Q 2 /x)(1 − x) and σ

= 2α

1 1 + 1ε 2




ε x 1 x (1 + 1 ε) − 1+ 2 1−x y 1−x 2x


We see that the cut vertex has (in the Feynman gauge) both IR and mass singularity. Performing the integration over d 2 we obtain the following σ =2 + α 2π Q2 4π µ2


(1 + 1 ε) 1 2 (1 + ε) 1 + ε
ε/2 +

2 ε



π2 − 1 δ(1 − x) 6 (8.120)

2 −1 ε

x 1−x 1−x x

Final result for the quark structure functions Virtual corrections to the considered cross section can be easily calculated using the results of Chapter 5 for the electron self-energy correction ( p 2 ) and the vertex correction (q 2 ). For instance, the first two diagrams in Fig. 8.4 give together (in

8.5 Deep e–n scattering


n = 4 + ε dimensions) σ =− gµν W µν dn k −1 1 (2π )n δ (n) (k − p − q)2πδ+ (k 2 )(−i)2 = 1 1 4π (2π )n 1 + 2ε 1 + 2ε i × 1 Tr p γ µ/ γµ [−i ( p 2 = 0) p ] / k / 2 p / α 1 = ( p 2 = 0)δ(1 − x) = − δ(1 − x) (8.121) 2π ε

The same result holds for the next two diagrams of Fig. 8.4. Also, with ( p 2 ) changed into (q 2 ), each of the vertex correction diagrams can be calculated in a similar way. We are now ready to collect all the results (n = 4 + ε everywhere): Real σL = α 2π Q2 4πµ2

1 + 1 + 1 γ δ(1 − x) + 2 ε
ε/2 +

2 +1+γ ε

× (1 − x) σ σ = =2 α 2π α 2π

1−x x


Q 4πµ2 Q 4πµ2


1 1 1 1 + (γ − 1) δ(1 − x) + ε 2 2 1−x (1 + (1 + ε)
ε/2 + 1 ε) 2



2 ε



π −1 6

2 + −1 ε 1 2 F x Virtual

x 1−x 1−x x


α 3 [ δ(1 − x) + (3x)+ ] 2π 2

  1  δ(1 − x)   1+ε                    (8.122)


                         

α 1 σ =2 − δ(1 − x) 2π ε σ =2 × α 1 1 (1 + 1 ε) Q 2 2 − 2π ε ε (1 + ε) 4π µ2 2 ε
2 ε/2


Q2 4πµ2



π2 6

δ(1 − x)

         1 (1 + 2 ε) (−1)  (1 + ε) 1 + ε          (8.123)

The first two terms in σ correspond to the renormalized


(Section 5.3) written


8 Quantum chromodynamics

in n = 4+ε dimensions. The final result for (1/x)F2 in the lepton–quark scattering reads 1 α F2 = CF x 2π 1 1 + 2 1−x 2 +γ ε + 3x

1 + x2 1−x Q2 4π µ2


1 + x2 1−x ln − 1 + 2(1 − x) 1−x x + δ(1 − x) + O(α 2 ) (8.124)

We now add several important comments on the result obtained to clarify its physical sense. As we already know from Section 5.5, the IR singularities cancel between real and virtual corrections and they are not present in (8.124). However, this is not the case with mass singularities (due to the decay of a massless particle into two parallelly moving massless particles) which explicitly reflect themselves in the final result as poles in ε. We certainly have to cut them off by some physical parameter before inserting result (8.124) into (8.99) to compare our calculation with experiment. The trick is to ‘factorize out’ the mass singularities by writing the result (8.124) in the limit ε → 0 as (see (8.71)) 1 F2 (x, Q 2 ) = x
1 0

dz z

x 1 , Q2, Q2 F2 (z, Q 2 ) + O(α 2 ) 0 0 z z Q2 α ln 2 2π Q0 1 + t2 1−t


where Q 2 is an arbitrary parameter and from (8.124) 0 (t, Q 2 , Q 2 ) = δ(1 − t) + CF 0

+ O(α)


where the O(α) terms are finite when Q 2 → ∞. The meaning of (8.125) is clear: the structure function for any Q 2 is expressed in terms of the structure function for Q 2 and of the theoretically calculated function (x/z, Q 2 , Q 2 ) free of mass 0 0 singularity. Hadron structure functions; probabilistic interpretation To calculate hadronic structure functions we use the convolution (8.99) and get 1 had F (x, Q 2 ) = x 2
1 x

z x dz f (z) F2 , Q 2 z x z

1 = x


(t, Q 2 , Q 2 ) 0

1 x/t

dz f (z)F2

x , Q2 0 zt


where x = Q 2 /2Pq is the Bjorken variable for the lepton–hadron scattering. Consequently 1 had F (x, Q 2 ) = x 2
1 x

dz z

x 1 had , Q2, Q2 F (z, Q 2 ) 0 0 z z 2


8.5 Deep e–n scattering


and interestingly enough we again recover the result (8.125). The effect of the had intrinsic quark distribution in the nucleon is included in F2 (z, Q 2 ). We do not 0 have to know it theoretically to compare the theory with experiment as far as Q 2 evolution of the structure functions is concerned. By introducing an intermediate scale Q 2 we escape two problems: the intrinsic 0 parton distribution which originates from non-perturbative effects need not be known theoretically and the perturbative mass singularities are not present in had our final formula. They have been absorbed into F2 (z, Q 2 ), which we take 0 as the experimentally measured structure function. Our theoretical prediction is had the evolution function (x/z, Q 2 , Q 2 ) and we do not worry that F2 (z, Q 2 ) is 0 0 actually singular in our calculation since this is due to our ignorance of how to include correctly the soft binding effects. It is also worth observing that the leading Q 2 dependence of the result (8.124) for the quark structure functions and therefore also for the hadron structure functions (8.128) can be interpreted in a very transparent probabilistic parton-like way. To this end let us formally rewrite (8.124) as follows 1 F2 (x, Q 2 ) = x
1 0

dz 1 x F2 (z, Q 2 )δ 1 − z z z


Formally, we obtain an expression for F2 which is the same as (8.104) for the structure function of a composite object consisting of a set of partons which scatter elastically at large Q 2 . We are tempted to interpret the quantity (1/z)F(z, Q 2 ) as the average number of partons, with momentum in the interval zp, (z + dz) p, seen in the initial quark with momentum p by a probe (electron) in the process of deep inelastic scattering with momentum transfer Q 2 . The theory predicts its evolution with Q 2 . Let us assume for the moment that our interpretation is sensible. had The hadronic structure function (1/x)F2 (x, Q 2 ) is then a convolution of the intrinsic parton momentum distribution in a hadron (Q 2 -independent) and the parton momentum distribution in a parton when probed with large Q 2 . Technically, the second effect is present in perturbation theory, whereas the first one has its origin in non-perturbative binding effects. Is our interpretation of (8.129) sensible? It is for the leading Q 2 dependence. This point will become clearer in the next section when we repeat our calculation in a so-called axial gauge. The reason is that interference diagrams with a gluon emitted or absorbed by the final quark contribute to the final result in the present calculation. It turns out, however, that with the proper choice of gauge the leading effect comes from the ladder diagram and so our interpretation is possible. It applies, however, only to the leading Q 2 dependence because to interpret (8.129) in terms of the elastic scattering of a set of partons with momentum distribution (1/z)F2 (z, Q 2 ) we have to worry, as before, about corrections O( p⊥ /zp). But


8 Quantum chromodynamics

now the transverse momentum of a parton is limited only by kinematics and can be as large as O((Q 2 )1/2 ). For the probabilistic interpretation we have not only to stay in an infinite momentum frame but also in a limited region of phase space ( p⊥ small). This we can do by writing ln(Q 2 /Q 2 ) = ln(ε Q 2 /Q 2 ) + ln(Q 2 /ε Q 2 ), 0 0 where ε is some fixed arbitrarily small number. The first logarithm comes from the region where p⊥ (ε Q 2 )1/2 is small and at the same time it gives the leading Q 2 dependence correctly (for Q 2 → ∞). The function P(z) = 1 + z2 1−z (8.130)

is called the Altarelli–Parisi function. One often considers the Q 2 evolution of moments of the structure function, e.g. M N (Q 2 ) =
0 1

dx x N −1 F(x, Q 2 )


where again F(x) = F1 (x) or F(x) = (1/x)F2 (x). From (8.128) and (8.126) one gets MN =
0 1

dx x N −1





(t, Q 2 , Q 2 )δ(x − zt)(1/z)F2 (z, Q 2 ) 0 0 (8.132) α Q2 PN ln 2π Q2 0

or in the leading Q 2 dependence approximation M N (Q 2 ) = M N (Q 2 ) CF 0 +1 (8.133)

where PN is the N th moment of the Altarelli–Parisi function: PN =
0 1

dx x N −1

1 + x2 1−x





1 + x 2 N −1 − 1) (x 1−x


8.6 Light-cone variables, light-like gauge In this section we describe briefly the calculation of dσ q given by the diagrams in Figs. 8.4 and 8.5 using the light-cone parametrization of the momenta and the light-like gauge. To be specific let us consider first the ladder diagram in Fig. 8.6. The gluon propagator in the n 2 = 0 gauge is −i gµν − lµ n ν + lν n µ δαβ l ·n (8.135)

8.6 Light-cone variables, light-like gauge


and we parametrize the momenta as follows p = (P + p 2 /4P, O⊥ , P − p 2 /4P)
2 2 l = (z P + l⊥ /4z P, l⊥ , z P − l⊥ /4z P) 2 2 t = (y P + (t 2 + t⊥ )/4y P, t⊥ , y P − (t 2 + t⊥ )/4y P)

                   (8.136)

q = (η P + q 2 /4η P, O⊥ , η P − q 2 /4η P) y = 1 − z,
2 t 2 z = −l⊥

For the gauge four-vector n we take n = ( p · n/2P, O, − p · n/2P) (8.137)

We can interpret (8.136) and (8.137) noticing that P is an arbitrary boost parameter defining the frame (invariant quantities are P-independent) so that we may start, for example, from the p rest frame (4P 2 = p 2 ) and make an infinite boost P → ∞ in the direction opposite to n. We observe that:
(i) l · n/ p · n = z; (ii) η = −x for p 2 = 0 because 2 p · q = q 2 /η;
2 2 (iii) 1 − z = x + O(l⊥ (−q 2 )) because ( p − l)2 = −l⊥ /z and ( p − l + q)2 = 0.

The two-body phase space d


dn l dn k 2πδ+ (l 2 ) 2π δ+ (k 2 )(2π )n δ (n) (q + p − l − k) (2π)n (2π)n


can be written as d From
2 δ+ (( p − t)2 ) = δ(−(1 − y)t 2 /y − t⊥ /y), 2


dn t δ+ (( p − t)2 )δ+ ((q + t)2 ) (2π)n−2


( p − t)0 > 0

2 δ+ ((q + t)2 ) = δ((y − x)(t 2 /y − q 2 /x) − xt⊥ /y), (q + t)0 > 0


one concludes that x <y<1 y − x q2 t2 = , 1−x x

q2 < t2 < 0 x



8 Quantum chromodynamics

Furthermore, dn t 1 dy n−2 = dt 2 d t⊥ n−2 n−2 (2π) (2π ) 2|y| and dn−2 t⊥ =
n−3 t⊥ dt⊥ d n−2



π (n−2)/2 1 (n − 2) 2

2 2 dt⊥ (t⊥ )(n−4)/2


2 So finally, integrating over d(t⊥ /y) by means of the δ(( p − t)2 ), we get (n = 4 + ε)



1 8π 1 8π

1 (1 + 1 ε) 2



dt 2
q 2 /x x

dy (1 − y)|t 2 | y y 4π


δ t 2 (1 − x) −

(y − x)q 2 x (8.144) (8.145)


1 1 1 (1 + 2 ε) 1 − x



(1 − y)(y − x) −q 2 x(1 − x) 4π

For the Feynman part of the ladder diagram, using (8.107) and (8.108) together with (8.144) we recover the previous result (8.110). The axial part of the ladder diagram
A σL = CF

1 αµ−ε 1 + 1ε 2 d
1 2 2


1 1 1 1 / // l Tr p/ γµ (/ + / )γ µ n + p n γµ (/ + / )γ µ / /l t q t q / t / t / t / t

1 l ·n (8.146)

can be calculated using the expansion
2 t 2 + t⊥ n / /= yp + t / + /⊥ t 2y p · n

         (8.147)

/ −q 2 n / = −x p + q / x 2p · n and (8.144). After some effort one gets
A σL = CF

α 2π

−q 2 4π µ2 2 +γ ε



(1 + 1 ε) 2 (1 + ε)

2 ε


2 π2 + ε 6 (8.148)

×δ(1 − x)

2x 1−x 1−x x


8.6 Light-cone variables, light-like gauge


We see that
F σL


A σL

= CF

α 2π

2 +γ ε

1 + x2 1−x


+ A(ε)δ(1 − x) + · · ·

−q 2 ε/2 4π µ2 (8.149)

where the neglected terms do not have a pole in ε and are polynomials in x and logarithms in x and 1 − x. Thus in the axial gauge the ladder diagram by itself gives the leading logarithmic contribution (8.126) to the quark structure function. (Remember that we are interested in the finite x = 0 region so ln x and ln(1 − x) can be dropped.) We expect therefore that the contribution from the remaining diagrams with real emissions is non-leading. By explicit evaluation of the axial parts of these amplitudes one gets: σ Ac = so that
F A F σL + σL + σ Fc + σ Ac + σ Fc + σ Ac = σL + σ Fc + σ Fc

1 2 F x


= 0,

A σ Ac = −σL


and (σ Fc + σ Ac ) does not contribute to the leading logarithm, i.e. it is free of mass singularities. Actually, one can support this conclusion by the following general argument. The leading logarithm reflecting the presence of mass singularity obviously comes from the integration over t 2 in the region εq 2 < t 2 < 0, i.e. from the region where t and l are almost collinear with p: t ∼ yp. Part of the = Dirac algebra involves 2y lµ p /γµ p ∼ y p γµ p = y2 pµ p = t /= / / / / 1−y


and since in the axial gauge the on-shell gluons have only physical polarization the contraction with the gluon tensor dµν (l) = λ=1,2 ε ∗µ (λ, l)ε ν (λ, l), l 2 = 0 gives lµ dµν (l) = 0 (8.152)

Actually, one can check that the vertex vanishes linearly in θ, where θ is the gluon emission angle. So the ladder diagram gives d

(1/t 2 )2 θ 2 ∼

dθ θθ 2 /(θ 2 )2 ∼ ln θ



8 Quantum chromodynamics

whereas the cut-vertex diagram gives d

(1/t 2 )θ ∼

dθ θ(θ/θ 2 )


and hence no logarithm. The poles in ε in the coefficient of the δ(1 − x) are cancelled by the virtual corrections. Correspondingly the situation with real emissions in the axial gauge, mass singularity is present only in the quark self-energy correction. It is instructive to calculate the virtual corrections in the light-like gauge and using the light-cone parametrization of the momenta. This we leave as an exercise for the reader limiting ourselves to the following two remarks. Take, for instance, the self-energy correction = −i where can be decomposed as follows = Ap + B / with n / 2p · n pn p // / A + B = 1 Tr 2 2p · n A = 1 Tr 2 The full propagator reads G( p 2 ) = i p / n / B − 1 − (A + B) p 2 1 − A 2 p · n ip / = [1 + (A + B) + · · ·] 2 + · · · p i p− / = (8.156) 1 p2 (8.157) p2 n / 2p · n (8.155)


Apart from IR and mass singularities is also UV divergent, i.e. contains poles in εUV where n = 4 − εUV . The necessary counterterms are of the form a p + bn / // The term proportional to n reflects the pathology of the n 2 = 0 gauge because / d4l 1 (n · l)l 2 ( p − l)2

9. However. Our second remark is that an explicit calculation of the virtual corrections using the light-cone parametrization of the momenta is called the Sudakov technique. 8.161) where σL (yp. Since in the leading logarithmic approximation.160) and the integral over the semi-circle.159) 2 2 can be evaluated by integrating over dl⊥ first. q) (8. q ) = CF 2π 2 q2 p2 dt 2 t2 1 x dy 1 + y 2 y 1−y + σL (yp. we can replace 2 the denominator by δ(z(1 − z)( p 2 − l 2 /z) − l⊥ ).9. q) = δ(1 − x/y) † In the Feynman parametrization such integrals are discussed in Appendix B. its probabilistic interpretation is quite natural.6 Light-cone variables. in order-by-order calculation one can check explicitly that such terms cancel out in the full vertex function. 8. In fact this result has the following structure α σL ( p. light-like gauge 311 Fig.8. . only the ladder diagram contributes to the final result. see Fig. Going to the complex l⊥ -plane. It consists of noting that the integral dn l 1 1 ∼ (2π)n l 2 + iε ( p − l)2 + iε dl 2 dz 2 2 (n−4)/2 dl (l ) l 2 2|z| ⊥ ⊥ z × 2 − l 2 /z) − l 2 + iε z(1 − z)( p ⊥ (8. should be convergent on dimensional grounds. which is at infinity. since there is a pole at 2 l⊥ = z(1 − z)( p 2 − l 2 /z) + iε (8.† and it makes the general proof of the renormalizability in this gauge difficult. vanishes.

For the regions x = 0 and x = 1. Marchesini & Mueller (1982).129) and Pq→q (y) 1 + y2 1−y + can be interpreted as the variation of the probability density per unit t 2 of finding a quark in a quark with fraction y of its momentum. (8. This coincides with (8. the summation of the so-called leading logarithms α n lnn (q 2 / p 2 ) and also of the next-to-leading α n lnn−1 (q 2 / p 2 ) terms has been explicitly performed and compared.129). Llewellyn Smith 1978. is still valid. the next-to-leading one and so on (Gribov & Lipatov 1972. Of course. Dyakonov & Troyan (1980).e. whenever possible. Another one.7 Beyond the one-loop approximation Since αstrong ∼ O(1) the relevance of perturbative QCD to the description of the actual experimental data often relies on our ability to go beyond the one-loop approximation. In the context of QCD pioneered by Furma´ ski. Further very interesting and more recent progress consists of extending the QCD perturbative techniques to processes where there are two types of large logarithm (for example. In the leading logarithm approximation it is concisely formulated by means of the Altarelli–Parisi equations (Altarelli & Parisi 1977). Parisi & Petronzio (1979)) and to the so-called doubly-logarithmic region of the structure and fragmentation n functions (x ∼ 0). Dokshitzer. It is very interesting that these results can also be given a probabilistic interpretation. Ciafaloni. In addition we must also include in considerations the singlet and the gluon structure and fragmentation functions. Let us only briefly indicate the techniques that have been developed to include higher order effects in αstrong . Dokshitzer. applicable to a wider class of problems. 8. One approach is based on the OPE and on the RGE for the Wilson coefficient functions (Chapter 7). consists of summing infinite sets of Feynman diagrams by cleverly picking out the leading contribution. Dyakonov & Troyan 1980). . this extension is. i. important for studying new particle production in hard collisions in very high energy accelerators. neglecting ln x and ln(1 − x) terms as compared to ln(q 2 / p 2 ) terms. for instance. our previous discussion concerning the hadron structure functions.312 8 Quantum chromodynamics and we have chosen to regularize the mass singularity with p 2 = 0. This vast area of technical and phenomenological activity goes far beyond the scope of the present book and can by itself be a subject of an extended monograph.127)–(8. Petronzio & = Pokorski (1979) and Bassetto. Ciafaloni & Marchesini (1980) and fully clarified by Bassetto. with the OPE.

The dominant region of integration is where | p2 | 2 |t1 | 2 |t2 | ··· 2 |tn | |q 2 | Furthermore. It turns out that the dominant diagrams in the axial gauges are the generalized ladder diagrams. i.7 Beyond the one-loop approximation 313 Fig.162) we can easily calculate the moments of (1/x)F2 .8.161) and reads n σL = |q 2 | p2 2 2 dtn α(tn ) CF 2 tn 2π 2 |tn | 2 dtn−1 2 tn−1 CF p2 2 α(tn−1 ) ··· 2π × y1 2 2 dt1 α(t1 ) 1 dy2 y2 CF dy1 Pq→q (y1 ) Pq→q ··· 2 2π x y1 y1 t1 x p2 yn−2 yn−1 1 dyn−1 dyn yn−1 yn Pq→q Pq→q × δ(yn − x) yn−2 yn−2 yn−1 yn−1 yn x x (8. 8. ladder diagrams with vertex and self-energy insertions.10. Fig. it can be shown that the generalized ladder can be replaced by a ladder with the coupling constant α(ti2 ) at each vertex. 8. Thus in the leading logarithm approximation the cross section is a generalization of (8. First we note that CF 2π n |q 2 | p2 2 dtn 2 α(tn ) 2 tn 2 dtn 2 α(tn ) 2 tn × |q 2 | p2 1 2 α(tn−1 ) · · · → (CF /2)n 2 n! tn−1 |q 2 | dt 2 ln(−q 2 / 1 n−1 2 α(tn−1 ) · · · = ln 2 n! ln(− p 2 / tn−1 p2 p2 2 |tn | 2 dtn−1 2 ) 2) n − CF n b1 (8.163) .e.162) 2 |t2 | Using (8.10. Let us be a little bit more explicit concerning the summation of the leading logarithm terms α n lnn (q 2 / p 2 ) generated by mass singularities in the quark structure function.

.e. In QCD it has been demonstrated that IR divergences cancel for processes with none or one coloured particle in the initial state (for a review see.. Of course. for example. Changing variables 3 z 2 = y2 /y1 . for the moments of the hadron structure function 0 and it coincides with (7. .314 8 Quantum chromodynamics 2 where α(t 2 ) = −2π/[b1 ln(t 2 / z 1 = y1 . .165) M N (Q 2 ) ln(Q 2 / = M N (Q 2 ) ln(Q 2 / 0 0 where 0 −γ N = CF 1 0 ) 2) 2 0 γ N /b1 . z n ) = z1 . 11 2 − 1 n f . Comments on the IR problem in QCD The Bloch–Nordsieck theorem in QED states that IR divergences cancel for physical processes. . Q 2 = −q 2 (8. q 2 ) x 1 ln(−q 2 / 2 ) ln n! ln(− p 2 / 2 ) 2 − CF b1 1 0 dz z N −1 Pq→q (z) n = and ln(−q 2 / ln(− p 2 / ) 2) −(CF /b1 ) 1 N −1 P q→q (z) 0 dz z (8. in the process qq → l+ l− X these divergences arise at the two-loop level.. However. . for processes with an arbitrary number of undetectable soft photons.166) dz z N −1 1 + z2 1−z + =− N −2 1 2 2 +4 1− 3 N (N + 1) j j=2 (8. zn dz i z iN −1 Pq→q (z i ) (8.124) obtained using the OPE. z n = yn 1 i 0 dx x N 0 i dz i Pq→q (z i ) 1 δ(x − z 1 . −b1 = .. . . i.99) in the coupling constant.167) The same result holds. Sachrajda (1983b)). it has been shown by means of an explicit example that such a cancellation does not occur for processes with two coloured particles in the initial state even if an average over different colour states is taken.164) Therefore. we get 1 )]. We see that γ N are the g 2 coefficients in the expansion of the anomalous dimensions of the operators in (7. For instance. of course. z n = yn /yn−1 z 1 z 2 . finally MN = 0 1 = n 1 dx x N F2 (x.

with a continuously deformable to it. In QED the cancellation of IR singularities occurs separately in the final state and there is no need to invoke the KLN theorem. U (x) belongs to N = 1 class. Check that the winding number can be written as ∞ −i dα(x) 1 dx (8. Problems 8. transition probabilities are finite when the sum over all degenerate states (final and initial) is taken. This is true order-by-order in perturbation theory in bare quantities or if the minimal subtraction renormalization is used (to avoid IR or mass singularities in the renormalization constants). However.Problems 315 in physical processes coloured partons never appear as free on-shell particles and these divergences are regulated by a mass-scale characteristic of the hadronic wave function which is of the order of the inverse hadronic radius. Fortunately enough it has been shown that such large IR logarithms are suppressed by at least one power of Q 2 and asymptotically the factorization property is still maintained. What is U∞ ? Can such a transformation be built up from infinitesimal transformations a (x) → δ a (x) for all x? . then a (x) a (∞) a a a σ )(Aµ + ∂µ ) exp( 1 i 2 a a σ ) ≡ 0. In non-abelian gauge theories this does not happen but the initial state is determined by the non-perturbative confinement effects and the relevance of the KLN theorem is doubtful. (b) For S 3 → SU (2) the transformation Aµ → exp(− 1 i 2 with a (x) ≡ 0 or N = 0. to be unclear. The KLN theorem is a mathematical theorem saying that. spoil the factorization property discussed in Section 8.4 which states that all long distance effects can be absorbed into the quark and gluon distribution functions and is crucial for the sensible use of perturbative QCD. The physical meaning of this theorem seems. residual large logarithms remain and could. belongs to continuously deformable to = 0. In the context of the IR problem in QCD it is worth recalling the Kinoshita– Lee–Nauenberg (KLN) theorem (Kinoshita 1962.168) N= 2π −∞ α(x) dx Check that N = 1 for α(x) = exp[iπ x/(x 2 + a 2 )1/2 ]. Lee & Nauenberg 1964). as a consequence of unitarity. Clearly. in principle. Check that with (x) = 2π x a /(x2 + a 2 )1/2 a (x) or. however.1 (a) Consider mappings of the real axis with the end-points identified into a set of unimodular complex numbers z = exp[iα(x)].

the gauge-fixing term (8. d3 x K 0 i d3 x Dab b · i a .34) and complete the calculation of the β-function in Yang–Mills theories in the one-loop approximation (Abbott 1982). Aµ ] and δ Aµ given by (8.22).3 Derive the Feynman rules for calculating the 1PI Green’s functions generated by ˜ [0.) a 8. Aµ ] defined in Section 8. show that exp(iQ G )W (A) exp(−iQ G )W (A) contradicting the correct result exp(iQ G )W (A) exp(−iQ G ) = W (U −1 AU − U −1∇ U ) = W (A) + N (Use δW (A)/δ Ai = (1/8π 2 )Bia .24) and the Faddeev–Popov term obtained from (3.15) and (3. Prove (8.64).316 8 Quantum chromodynamics 8. Use the shifted action S[Aµ + Aµ ].2.39) with F α → ˜ F α [Aµ .2 Show that gauge transformations which belong to the homotopy class N = 0 cannot be built up by iterating the infinitesimal gauge transformations. For this assume that a ‘large’ (N = 0) gauge transformation U = exp(− 1 i 2 where a (x) a a σ ) approaches a non-zero limit at spatial infinity is given by the operator exp(iQ G )A exp(−iQ G ) = U −1 AU − U −1∇ U where exp(iQ G ) = exp −i Applying it to W (A) = where K 0 is given by (8.

spontaneous symmetry breaking Chiral symmetry plays an important role in the dynamics of fermions coupled to gauge fields. its importance for particle physics originates in the idea that SU (2)L × SU (2)R is an approximate symmetry of the strong interactions. This chapter serves as an introduction to the subject of chiral symmetry and to techniques used in its exploration. 9. = (x) (9. realized as spontaneously broken symmetry with pions being Goldstone bosons in the symmetry limit. Historically. In fact. Consider the general massless Dirac lagrangian of fermion fields ri coupled to gauge fields Aα µ L= r n(r ) i ¯ ri iγ µ Dµ 317 ri . the masses of the light quarks are small in a sense which will be specified later on and we shall assume in the following that the theory with the quark masses neglected is a correct first approximation.1) . secondly. to find the dynamical theory of the fermion mass matrix which might eventually lead to the true theory of fundamental interactions. Much of the progress of particle physics since 1960 is related to understanding the phenomenological aspects of spontaneous and explicit chiral symmetry breaking.9 Chiral symmetry. The successes of gauge theories in describing the fundamental interactions further stimulate interest in chiral symmetry and this is for at least two reasons: one may hope firstly to understand the underlying mechanism of spontaneous chiral symmetry breaking in the strong interactions and in gauge theories in general and.1 Chiral symmetry of the QCD lagrangian We begin our discussion by considering the QCD lagrangian with two quarks u and d in the theoretical limit in which the quark mass parameters in the lagrangian are put equal to zero.

3 and σ a are Pauli matrices for a = 1.5) conservation. 2 a = 0. For QCD with two flavours r ≡ 3. Indeed. the massless theory (9. 2 and i (we now omit the subscript r ) denotes a vector in the colour space.2). since {γ5 .7) . the matrices trα represent these generators in the representation r of the gauge group to which the fermions are assigned and the index i runs over all n(r ) flavours belonging to the same representation r . 3 (9. In this case lagrangian (9. 1. involving γ5 1 → exp(−iα a 2 σ a γ5 ) ¯ → ¯ exp(−iα a 1 σ a γ5 ) 2 (9.1) is invariant under the general unitary transformation in the two. 3 and σ 0 = 1. 2.3) where a = 0.6) where a = 0.2) also has another global symmetry. 2.6) is introduced into lagrangian (9. The axial currents read 1 Aa = ¯ γµ γ5 2 σ a µ (9. 2.4) and the SU (2) × U (1) invariance gives the isospin and the baryon number Qa = d3 x ¯ (x)γ0 1 σ a (x).2) which is invariant under SU (2) × U (1) transformations 1 → exp(−iα a 2 σ a ) ¯ → ¯ exp(+iα a 1 σ a ) 2 (9. γµ } = 0 the exponentials again cancel when (9. 2. 1.dimensional flavour space. The corresponding conserved currents are a 1 Vµ = ¯ γµ 2 σ a (9. i = 1.318 9 Chiral symmetry. However. Introducing = L= ¯ 1 2 we can write / iD / iD (9. spontaneous symmetry breaking where Dµ = ∂µ + ig Aα trα µ and the index α runs over the generators of the gauge group. 1. 3.

2) we introduce some useful notation. This is discussed in Appendix A.R . c = 1.10) PL PR = 0 / l + ¯ R iD 1 and the lagrangian (9. Q b ] = [Q a . 3  L R 5 5 2 2   a b abc c (9. .9. j are flavour indices and the unitary transformations UL and UR read a a UL = exp(−iαL TL ) (9. [Q a .10) has a clear interpretation in terms of the free particle solutions of the massless Dirac equation. Q a = 1 (Q a + Q a ).R ] = iε Q L. hence the name chiral R.8) and the equal-time canonical commutation relations between the fields)  Q a = 1 (Q a − Q a ).R R (for completeness: [Q a . Before we proceed to explore the chiral symmetry of the massless QCD lagrangian (9. 3 L. Q b ] = iεabc Q c . 2.5) and (9. An 5 5 5 5 important property of the generators Q a and Q a is that they transform into each L R other under the parity operation: P Q a P −1 = Q a . 2.1 Chiral symmetry of the QCD lagrangian 319 and the axial charges† Qa = 5 d3 x ¯ (x)γ0 γ5 1 σ a (x) 2 (9.5) form the chiral SUL (2) × SUR (2) × U (1) × UA (1) algebra (as can be seen using (9. ‡ Decomposition (9. This remark is relevant for the case of spontaneous symmetry breakdown. Q b ] = 0. their commutators exist provided we first commute under the space integrals and then do the integrations. a = 0.2) can then be rewritten as / l L = ¯ L iD 1 L R (9. 1. Q b ] = iεabc Q c ). 3    [Q a .8) together with the charges (9. Even so. 2. l Chiral symmetry of the lagrangian (9.R . b = 0.11) (1 is unit matrix in the flavour space). L + R = 1 (1 − γ5 ) 2 + 1 (1 + γ5 ) 2 = PL + PR (9. a. b. Any spinor field can be decomposed into ‘chiral’ fields‡ as follows = where 2 2 PL + PR = 1.13) a a UR = exp(−iαR TR ) † Charges may be formal in the sense that the integral over all space may not exist.9) [Q L.R transformations. a.11) means invariance under the transformations j i i i L → L = (UL ) j L (9.12) j i i i R → R = (UR ) j R where indices i. Q L. 1.L L.

spontaneous symmetry breaking a a with TL = T a ⊗ PL and TR = T a ⊗ PR . At this level we should also conclude that because of the U (1) × UA (1) invariance the fermion numbers of R and L (corresponding to U (1) ± UA (1) charges) are separately conserved.14) Then we can formulate our theory in terms of L and ˜ L . In many problems we work with one type of chiral field R or L only.16) and the transformation properties of the field . i.2 Hypothesis of spontaneous chiral symmetry breaking in strong interactions We have assumed that the massless QCD with its chiral SUL (2) × SUR (2) global symmetry is a good approximation to the real world. H ] = 0 (9. Indeed. We see that multiplet L ( R ) is a singlet with respect to SUR (2)(SUL (2)) transformations. 513): ‘a symmetry of the vacuum is a symmetry of the world’ or Q i |0 = 0 ⇒ [Q i . The matrices T form a representation of the corresponding generators Q in the space of fermion multiplets . in grand unification schemes) only L (or only R ) fields can be placed in irreducible multiples of the gauge group. Itzykson & Zuber (1980). p. for example. We refer the reader to Appendix A for further details on both subjects.15) We can invoke Coleman’s theorem (see.e. For the time being we will be concerned with the chiral SU (2) × SU (2) only. the left-handed fields can be obtained if we replace R by their charge conjugate partners ( R) C =( C )L = ˜ L df (9. The description in terms of. The problem of the UA (1) symmetry will be briefly discussed in Chapter 13. Now let us imagine in addition that the physical vacuum defined by the minimum of the expectation values of the hamiltonian 0|H |0 = H min is invariant under the chiral transformations. for example.L be seen by constructing them from the vacuum by means of the field operators and then using the properties (9.320 9 Chiral symmetry. For instance in ‘chiral’ gauge theories such that R and L transform differently under the gauge group (for example. Some useful properties of the chiral fields L and R are given in Appendix A.16) to conclude that the physical states in the spectrum of H can be classified according to the irreducible representations of the chiral group generated by Q a (as can R. Another notation worth mentioning is the one using the two-component Weyl spinors. 9. ¯ R γµ L currents vanish and we would not be able to introduce symmetry operations between R and L . Q a |0 = Q a |0 = 0 R L (9.

corresponding to each generator (with its quantum numbers) which does not leave the vacuum invariant.17) and explore its consequences in the hope of finding support for the assumption. or Nambu–Goldstone realization (Nambu 1960. Of course. It is easy then to see that all isospin multiplets would have to have at least one degenerate partner of opposite parity.15). It is then interesting to begin as we do: assume the spontaneous breakdown (9. Let the state | be an energy and parity eigenstate: H| = E| and P| =| Since Q L and Q R commute with the hamiltonian we also have H QL| In addition P QL | R and therefore the state | degenerate with | 1 = √ (Q R − Q L )| 2 = P Q L P + P| R = QR| L = E QL| and H QR| = E QR| is also a parity eigenstate P| = −| The presence of parity degenerate states is not a general feature of the known hadron spectrum. A very simple heuristic argument is as follows: . 5 Q a |0 = 0 (9. Instead we assume that the physical vacuum is not invariant (but the hamiltonian remains invariant) under the full chiral group and Q a |0 = 0. tell us whether the symmetry or part of it is spontaneously broken or not. scalar or pseudoscalar. Therefore. The proof of the theorem will be given later. if we wish to maintain our assumption about massless QCD being a good approximation to the strong interaction we must give up assumption (9. Goldstone 1961). The most important consequence of the spontaneous symmetry breakdown is summarized in Goldstone’s theorem: if a theory has a global continuous symmetry of the lagrangian which is not a symmetry of the vacuum there must be a massless (Goldstone) boson.17) is no longer free for us. given a theory like massless QCD the choice between (9.17) One then talks about spontaneous breakdown. The theory itself should. in principle.2 Spontaneous chiral symmetry breaking 321 operators under the chiral group). of the chiral symmetry SUL (2) × SUR (2) into isospin symmetry SU (2) generated by Q a . the problem of calculating the spontaneous chiral symmetry breaking in QCD has not yet been satisfactorily solved.15) and (9.9. Unfortunately.

. Another consequence of spontaneous chiral symmetry breaking which agrees with the experimental data is the Goldberger–Treiman relation.18) (we neglect the Cabbibo angle). π 3 = π 0 (therefore. the resulting commutators may be well defined if we first commute currents and then integrate over space. 5 Very plausible candidates for Goldstone bosons in the strong interactions with spontaneously broken chiral symmetry are pseudoscalar mesons.18) and (9. We assume the axial charge does not annihilate the vacuum. π → lν. That they are not exactly massless (but nevertheless much lighter than other hadrons) is attributed to an explicit chiral symmetry breaking by the quark mass terms in the lagrangian. This identification is explicit in the SU (2) × U (1) Glashow–Salam–Weinberg theory of weak interactions. Nevertheless. In the Born approximation for the electroweak decays (see Chapter 12) the pion decay. H ] = 0. π (2) = −(i/ 2)(π + − π − ). 0|Aa (0)|π b (q) = i f π qµ δ ab µ (9. It means 5 that U = exp(−iα a Q a ) is not a unitary operator. The derivation of the Goldberger–Treiman relation in the limit of exact chiral symmetry (with quark masses and pion mass neglected) which is spontaneously broken. the π ± decay constant √ is 2 f π ).19) we get µ 0|∂ µ Aa (x)|π b (q) = exp(−iq x) f π m 2 δ ab µ π (9. then goes as follows: consider the matrix element 0|Aa (x)|π b (q) = exp(−iq x) 0|Aa (0)|π b (q) µ µ where. expressions like U AU + can be meaningful. A massless boson with internal quantum numbers. from Lorentz covariance. Assuming C invariance and isospin † This interpretation requires some caution (Bernstein 1974) since Q a |0 is not a normalizable state. In our approximation the axial current is conserved: ∂ µ Aa (x) = 0 and from (9.322 9 Chiral symmetry.19) (9. therefore f π = 0. By successive application of Q a one can construct an infinite number 5 5 of degenerate states. Next consider nucleon π matrix elements of the axial current (β-decay).20) This is Goldstone’s theorem: f π = 0 therefore m 2 = 0. Restricting ourselves to the SU (2) × SU (2) case we see that the triplet of pions has the right quantum numbers for such an interpretation. amplitude is proportional to the matrix element (9.18) √ The constant f π is called the pion decay constant and π (1) = (1/ 2)(π + + √ π − ). This is based on the identification of the axial current of the two-flavour QCD with the axial current of the weak interactions for the first generation of quarks. 5 If the exponentials are expanded. spin and parity of Q a must exist to account for the degeneracy. spontaneous symmetry breaking if Q a |0 = 0 then there is a state Q a |0 degenerate with the vacuum† because 5 5 [Q a .

9. One can now ask whether the relation (9. π 2 Experimentally. and it is assumed that the pion field so defined is . Hence m N gA (0) = gπN f π (9. This agreement supports our main assumption: massless QCD with spontaneously broken chiral symmetry is a good approximation for the strong interactions.24) happens to be satisfied within 7%. Relating physical quantities to the results obtained in the exact symmetry limit m 2 = 0 is called the PCAC (partially conserved axial vector current) π approximation. the PCAC approximation has been formulated in another. Relation (9.20).25) is an identity on the pion massshell.24) is a good approximation to the real world where m 2 = 0. from (physical) π → lν | f π | = 93 MeV.22) where m N is the nucleon mass.24 and relation (9.25) where m 2 is the physical pion mass. way as an operator equation ∂ µ Aa (x) = m 2 f π π a (x) µ π (9. q = p − p Current conservation implies 2m N gA (q 2 ) + q 2 h(q 2 ) = 0 (9. equivalent.6.21) where √ σ − = (1/ 2)(σ 1 − iσ 2 ).23) where gπN is the pion–nucleon coupling constant with the same normalization convention as for f π . Originally.2 Spontaneous chiral symmetry breaking 323 invariance one has ¯ N( p )|A− (x)|P( p) = exp(iq x)u( p ) 1 σ − [γµ γ5 gA (q 2 ) + qµ γ5 h(q 2 )]u( p) µ 2 (9. For q 2 = 0 this relation can be satisfied either for gA (0) = 0 or if h(q 2 ) has a pole for q 2 = 0. from π N scattering gπN /4π = 14. as can be seen from (9. from β-decay gA (0) = 1. π a (x) is the pion field normalized as follows: π π a ( p)|π b (x)|0 = δ ab exp(i px).23) = √ √ 2gπN (i/q 2 )i 2 f π qπ uγ5 u exp(iq x) ¯ (9.24) which is the Goldberger–Treiman relation derived from the assumption of spontaneously broken chiral symmetry of strong interactions. If pions are Goldstone bosons they contribute such a pole as can be seen from (9.

proved instructive in several aspects to study a phenomenological chirally symmetric field theory model of the strong interactions. σ 3 ) (9. massless to start with so in fact we have two doublets NR and NL transforming like (1. i. The fourth field σ has been introduced to get a chirally symmetric theory. isotopic spin group α π → π + δα × π σ →σ under axial vector rotations       (9. π 3 ). The elementary fields in the σ -model are: the nucleon doublet. The postulated lagrangian is as follows ¯ ∂ ¯ σ L = N i/ N + g N (σ + iσ · π γ5 )N + 1 [(∂µπ )2 + (∂µ σ )2 ] 2 − 1 µ2 (σ 2 + π 2 ) − 1 γ (σ 2 + π 2 )2 2 4 where π = (π 1 . In this model the scalar fields are introduced as elementary fields and their interactions are arranged to produce the spontaneous breakdown of the symmetry.27) π to see that for chiral invariance of the lagrangian the (σ + iπ · σ ) must transform as the (2. For a more extensive discussion of the derivation of the Goldberger–Treiman relation based on relation (9. 2) and (2. 1) under SUL (2) × SUR (2). calculating spontaneous chiral symmetry breaking in QCD is a difficult problem which has not yet been solved. therefore. the so called σ -model. This corresponds to the following transformations (see Appendix E): under vector SU (2).324 9 Chiral symmetry.26) This is the usual pseudoscalar pion–nucleon coupling. 9. σ 2 .3 Phenomenological chirally symmetric model of the strong interactions (σ -model) As we have already mentioned. π 2 .25) and its connection to the chiral derivation see. spontaneous symmetry breaking a good interpolating field off-shell at p 2 = 0. the triplet pion field π (0− ) and σ -a scalar 0+ meson. Pagels (1975). 2) representation of the SU (2) × SU (2) (remember that 2∗ ≡ 2). for example.e. It has. σ = (σ 1 .28)    α π → π + δα σ   α σ → σ − δα · π . We rewrite ¯ ∂ ¯ ∂ ¯ ∂ N i/ N = NR i/ NR + NL i/ NL ¯ ¯ ¯ π π π N (σ + iπ · σ γ5 )N = NL (σ + iπ · σ )NR + NR (σ − iπ · σ )NL (9.

σ ] = ∓ 1 iπ a  L 2 R (9.30)  [Q a . The vacuum is defined by the minimum of the expectation values of the hamiltonian. = −Tia j j (x) (9. The physical vacuum is chosen (defined) from the degenerate ground states given by the condition (9. Q a |0 = 0 5 (9.3 σ -model 325 α (δα is an infinitesimal vector in the isospin space).35) . i ] = i (x)] (9. Chiral symmetry implies conserved axial and vector currents. Q R and using canonical commutation 5 relations for fields one gets   [Q a . In particular we would like to preserve unbroken isospin symmetry. As we know. We can rewrite the potential as V (σ . it must be that [Q a . Q a . in the tree approximation the vacuum expectation values of the fields are equal to the values of the classical fields in the minimum of the potential V (σ . π ).32) (matrices T a form a representation of generators Q a ).31) holds. j since on the other hand the operator relation i → i = exp(iα a Q a ) i (−iα a Q a ) ∼ i + iδα a [Q a . π ) = − 1 λ(σ 2 + π 2 + µ2 /λ)2 4 (9. Q L .33) to see that for µ2 /λ > 0 the minimum occurs for σ = π = 0. It is now necessary to establish the correct ground state of the model. π b ] = 1 iεabc π c ± 1 iδ ab σ  L 2 2 R Note (Appendix E) that in general if a field multiplet transforms as i → i = (δi j − iδα a Tia ) j under linear transformations generated by generators Q a then.34) With this second choice the model accounts for the spontaneous breakdown of chiral symmetry.34). The physical vacuum must then satisfy (9.34) and also Q a |0 = 0.29) Constructing then charges Q a .9. They read 1 a ¯ π Vµ (x) = N (x)γµ 2 σ a N (x) + [π (x) × ∂µ π(x)]a ¯ Aa (x) = N (x)γµ γ5 1 σ a N (x) + σ (x)∂µ π a (x) − π a (x)∂µ σ (x) µ 2 (9. for example. whereas for µ2 /λ < 0 it exists for σ 2 + π 2 = |µ2 /λ| (9. from the effective potential formalism.

give to them and to the fields the desired physical interpretation.29) give a vanishing contribution to this matrix element. where σ belongs to the chiral doublet). Our last remark in this section is as follows: the model can be extended to account for an explicit breakdown of the chiral symmetry. In view of (9.36) we must. To do this in an (9. Another way 5 is to consider 0| exp(iα a Q a )σ exp(−iα a Q a )|0 to conclude that generators of the 5 5 transformations which do not leave 0|σ |0 invariant are spontaneously broken.36) π Indeed the vacuum must be an isospin singlet (hence 0|π |0 = 0) but not a chiral singlet (hence 0|σ |0 = 0. Comparing with (9. The pattern of the spontaneous symmetry breaking from such a general viewpoint will be discussed in Section 9.29) 0|Aa (x)|π b (q) = 0|σ |0 0|∂µ π a (x)|π b (q) µ = − 0|σ |0 iqµ δ ab exp(−iq x) In the tree approximation the other terms in the axial current (9.34) is no longer arbitrary if we want to keep the isospin unbroken. a σ meson of mass m σ = 2(−µ2 /2)1/2 = (2v 2 λ)1/2 and the isospin triplet of massless pions. Formally. spontaneous symmetry breaking These conditions are satisfied with the choice π 0|π |0 = 0. define the physical σ field as σ =σ −v The lagrangian (9. π b ] = −iδ ab σ . to ensure the orthogonality of the vacuum to the one-particle state. In addition we observe that in the tree approximation.38) It describes nucleons of mass m N = g(|µ2 /λ|)1/2 = −gv. from (9. 0|σ |0 = −(|µ2 /λ|)1/2 ≡ v (9.26) can now be rewritten in terms of the field σ ¯ ¯ ∂ ¯ σ L = N i/ N + g N (σ + iσ · π γ5 )N + gv N N + 1 [(∂µπ )2 + (∂µ σ )2 ] 2 π − |µ2 |σ 2 − 1 λ(σ 2 + π 2 )2 − λv(σ 3 + σπ 2 ) + const.19) one gets f π = − 0|σ |0 The v or f π is the only mass scale of the model. 4 (9.326 9 Chiral symmetry. find the broken and unbroken generators of the theory. In the basis so defined the choice of the physical vacuum among the degenerate ground states given by the condition (9. However.5.37) . we could also proceed in another way: define any of the a priori equivalent degenerate ground states as the physical vacuum. It is worth stressing at this point that in our discussion of the σ -model we have fixed at the very beginning the physical interpretation of the group generators and of the fields. one can see that 0|σ |0 = 0 implies Q a |0 = 0 (and vice versa) by taking the 5 vacuum expectation value of the commutator [Q a .

Goldstone bosons as eigenvectors of the mass matrix First we generalize the discussion of Section 9.3 and show in the tree approximation that in a theory with spontaneously broken continuous global symmetry the mass matrix of the scalar sector has eigenvalues zero. . The corresponding eigenvectors are Goldstone boson fields. This vacuum alignment problem will be discussed in more detail later on. the theory contains elementary scalar fields i .25) introduced earlier.39) which is the operator PCAC relation (9.40) (9. any complex . i = 1. n.37) one now gets m 2 = µ2 + λv 2 = −ε/v = ε/ f π π The axial current is no longer conserved ∂ µ Aa (x) = επ a (x) µ and finally we get ∂ µ Aa (x) = m 2 f π π a (x) µ π (9. . Then minimizing the potential (9. . This proof can also be easily extended to theories without elementary scalars. . Next we give the general proof of Goldstone’s theorem (Goldstone. Another effect of the εσ term is that it defines the unique physical vacuum. Salam & Weinberg 1962) which does not rely on perturbation theory. which we take as real.6) in a more general context. Consider a theory described by a lagrangian L which has some global continuous symmetry G generated by charges Q a .4 Goldstone’s theorem 327 isotopically invariant manner one adds to L a term −εσ .33) plus the extra term one finds instead of (9.9. Among other things. 9.36) the following equation for the vacuum expectation value v: −ε − λv 3 − µ2 v = 0 With the shift (9. Here we study the problem of Goldstone bosons in theories with elementary scalars (QCD-like theories are discussed in Section 9.4 Goldstone bosons as eigenvectors of the mass matrix and poles of Green’s functions in theories with elementary scalars We have learned in the previous section that spontaneous breakdown of the chiral symmetry in the σ -model gives massless Goldstone bosons in the physical spectrum.

i (x. if the theory is to be renormalizable.45) ∂ i ∂ i must have solutions with i = vi = 0 for some of the i s. t) (9. Notice that if the vector v = (v1 . t)] = −Tia j j (x.42) Matrices T a form the representation of charges Q a in the n-dimensional space of scalar fields i . for a given potential V which gives the spontaneous breakdown of the symmetry G the vacuum is infinitely degenerate and any choice of it is physically equivalent (a choice of a solution v is a choice of the vacuum). different potentials may.44) In the tree or classical approximation. . The V ( i ) is a real polynomial in fields i .43) We are interested in theories in which the global symmetry G is spontaneously broken by the vacuum which is such that some of the fields i have non-vanishing vacuum expectation values 0| i |0 = vi = 0 (9.328 9 Chiral symmetry. Thus. Of course.41) where the unspecified terms.32)): [Q a (t). are possible couplings of the scalar fields to other fields of the theory. Because iT a must be real. irrelevant for our discussion in the tree approximation.46) is another solution to this equation.45) then it follows from the G invariance of the potential V that the vector v = exp(−i a T a )v (9. of the fourth order at most. vn )T is a solution of (9. in principle. give physically . spontaneous symmetry breaking representation can always be turned into a real one by doubling the number of the basis vectors. when the energy density of the vacuum state is just given by the minimum of the potential V ( i ). The transformation rule (9.42) is equivalent to the following commutation relation between charges Q a and fields i (see (9. . T a is purely imaginary and being hermitean it is antisymmetric. We assume furthermore that the set i forms a multiplet transforming under some irreducible representation of the group G of the global continuous symmetry of the lagrangian L (if the original fields form a reducible multiplet we change the basis to write it as a direct sum of irreducible multiplets) i → i ≈ i −i a Tia j i (9. The piece Ls of the lagrangian containing the scalar fields i is Ls = 1 ∂µ 2 i∂ µ i − V( i) + ··· (9. one then concludes that for the spontaneous breakdown of the symmetry G the condition for the minimum of V ( i) ∂V ∂V a a 0 = δV ( i ) = δ i = −i Ti j j (9.

The generators which remain unbroken must form the algebra . To discuss this point more precisely we divide the generators of G into two groups.47) and to rewrite in terms of them the potential V V( where 2 Mik = i) 2 = const + 1 Mik ( 2 − v)i ( − v)k + · · · (9. .4 Goldstone’s theorem 329 different. We conclude these considerations with several useful remarks. The massless boson fields (normal coordinates) are then the following: l = i(X l v)i i (9.48) ∂2V ∂ i∂ (9. .45) as the physical vacuum of our theory it is convenient to define the physical fields i = i − vi (9. Differentiating (9. Denoting their matrix representation by X i . Obviously. These degrees of freedom are used for longitudinal components of the massive gauge boson.9.1. Choosing some solution v of (9. N (note the values of the index i) we have X i v = 0 and (9. . Denoting their matrix representation by Y i .51) We recall at this point that our discussion is for spontaneously broken global symmetries. some generators of G must be broken. The term linear in fields does not appear in (9. For spontaneously broken gauge symmetries the Goldstone boson fields can be gauged away and do not appear in the physical spectrum. n. i = 1. of the eigenvectors X i v corresponding to zero eigenvalues of the mass matrix. for each broken generator of G. non-degenerate.49) k =v is the mass matrix for the physical scalar fields.50) which suggests that the mass matrix has eigenvectors T a v corresponding to zero eigenvalues. . .50) implies the existence. .50) does not give any new information. To the second group belong generators which are spontaneously broken. vacua. Possible patterns of spontaneous symmetry breaking are discussed in more detail in the next section. the vacuum is invariant with respect to Y i and (9.45) one gets the following result: 2 Mik iTia v j = 0 j (9. . This so-called Higgs mechanism is discussed in Section 11. i = n + 1.48) because of (9. . if some of the fields have non-vanishing vacuum expectation values. one has Y I v = 0.45). The first one consists of those which remain unbroken with our choice of the vacuum.

54) This follows from the property (9. k. One can show that spontaneous breakdown of some global continuous symmetry G of the lagrangian implies poles at p 2 = 0 in certain Green’s functions and consequently implies the existence of massless bosons in the physical spectrum. of (9.52) on the vector v we get a contradiction. Let us consider the Green’s function a G a (x − y) = 0|T jµ (x) µ. spontaneous symmetry breaking of some subgroup H of group G (it may. P X i P −1 = −X i (9. Salam & Weinberg 1962) in the class of theories specified in the beginning of this section. j ≤ n. l>n (9.55) This is the case. Therefore ci jl = 0.57) with proper µ. j > n.s. Finally. Eqs. i ≤ n.52) with some of the ci jl different from zero.54) cannot contain a term ci jl Y l .k account of the -functions involved in the definition of the T -product µ a ∂(x) G a (x − y) = δ(x 0 − y 0 ) 0|[ j0 (x). the r. Acting with both sides of (9.56) General proof of Goldstone’s theorem We turn now to the general proof of Goldstone’s theorem (Goldstone. X j ] = ici jk Y k . j > n.57) a where jµ (x) is the current corresponding to a generator Q a of the symmetry G and k (y) belongs to an irreducible multiplet of real scalar fields. Indeed. j. be that H ≡ 1). for chiral symmetry. k≤n (9. it is easy to see that broken generators transform under some representation of the subgroup H [Y i . where i.k k (y)]|0 (9. µ.330 9 Chiral symmetry. (9. l ≤ n.k k (y)|0 (9.53) and the unbroken generators form the algebra of a group. Thus.h. i. Then there is one more useful and obvious relation [X i . j. for example. k ≤ n. j.51) taken together tell us that Goldstone bosons transform under some representation of the unbroken subgroup H . Furthermore. X j ] = ici jk X k . if it is not the case then [Y i . i.53) and from the antisymmetry of the structure constants in indices i.54) and (9.1.58) . of course. Y j ] = ici jk Y k + ici jl X l . see Section 9. in many physical problems one can define a parity operation P which leaves the Lie algebra of the group G invariant and such that PY i P −1 = Y i . l>n (9. The Green’s function G a (x − y) satisfies a Ward identity obtained by differentiating (9. i. k > n (9.

from Lorentz invariance the general ˜ µ. are assumed. for the discussion of anomalies in theories with fermions see Chapter 13.4 Goldstone’s theorem 331 This is an example of the so-called non-anomalous Ward identity (see Section 10.k structure of G a ( p) can be written as ˜ µ. see (9. in principle.1). Let us take it as a relation for ˜ µ.59) which follows from the assumed transformation properties. related to these 0| j |0 by (9.61) one gets from (9.60) (see also the next chapter). Thus. k (y. t)δ(x − y) (9.k G a ( p) = pµ Fka ( p 2 ) and therefore Fka ( p 2 ) = −iTkaj 0| j (0)|0 (9.62) This last equation implies that if some of the vacuum expectation values 0| j |0 ˜ µ.k d4 p ˜ µ.63) (1/ p 2 ) (9. However.k renormalized quantities. Using the relation a [ j0 (x.k i p µ G a ( p) = Tkaj 0| (9.60) A few words are worthwhile here about renormalization properties of (9.k exp[−i p(x − y)]G a ( p) (2π )4 j (0)|0 (9.64) Our last step is to show that poles at p 2 = 0 imply the existence of massless bosons in the physical spectrum.59).60) the following: ˜ µ. Introducing the Fourier transform G a ( p) G a (x − y) = µ. Indeed. the conserved currents are not subject to renormalization and the field renormalization 1/2 R 1/2 R constants iB = Z i i = Z i do not carry the group index because the lagrangian is G invariant.43).62). one gets the following Ward identity (translational invariance is assumed) µ ∂(x) G a (x − y) = −∂(x − y)Tkaj 0| µ. the renormalized fields transform under the group G in the same way as the bare fields and the same Ward identity (9.k do not vanish then there are poles at p 2 = 0 in the Green’s functions G a ( p). derived for bare currents and fields for which transformation properties (9.65) where the vector |π k ( p) describes a particle of mass m k which is a quantum of the . of fields i under the group G. To this end we consider the matrix element a 0| jµ (x)|π k ( p) = i f ka pµ exp(−i px) (9.60) holds both for bare and renormalized Green’s functions. The Ward identity is. t).k j (0)|0 (9. t)] = −Tkaj j (y.9.

65) gives us our final equation: p 2 →m 2 k (9.66) the following result: a ˜ µ.69) ˜ µ.69) implies m 2 = 0 and k f ka = 0: f ka = iTkaj 0| j (0)|0 (9.k = lim −i exp(−i px)G a ( p)( p 2 − m 2 ) k Comparison with (9.k ˜ kk 0| jµ (x)|π k ( p) = lim exp(−i px)G a ( p)iG −1 ( p) p 2 →m 2 k p 2 →m 2 k ˜ µ.7) one can relate the matrix a ˜ µ. If the broken generators and the corresponding currents form a single real irreducible representation of the unbroken subgroup H which leaves the vacuum invariant then from Schur’s lemma a 0| jµ |π k ( p) ∼ δ ak (9.k kk (9. Using the reduction formula (see Section 2.71) and f ka = f a δak .70) Thus there must be massless bosons | a ( p) = iTkaj 0| j (0)|0 |π k ( p) in the physical spectrum corresponding to each broken generator which does not leave the vacuum invariant Tkaj 0| j (0)|0 = 0.k element 0| jµ (x)|π k ( p) to the Green’s function G a ( p) a 0| jµ (x)|π k ( p) = d4 y d4 z f p (z)G a (x − y)iG −1 (y − z) µ.66) where f p (z) = exp(−i pz) and  −δk k d4 q   G k k (y − z) = exp[−iq(y − z)]  4 q 2 − m 2 + iε (2π ) k    d4 y G −1 (x − y)G(y − z) = δ(x − z) (9.68) ˜ µ. spontaneous symmetry breaking field k .64)) have for spontaneously broken symmetry a pole at p 2 = 0 then (9.67) Writing exp(−i pz) as exp(−i px) exp[i p(x − y)] exp[i p(y − z)] we get from (9.332 9 Chiral symmetry.k If G a ( p) is one of those Green’s functions which (according to (9.k lim G a ( p)( p 2 − m 2 ) = − f ka pµ k (9.

5 Patterns of spontaneous symmetry breaking 333 For instance. The breakdown of the presented proof is most easily seen if we work in a ‘physical’ gauge in which the gauge field has only physical degrees of freedom. A choice of the vacuum specifies the pattern of the spontaneous symmetry breaking which may. be such that some subgroup H of G remains unbroken. For a discussion in a covariant gauge see. in the absence of reliable methods for calculating non-perturbative effects. 0| ¯ |0 ) which. if they exist. Although different patterns.70). in general. Gauge fixing then requires a specification of some four-vector n µ and the general form of the matrix element involving the gauge source current is no longer given by (9. Goldstone’s theorem does not hold for spontaneously broken gauge symmetries (see Section 11.5 Patterns of spontaneous symmetry breaking For a theory with some global symmetry G and a given multiplet content of fields. 3 )T . patterns of spontaneous symmetry breaking are determined by the properties of the vacuum defined by the non-vanishing of the vacuum expectation values of some fields.2). As we have already mentioned in the previous section for any choice of the vacuum specified by the non-vanishing vacuum expectation values there exists a set of physically equivalent vacua.63): a term proportional to n µ can be present. in general. In theories with elementary scalars the non-vanishing vacuum expectation values may already appear in the tree approximation as a solution minimizing the assumed potential V ( ) or may be generated by radiative corrections (Section 11. with matrices Tkaj from Appendix E. no explicit reference to the potential is necessary for our discussion. correspond to different potentials V ( ). are usually introduced by assumption. must. obtained from the original one by the set of unitary transformations belonging to G. Our first example is a group of orthogonal transformations. which also minimize the same potential V ( ). for the σ -model with 0|σ |0 = 0 we get from (9. Different choices of the physical vacuum among the degenerate vacua give the same pattern of symmetry breaking. for example. The question we would like to discuss here is what are the possible patterns of spontaneous breaking of a given symmetry G. Bernstein (1974). 2 .1). f π = − 0|σ |0 (9.9. In theories without elementary scalars the symmetry is spontaneously broken by some condensates (for example. O(3) for definiteness.72) as the generalization of the result obtained in the tree approximation. 9. and the scalar fields are assumed to form a real multiplet = ( 1 . they simply correspond to different orientations of the unbroken subgroup H in the group G which can be transformed one into the other by a redefinition of the basis.

The matrix representation of generators Q i in the space of s is       0 0 0 0 0 i 0 −i 0 Tx =  0 0 −i  Ty =  0 0 0  Tz =  i 0 0  (9. a physically distinguished direction. ]|0 = − iαTz Thus the group O(3) is broken to the group O(2) of rotations around the z axis.75) 0 0 0 0 0 O| exp(iα Q z ) exp(−iα Q z )|0 ∼ = + iα 0|[Q z . the vacuum defined by (9. This would not be the case if there was. This point will be discussed in detail in the next chapter. spontaneous symmetry breaking transforming under the vector representation. Both theories are physically equivalent and transform one into the other by a redefinition of the basis.334 9 Chiral symmetry. Any other choice of the vacuum expectation value   v1  v2  0| |0 = (9.32) we get = 0 (9.77) √ χi = (1/ 2i)( i − i∗ ) .74) breaks spontaneously the other two generators:     0 −v iTx 0 =  v  . Consider now unitary groups with scalars in a real representation. There are two Goldstone bosons corresponding to two linearly independent vectors iTx 0 and iTy 0 .76) v3 which can be obtained from the original one by unitary transformations belonging to G gives the same pattern of symmetry breaking but this time with the combination v1 Tx + v2 Ty + v3 Tz unbroken.74) However.   0 0| |0 =  0  ≡ 0 v is. Any complex field can be written in a real basis by defining two real fields √ ϕi = (1/ 2)( i + i∗ ) (9. iTy 0 =  0  (9. for The vacuum defined this way is invariant under the transformations generated by Q z : taking the transformed vacuum exp(−iα Q z )|0 and using (9.73) 0 i 0 −i 0 0 0 0 0 A possible choice for the vacuum expectation value of the multiplet instance. that is if the O(3) symmetry was also broken explicitly. in space.

. v. thus. Our next example is the group SU (2) with a doublet of complex fields 1 and 2 . there is just one pattern of spontaneous symmetry breaking and any non-zero choice of the vacuum expectation value 0| r |0 = (v1 v2 )T is equally good. 0| r |0 = (0.79) Let us take. 0)T . . T1 =  T2 =    i   i 0 −i i   i 0   −i  T3 =   −i  0 i (9. 0. All other possible non-vanishing vacuum expectation values can be obtained from the one above by unitary SU (2) transformations and.5 Patterns of spontaneous symmetry breaking 335 For the group U (1) of the unitary transformations → the real representation is r = exp(−i ) → r = exp(−i T ) ϕ χ r where r = (9.78) and T is purely imaginary T = 0 i −i 0 because r must be real. In the real representation   ϕ1  χ  r = 1   ϕ2  χ2 the generators of the SU (2) transformations are     −i i 0    −i  −i . give the same pattern of symmetry breaking. Obviously.9. Following the discussion of our first example we see that the group is broken completely because the three vectors iT a 0| r |0 = 0 are linearly independent. for instance. in this case.

8 (9. In this case we have one more generator   i 0   −i  T =  i  0 −i but the vector iT 0| r |0 is not linearly independent from the other three vectors iT a 0| r |0 . Then the generators 5 T 2 . . 0)T . not always the case. However. thus obtaining the same theory as before. for 0| a a the generators 5 T are broken and the T s remain unbroken: SUL (2) × SUR (2) → SUL+R (2). spontaneous symmetry breaking It is easy to extend the analysis to the SU (2) × U (1) group with one complex doublet of scalar fields (Glashow–Salam–Weinberg model). 0. 0. 0| r |0 = (v. For instance. b = 1. 5 T 3 and T 1 remain unbroken and we see from the commutation relations (9.336 9 Chiral symmetry. 0. This is. where a Tbc ≡ −icabc b] a = −Tbc c (9. Take. So far we have been discussing examples with only one possible pattern of spontaneous symmetry breaking. of course. for 0| r |0 = (0. . We can call this group the isospin group and interpret as physical pions those combinations of the fields π i and σ which transform as a triplet under this new SU (2). v.80) It is convenient to write the scalar multiplet in the matrix form ˆ = 1 2 b b λ . Thus. 0)T we have T 0| r |0 = −T 3 0| r |0 and the combination T + T 3 remains unbroken. v)T 5 T collected in Appendix E we see that. for example. 0. for example.9) that they also form an SU (2) algebra. 0.81) . The spontaneous breaking of chiral symmetry in the σ -model can be discussed in a similar manner. . again only one pattern of symmetry breaking is possible. any other choice of vacuum among the SU (2) × SU (2) degenerate set gives the same pattern of symmetry breaking. As our last example we consider the group SU (3) with a real multiplet of scalar fields transforming under the adjoint representation of the SU (3) [Q a . Using the real basis r = (π1 π2 π3 σ )T and the matrices T a and a r |0 = (0. Other choices of the vacuum give only different unbroken combinations of the generators T and T a but a redefinition of the basis transforms one case into another. This corresponds to the standard choice 0|σ |0 = 0 and to the π s forming a triplet under the SU (2). .

ˆ] (9. with no elementary scalars. However. for each pattern there is an infinitely degenerate set of possible vacua obtained from (9. It is also worth noticing that a vacuum expectation value giving SU (3) → SU (2) × U (1) cannot be transformed by unitary transformations belonging to SU (3) to one which gives SU (3) → U (1) × U (1) and they must correspond to different potentials V ( ). In this case we have six Goldstone bosons. λ7 . The matrix ˆ is hermitean and ˆ ii = 0 so it has eight independent elements. of the fact.82) and because of (9. The transformation rule for the matrix ˆ is ˆ → ˆ = exp(−i a1 a λ ) 2 ˆ exp(i a1 a λ ) 2 ∼ ˆ −i = a 1 a [2λ .6 Goldstone bosons in QCD 337 and λb are the Gell-Mann matrices. 9. as we do for the σ -model in Section 9. However. Taking them as λ3 and λ8 we can write the most general form of the vacuum expectation values of the scalar fields as follows: 0| ˆ |0 = 1 v3 λ3 + 1 v8 λ8 2 2 (9. λ2 .1.83) (with specified v3 and v8 ) by unitary SU (3) transformations.81) it is equivalent to → = exp(−i a T a) with T a given by (9. If we assume v3 = 0.80). that our understanding of . namely SU (3) → SU (2) × U (1).82) that λ3 and λ8 remain unbroken for any choice of v3 and v8 . if v3 = + 3v8 √ √ the λ6 . λ3 (SU (2)) and λ8 (U (1)) remain unbroken.83) First of all we notice that in our example the maximal possible breaking is SU (3) → U3 (1) × U8 (1) because it is clear from (9.3 and for QCD in Section 9. if 2 √2 √ 8 √ v3 = − 3v8 the λ4 . 1 (λ3 + 3λ ) (SU (2)) and 1 ( 3λ3 − λ8 ) (U (1)) 2 2 are unbroken. Because the matrix is hermitean it can be diagonalized by unitary transformations and written in terms of commuting generators of SU (3).9. λ5 . 1 (λ3 − 3λ8 )(SU (2)) and 1 ( 3λ3 + λ8 ) (U (1)) are unbroken.6. another pattern of symmetry breaking is also possible. Chiral symmetry of the massless QCD lagrangian has been discussed in Section 9.6 Goldstone bosons in QCD The purpose of this section is to extend the proof of Goldstone’s theorem to theories like QCD. v8 = 0 the √ generators λ1 . however. For a given pattern of spontaneous symmetry breaking one usually works from the beginning in a convenient basis with a specified physical interpretation of the group generators and of the fields. One should be aware. Considerable phenomenological evidence has then been summarized that the group SU (2) × SU (2) is an approximate symmetry of the strong interactions which is spontaneously broken by the vacuum. for the multiplet .83) and (9.

2 are flavour indices). Let us introduce scalar operators and † ij = ¯ Lj Ri . Thus we find that the vacuum |0 has the property that operators which destroy or create such a fermion pair have nonzero vacuum expectation values 0| ¯ R L |0 = 0. we shall assume the spontaneous breakdown of chiral symmetry and then show that it implies the existence of massless bosons (composite states) in the physical spectrum. 0| † i j |0 =0 0| i j |0 = 0| † i j |0 .84) (i.338 9 Chiral symmetry.85) i j |0 = 0. We consider QCD with two flavours and the chiral symmetry SUL (2) × SUR (2). we omit indices here. It has at least been realized that only asymptotically free theories allow spontaneous chiral symmetry breaking at reasonable momentum scales (Lane 1974. Gross & Neveu 1974). 0| ¯ L R |0 = 0. In the following we assume that the chiral symmetry of QCD is indeed spontaneously broken by non-vanishing vacuum expectation values of some fermion condensates. Assuming that it becomes arbitrarily strong we expect that the ground state of the theory has an indefinite number of massless fermion pairs which can be created and annihilated by the strong coupling. ( † )i j = ¯ R j Li (9. We still expect the ground state to be invariant under Lorentz transformations so these pairs must have zero total momentum and angular momentum. j = 1. This relies on the fact that the gauge coupling in QCD becomes strong at large distances. From the transformation properties of the fields Li and R j under the chiral group Ri † Lj = (UR )ik = Rk † † Lk (UL )k j it follows that † = UR UL ( We assume 0| and in addition † ) = UL † † UR (9. It is nevertheless useful to give at least an intuitive argument in favour of the spontaneous breaking of chiral symmetries in asymptotically free theories. It is not our aim here to review partial results suggesting that this indeed happens. spontaneous symmetry breaking the underlying mechanism of spontaneous symmetry breaking in theories without elementary scalars is not yet satisfactory.

First we collect the useful equal-time commutation relations following from (9. can be most generally written as (see the discussion preceding (9.4 and it is based on Ward identities for certain appropriate Green’s functions.85) that the vacuum expectation value v1 is invariant under the simultaneous UL = UR l transformation. For the proof of Goldstone’s theorem we use a Ward identity for Green’s functions involving composite operator a (x) defined as follows: a j0L a a ( j0R ) (x) = Tr[ (x) · T a ] . v3 = 0 in (9. t). We can also write 0| i j |0 = 1 0| ¯ |0 δi j (9. t)δ(x − y)    [ j a (x. From now on we always refer to renormalized quantities.86) (we always work in a specified basis of chiral fields). We want the chiral SUL (2) × SUR (2) symmetry to be broken to the SUL+R (2) symmetry which is the isospin symmetry in our basis. t)T a δ(x − y)    a a [ j0R (x. t)δ(x − y)  (9. now the 0| |0 is a 2 × 2 complex hermitean matrix. t).6 Goldstone bosons in QCD 339 The latter equality ensures that parity is not spontaneously broken since then 0| ¯ i γ5 j |0 = 0|( − † ) ji |0 = 0 (9. t)] = −T a † (x. Therefore we assume v = 0. (y. t)] = † (x.87) where σ 3 is the diagonal Pauli matrix.87a) 2 where ¯ = ¯1 1 + ¯2 2 We are ready to prove Goldstone’s theorem. t)T a δ(x − y)  0R where the are the zeroth components of the currents of the SUL (2) (SUR (2)) transformations and the matrices T a represent the generators of the SUL (2) (when to the right of the operator ) and of the SUR (2) (when to the left of the ) in the space of chiral doublets L and R .88) a  [ j0L (x. The above commutation relations are valid both for the bare and the renormalized quantities because chiral symmetry is an exact symmetry of the lagrangian (see also the next chapter). t). being a hermitean 2 × 2 matrix. where the SU (2) × SU (2) → SU (2) is the only possible pattern of symmetry breaking.87): it is clear from (9. Notice the difference with the σ -model with one complex doublet. t). respectively.59) and (9. † (y.85)  a  [ j0L (x. † (y. or four real fields. The vacuum expectation value 0| |0 .9. t)] = (x. (y.83)) 0| |0 = v1 = v3 σ 3 l (9. t)] = −T (x. The proof is similar to that in Section 9.

96) . The only modification is that some care must be taken about the interpretation of the operator (x).62)–(9.95) d4 x G ab (x) exp(i px) ≈ p 2 →m 2 iZ 2 δ ab p2 − m 2 (9. The proof goes as in Section 9. consider the Green’s function G ab (x − y) = 0|T a (x) b (y)|0 (9.95) A single particle state with appropriate quantum numbers gives a pole at p 2 = m 2 in the Fourier transform of (9. exp[−i p(x − y)]G ab ( p) (2π )4 (9.90) where the curly bracket on the r. µ b (y)|0 = d4 p ˜ µ. t). (y)+ † (y)}T b ]|0 (9. σ a are Pauli matrices). 0 (y.h. Tr[T a T b ] = 1 δ ab and (9. Using the commutation relations (9.4.87) one gets 2 µ ∂(x) 0|T Aa (x) µ b (y)|0 = −2viδ(x − y)δ ab This result is analogous to (9.60) and it implies a pole at p 2 = 0 in the Fourier ˜ µ. Indeed. t) + † (y. (y. G ab (x − y) = 0|T Aa (x) µ.93) Following the steps (9. spontaneous symmetry breaking where (x) = i( and therefore a − − † ). t)}δ(x − y) (9. G ab ( p) = pµ 2 δ ab = pµ 2 δ ab p p (9. transform G ab ( p) of the Green’s function G ab (x − y) µ. means anticommutator.91) Using (9.89) (x) = i( ij † a i j )T ji = i ¯ T aγ 5 (T a = 1 σ a .s. Using this last result we finally arrive at the desired Ward identity: µ ∂(x) 0|T Aa (x) µ b (y)|0 = −iδ(x − y) 0| Tr[{T a . t)] = −i{T a .88) we get 2 the following result for the axial currents: [Aa (x.92) (it is unaltered by the chiral anomalies discussed in Chapter 13).64) we get ¯ 2v ˜ µ. † = (9.340 9 Chiral symmetry.94) ˜ µ. The pole at p 2 = 0 in the Green’s function G ab ( p) implies the existence in the physical spectrum of the massless Goldstone boson with the quantum numbers of the operators a .

in the sense of Chapter 2.9.100) ˜ µ. ≈ −i exp(−i px)Z −1 G ab ( p)( p 2 − m 2 ) (9. In this case we have two low energy parameters: f π and ¯ .68) 0|Aa (x)|π b ( p) µ to get p 2 →m 2 p 2 →m 2 a (x) (9. lim G ab ( p)( p 2 − m 2 ) = −Z f π δ ab pµ (9.97) (9.65) 0|Aa (x)|π b ( p) = i f π δ ab pµ exp(−i px) µ and the steps (9. (9. Since G ab ( p) has the pole (9. Introducing the physical.6 Goldstone bosons in QCD 341 where Z is a finite dimensionful constant ( a is a renormalized operator). .66)–(9. field π a (x): π a (z) = Z −1 we repeat the step (9.99) ˜ µ.101) This completes the proof of Goldstone’s theorem.94) due to the spontaneous breakdown of chiral symmetry. It does not carry the group index because G is an exact symmetry of the lagrangian.98) ˜ µ.100) implies m 2 = 0 and Z fπ = − ¯ (9.

Now we would like to introduce the possibility that the symmetry G of the lagrangian is also broken explicitly by a small perturbation. This will also be useful as a supplement to our use of Ward identities in the earlier chapters and as an introduction to Chapter 13 devoted to the problem of anomalies. In addition the symmetry G may or may not be spontaneously broken.1 Internal symmetries and Ward identities Preliminaries We are interested in theories described by lagrangians L such that L = L0 + L1 (or H = H0 + H ).10 Spontaneous and explicit global symmetry breaking So far we have been discussing theories described by lagrangians having some exact global symmetry G spontaneously broken by the vacuum. 10. 342 . where L0 is invariant under global internal symmetry G and L1 explicitly breaks this symmetry. We assume that L1 can be treated as a perturbation whose effect can be calculated as a perturbative expansion in an appropriate small parameter. We begin with a systematic discussion of Ward identities between Green’s functions involving global symmetry currents. Several interesting physical problems fall into this category. Furthermore. let the bare fields in the lagrangian L form a basis for some definite representation R of the symmetry group G of the lagrangian L0 δ ( i = −i a Tia j j ≡ a a δ i (10.1) stands for boson or fermion fields).

Whether G is spontaneously broken or not we are able to derive the same Ward identities.1 Internal symmetries and Ward identities a The bare currents jµ (x) constructed according to Noether’s theorem a jµ (x) = −i 343 ∂L Ta ∂(∂µ i ) i j j (10. In such theories some of the Ward identities derived here are afflicted with anomalies. The same discussion applies when the symmetry G is spontaneously broken since in this case also only symmetric counterterms are needed: this follows from (2.2) which are conserved currents of the G symmetric classical theory described by L0 .161).6) Remembering that L0 is invariant under global transformations (10. We assume in this section that the UV regularization of the theory does preserve its classical symmetries. .e. a ∂ µ jµ (x) = δ a L1 (x) (10. Both bare regularized and renormalized Green’s functions are G symmetric. Let us first consider the case in which the symmetry G is an exact symmetry of the lagrangian. As we will see in Chapter 13 this is not true in theories with chiral fermions.10. we assume this to be the case. If the regularization preserves this symmetry then the necessary counterterms are also G-invariant.5) In the following we shall also need the variation of the same lagrangian L under local transformations i (x) = i (x) + a (x)δ a i (x) (10. remain unchanged when L0 → L if L1 does not contain derivatives of the fields a present in the jµ (x).4) and. i.3) the lagrangian L transforms as follows: L (x) = L(x) + a a δ L(x) = L(x) + a a δ L1 (x) (10. L1 = 0. using the classical equations of motion. Under an infinitesimal global symmetry transformation i (x) = i (x) + a a δ i (x) (10.3) and that L1 does not contain derivatives of the fields we get δL(x) = a a (x)δ a L(x) + jµ (x)∂ µ a (x) (10.7) We want to derive Ward identities for bare regularized or for renormalized Green’s functions following from the symmetries of the lagrangian. The renormalized fields transform under the same representation R of G as the bare fields: the renormalization constants are G-invariant.

This is so-called soft explicit symmetry breaking. for such a ‘soft’ breaking the wave-function renormalization constants.† using (10. If the integration measure is invariant. We now define the Green’s functions 0|T A(x) i i (yi )|0 ∼ D i A(x) i i (yi ) exp(iS) (10.s.2) do not affect counterterms of dimension 4.1) are traceless. of (10. included in counterterms of dimension 4. the wave-function renormalization constants are mass-independent and thus the same as in the chirally symmetric theory.9) The crucial point in deriving (10. Our general remarks can be illustrated with the minimal subtraction renormalization scheme used for QCD with the mass term m ¯ .7) we get 0= d4 x ×[ a D i exp iS + i a d4 x J i i a a (x)δ a L + jµ (x)∂ µ (x) + J i (x) (x)δ a i (x)] (10. for example. † We consider cases in which det(D i /D j ) = 1 + O( 2 ). This is true when generators T a in (10. in this case there is no regularization of the path integral and no renormalization prescription for the Green’s functions which preserve the full chiral symmetry and anomalies are present.10) is just a short-hand notation for its r.8) The integral is invariant under the change of integration variables defined by (10. δ a L(x) and δ a i (x).10) a where A(x) denotes one of the jµ (x).6).6).h.9) is the invariance of the integration measure under the transformation (10. It breaks down for chiral transformations in theories with chiral fermions (see Chapter 13). remain symmetric under G (strictly speaking there exists a renormalization scheme such that they remain symmetric) and the renormalized fields of the theory with softly broken symmetry G transform under G in the same way as the bare fields. being of dimension lower than or equal to d (see the discussion in Section 4.344 10 Spontaneous and explicit global symmetry breaking With L1 present we consider the case in which symmetry breaking operators have dimension d < 4 (L1 then involves dimensionful constants).s. for bare regularized Green’s functions W [J ] = N D i exp i d4 x [L( i) + Ji i] (10. Ward identities from the path integral Given all these preliminary constraints we can proceed to derive Ward identities using the functional integral formulation of quantum theory. In other words. We also assume the absence of anomalies. Thus.h. . One should remember that the quantity on the l. Consider the generating functional W [J ]. Then the counterterms required by L1 .

an action which is exactly invariant under certain symmetry transformations.6) if simultaneous transformations on A.12) where S0 [ ] is the part of the action symmetric under global transformation (10.11) vanishes. Aa and K for the operators i . respectively. to make easy contact with the operator language. J. K ] is invariant under the local transformation (10. In our case the background fields are the sources j i . J.11) Ward identity (10. Action µ (10. We want to introduce them in such a way that the action S[ . µ a jµ and L1 . K ]} .9) by parts (remember the remarks following (2. A. This is achieved if S[ .12) is invariant under (10. A. of (10.10.h. however. Under the constraint of explicit breaking being only soft the Ward identity is valid both for bare regularized and for the renormalized Green’s functions. K ] = N D i exp{iS[ .11) is valid irrespective of whether the symmetry is spontaneously broken or not.1) and the background fields Aa have been introduced as gauge fields by changing µ the normal derivatives in S0 into covariant derivatives ∂µ → ∂µ + iAa T a .1 Internal symmetries and Ward identities 345 The notation has been chosen. A. J and K defined as δ Aa = +∂µ µ δJi i a (x) + cabc i. J. If the symmetry G is an exact global symmetry of the lagrangian the first term on the r.s. A] + d4 x (J i i + K · L1 ) (10. We take the latter to have well-defined transformation properties under the symmetry group. b Ac (10.13) = −J i δ δ K · L1 = −K · δL1 The generating functional W [A. K ] = S0 [ .6). Integrating (10. J. We may also be interested in Ward identities for Green’s functions involving a several symmetry currents jµ and/or the symmetry breaking operator L1 . The most compact way to derive such Ward identities is based on the general idea (see Chapter 8) of using background fields to get. by construction.55)) differentiating it functionally with respect to the sources Ji and then setting them to zero we get the following general Ward identity: a (∂/∂ xµ ) 0|T jµ (x) i i (yi )|0 = 0|T δ a L(x) i i (yi )|0 i (yi ) j=i j (y j )|0 −i i δ(x − yi ) 0|T δ a (10.

µ µ K = 1.15) Eq.17) Note that in the latter case the symmetry may also be spontaneously broken since . ]}(J i δ a i +K ·δ a L1 −∂µ j µa +cabc Ab j µc ) µ a (x) (10. J . K ] The last statement contains all the information equivalent to Ward identities which follow from the global symmetry of the original theory.15) generates the desired Ward identities valid to any order in L1 by differentiating with respect to sources J i . J. for instance. When the symmetry is explicitly broken by perturbation L1 we have δ a 0|T i i (yi )|0 = −i d4 x 0|T δ a L i i (yi )|0 i i (yi )|0 ≡ −i 0|T δ a S (10. For infinitesimal transformations the variation of the functional W [A. The method can be generalized to generate Ward identities involving other composite operators with well-defined transformation properties under the global symmetry group if we introduce sources for such operators. K . J. K and Aa and setting J i = Aa = 0.346 10 Spontaneous and explicit global symmetry breaking is then invariant under transformations (10. may also require subtractions in the form of polynomials in momenta in momentum space or of δ-functions and their derivatives in position space. in addition to multiplicative renormalization. Thus in the case of exact symmetry which is not spontaneously broken one gets δ a 0|T i i (yi )|0 =0 (10.13) of the background fields W [A. J. (10.14) a or specifying jµ . K ] reads 0= d4 x δW δW δW δ J i (x) + δ Aa (x) + δ K (x) δ Ji (x) δ Aa (x) µ δ K (x) µ (10.11) can be integrated over spacetime and the surface terms neglected.16) which simply means an exact invariance of the Green’s functions under the symmetry transformations. Finally we observe that in the special cases of no massless Goldstone bosons α coupled to the current jµ . In particular renormalization of Green’s functions involving products of composite operators. Ward identities like (10. K ] = W [A . 0 = d4 x D i exp{iS[A. We stress again that the validity of the symmetry relations for the Green’s functions relies on the existence of the regularization and renormalization procedures which preserve this symmetry. to flavour fermionic currents.

18) [Q a (t). Q b (t)] = icabc Q c (t) (10. However. .18) are true in the symmetric theory and also with the symmetry breaking term L1 included as long as such terms do not contain derivatives of the fields. t)] = icabc j0 (x. Q a (t)] (10.19) a d3 x j0 (x. t) (10. t) = i dt 0 d3 x [H (x. t) In accordance with the transformation properties (10. The time dependence of the charges is given by dQ a = dt d3 x d a j (x. and therefore a ∂ µ jµ (x. t).18) Relations (10. j0 (y.21) If the symmetry breaking is soft (10.20) = −Tia j j (y.10.23) (the surface terms at infinity can be neglected because due to perturbation L1 the would-be Goldstone bosons acquire masses). t)] = −Tia j j (x. by analysing the classical lagrangian. or a [ j0 (x. t)δ(x − y) (10. t). Q a (t)] (10. t)] i (x. already be seen at the level of the tree approximation.21) hold for the renormalized quantities as well because wave-function renormalization constants can remain invariant (for properly chosen renormalization schemes) and as will be seen shortly the currents are not subject to renormalization. i (y. t)δ(x − y) (10. t). t) = i[H (x. i.e.1 Internal symmetries and Ward identities 347 in the presence of perturbation L1 the massless Goldstone bosons acquire masses. This can.18)–(10.1) we have furthermore [Q a (t).22) where H is the symmetry breaking part of the hamiltonian. for instance. t). Comparison with the operator language The equal-time bare current commutation relations following from the equal-time canonical commutation relations for the bare field operators are a b c [ j0 (x. in the presence of the symmetry breaking term L1 the currents are no longer conserved and the corresponding charges are time-dependent. Defining Q a (t) = we get from (10.

24). i. The former are always understood to be regularized or renormalized covariant quantities. Similarly. for instance.26) (Lorentz and other symmetry group indices must. generally speaking.24) (with a straightforward generalization to products of more than two operators) by direct differentiation of the Green’s function and use of the commutation relations (10. the so-called seagull terms. The question thus arises of the role of such terms which in momentum space are polynomials in momentum.e. of course. the meaning of time-ordering as given by (10. For instance a a (∂/∂ xµ ) 0|T jµ (x) (y)|0 = 0|T ∂ µ jµ (x) (y)|0 a + δ(x 0 − y 0 ) 0|[ j0 (x).25) with (x) replaced by any operator O(x). This relation is obtained by differentiating the T -product defined by (10.21). Ward identities and short-distance singularities of the operator products Consider.5) for the operator ∂ µ jµ we recover the Ward identity (10. whereas the latter are not and they may differ by local terms. .25) a and using (10. Therefore there is an apparent ambiguity T A(x)B(0) → T A(x)B(0) + c1 δ(x) + c2 ∂x δ(x) + · · · (10. be properly taken into account).11) is the same as one derived for the time-ordered product of operators defined as a a T jµ (x) (y) = jµ (x) (y) (x 0 − y 0 ) + a (y) jµ (x) (y 0 − x 0 ) (10. at x0 = y0 .15) if we neglect in the current commutators the presence of the derivatives of the δ-functions (the so called Schwinger terms). in many cases these terms certainly do not vanish (see Jackiw (1972) for a systematic discussion) and one may wonder about the compatibility of the two methods. However.24) is clear everywhere except at coinciding points.348 10 Spontaneous and explicit global symmetry breaking The Ward identity (10. The path integral approach tells us that it is always possible to arrange such cancellation by a suitable renormalization prescription if there exists one which preserves the symmetry of the classical lagrangian. the Ward identity (10. which cancel the Schwinger terms.13) or (10.21) together with (10.11). However. Otherwise anomalies appear. (y)]|0 (10. The answer is that the Green’s functions defined by the path integral do not always coincide with the vacuum expectation values of the T -products defined by (10. one can derive Ward identities for Green’s functions involving products of currents and they are the same as the relations following from (10. We shall come back to these problems in some detail in the following.24).

29) We expect the meaningfulness of the Ward identity (10. i. In particular.e.h.10.25) are for distributions.28) where f˜(q) is the Fourier transform of f (x) and use the notation ˜ T (q) = d4 x exp(iq x)T (x) The Fourier transform (∂µ T ) of ∂µ T (x) is (2π)−4 d4 q f˜(−q)(∂µ T ) = =− i. the derivative ∂µ T (x) of the distribution T (x) is defined as follows: d4 x f (x)∂µ T (x) ≡ df d4 x [−∂µ f (x)]T (x) (10. Then the integrand .25). so that the lightcone singularities of the operator product collapse to the origin.32) Performing Wick’s rotation we can work in the Euclidean space.25) to be related to the short-distance singularity structure of the distributions on both sides of (10.31) (10.27) the l. ˜ (∂µ T ) = −iqµ T (q) In the other notation d4 x exp(iq x)∂µ T (x) = −iqµ d4 x exp(iq x)T (x) (10.27) We also define the Fourier transform of a distribution (2π)−4 ˜ d4 q f˜(−q)T (q) = d4 x f (x)T (x) (10. let us rederive (10.25) can be written as − lim ε→0 d4 x ∂µ f (x) 0|T jµ (x)O(0)|0 (|x0 | − ε) (10. To see this clearly.25) paying explicit attention to the singularities at x ∼ 0.s.1 Internal symmetries and Ward identities 349 To discuss this point we recall first that equations like (10. Using the definition (10.30) d4 x f (x)∂µ T (x) d4 x ∂µ f (x)T (x) (10. they hold in the sense that both sides are integrated over d4 x f (x) with f (x) being any ‘smooth’ test function vanishing at xµ → ±∞.e. of (10.

33) and (10.32) is finite and unambiguous for ε > 0. We have also assumed the absence of massless particles. O(0)] = a 1 O1 (x. −ε) − j0 (x.37) is a well-defined relation. with no additional subtractions the Ward identity (10. after excluding x = 0.35) with the equal-time commutator given by (10. 0). Thus.34) If the limit ε → 0 exists. em em Often T jµ (x)O(0) is more singular than O(1/x 3 ) (for example T jµ (x) jν (0) ∼ (1/x 6 ) · 1 Stronger than O(1/x 3 ) singularity in the OPE of the T -product of l). we have ˜ −i pµ T ( p) = ( p) + d3 x exp(ipx) 0|[ j0 (x.35) where ( p) is the Fourier transform of 0|T ∂ µ jµ (x)O(0)|0 . 0). then the surface terms at ±∞ do not contribute.36) we also get (see Section 7.34) using the OPE. operators translates into terms additional to the δ(x) term with a finite coefficient in .350 10 Spontaneous and explicit global symmetry breaking in (10.37) (no Schwinger terms.33) In the last term the operator ∂µ has been interchanged with T because.36) 1 where Cµ (x) ∼ O(1/x 3 ). Let us discuss the limit ε → 0 in (10. ε)O(0)|0 d4 x f (x) 0|T ∂ µ jµ (x)O(0)|0 (|x0 | − ε) (10.34) are convergent. O(0)]|0 (10. T jµ (x)O(0) is regular. and similarly for ∂ µ jµ (x). From (10. a 1 finite). Similarly. with O(1/x 3 ) singularity the integrals in (10. One can immediately convince oneself that the limit exists if 1 T jµ (x)O(0) ∼ Cµ (x)O1 (0) + less singular terms (10.33) or (10. we may consider the Fourier transform d4 x exp(i px)∂ µ 0|T jµ (x)O(0)|0 = lim ε→0 ε→0 d4 x [−∂ µ exp(i px)] 0|T jµ O|0 (|x0 | − ε) = − lim + lim ε→0 d4 x exp(i px) 0|T j0 (x)O(0)|0 [δ(x0 + ε) − δ(x0 − ε)] d4 x exp(i px) 0|T ∂µ j µ (x)O(0)|0 (|x0 | − ε) (10.4) [ jµ (x. Integrating by parts we get − lim −ε −∞ ε→0 ε→0 + ∞ ε dx0 d3 x ∂µ f (x) 0|T j µ O|0 = − lim + lim ε→0 d3 x f (x) 0|O(0) j0 (x. 0)δ(x) (10.33) and (10. Indeed.

e. This subtractive renormalization of the Green’s functions comes in addition to the multiplicative renormalization of operators. These are the so-called Schwinger terms.s.34).11) we discuss the renormalization properties of currents and their divergences. The finite constant c is to be fixed by some renormalization conditions. Renormalization of currents As the first application of (10. the following specific case of (10.11): µ a ∂(x) 0|T jµ (x) b a (y) ¯ c (z)|0 = 0|T ∂ µ jµ (x) a − δ(x − y)Tbd b (y) ¯ c (z)|0 0|T d (y) ¯ c (z)|0 b a + δ(x − z)Tdc 0|T (y) ¯ d (z)|0 (10.33) and (10. let T A(x)B(0) ∼ 1 2 − iε x 2 1 + ··· l Performing Wick’s rotation and integrating in Euclidean space we conclude that for a well-defined ε → 0 limit in the l. as if the Schwinger terms were absent. integral in (10.4).h. In other words by suitable renormalization the Schwinger terms can be cancelled by subtraction terms. Depending on the nature of the singularity.10.1 Internal symmetries and Ward identities 351 their equal-time commutators (see Section 7. However.11). Our last remark is that in applications of Ward identities to physical problems one should pay attention to the fact that part of their content may be subtraction dependent. Subtractions are of the form c1 δ(x) + c2 ∂(x) δ(x) + · · · possibly with divergent ci .33) a subtraction of the form iπ 2 ln εδ(0) + cδ(0) is necessary. i. for instance.38) . For instance. Let us consider. The path integral approach teaches us that when regularization and renormalization schemes exist which preserve the classical symmetry of the lagrangian it is possible to define the renormalized T -products so that they satisfy Ward identities in a form like (10. without subtractions these integrals may depend on the order of integration over x0 and x (and be then non-covariant) or may be divergent. at the same time subtractions at x = 0 are necessary to have a well-defined limit when ε → 0 in the integrals in (10.

even if the symmetry is broken explicitly but only softly the currents and their divergences are not subject to renormalization. exclude subtractive renormalizations for time-ordered products of such currents.352 10 Spontaneous and explicit global symmetry breaking We take (10. as discussed before.s.41) In the case of explicit symmetry breaking by soft (d < 4) terms one has ∂ jµ (x) = 0 but. of course. Writing in a a a a general jµB = Z j jµR and ∂ µ jµB = Z D ∂ µ jµR we first conclude. arguing as before. vanishes (explicit symmetry breaking accompanies the spontaneous symmetry breaking and therefore there are no massless bosons in the spectrum a coupled to jµ currents). for example.44) Absence of wave-function renormalization for conserved or partially conserved currents does not. In the symmetry limit a ∂ µ jµ = 0 a B (10. Next.42) a for the divergence of the axial currents j5µ we have (from the equations of motion) a ∂ µ j5µ = 2mi ¯ T a γ5 (10. a jµB = ¯ B T a γµ B = Z 2 ¯ R T a γµ R a = jµR (10. Fourier transforming (10.s.38) by the renormalized fields and dividing both sides by Z 2 we immediately conclude from the finiteness of the r.38) and taking the limit p = 0 ( p is conjugate to x) we get Z D = 1 since in the limit p = 0 the Fourier transform of the l.h. In conclusion.45) 5 (10. as discussed earlier.40) can still be assumed. For instance.43) and the non-renormalization theorem implies Z m = Z −1 ¯γ where ( ¯ γ5 T a )B = Z ¯ γ5 ( ¯ γ5 T a )R (10. for chiral symmetry broken by a mass term µ L1 = −m ¯ (d = 3) (10. that Z j = Z D . Thus the conserved currents are not subject to renormalization.40) where Z 2 is invariant under symmetry transformations.38) that a a jµB (x) = jµR (x) up to a finite multiplicative constant which is fixed as one by the current commutation relations. Expressing the bare fermion fields in (10. (10.39) a R = 1/2 Z2 (10. of (10.h.38) to be for bare quantities. .

43a) µ Instead. that the Ward identity (9. one can now calculate masses of the would-be (pseudo-) Goldstone bosons in terms of the explicit symmetry breaking parameter: quark mass m. Several techniques have been developed to expand the bound state masses. It is worth mentioning at this point that the subject of chiral perturbation theory is not limited to the problem of the pseudoscalar meson masses but covers the whole low energy physics of pions and kaons. Such calculation is based on perturbative treatment of the symmetry breaking interaction (chiral perturbation theory) but takes account of all orders of the interactions described by the chirally symmetric part L0 of the lagrangian.97) and (9.98) 0| and 0|Aa (x)|π b ( p) = i f π δ ab pµ exp(−i px) µ respectively. Gasser & Leutwyler 1984. Pagels 1975). Green’s functions and any other quantity in the low energy pseudoscalar meson sector (which in the limit of exact chiral symmetry depend on only two parameters f π and ¯ ) in powers of the light quark masses m u .2 Quark masses and chiral perturbation theory Simple approach Using the Ward identity for the Green’s function involving the axial current Aa (x) µ and the pseudoscalar density a (x) we have proved in Section 9.101) between the parameters f π and ¯ of spontaneous breaking of chiral symmetry Z fπ = − ¯ (9. Dashen & Weinstein 1969.2 Quark masses and chiral perturbation theory 353 10. Then the proof of Goldstone’s theorem breaks down because ∂ µ Aa (x) = 2mi ¯ T a γ5 = 0 (10.5 that in chirally symmetric QCD with chiral symmetry spontaneously broken there exist massless Goldstone particles in the physical spectrum. a (x)|π b ( p) = Z δ ab exp(−i px) . using the OPE for T Aa (x) b (y).) Imagine now a realistic case in which the chiral symmetry is also broken explicitly by the mass term (10. (In view of our discussion in the previous section check. m d and m s (Dashen 1969.101) where the constants f π and Z are defined by (9. b) relies on the effective lagrangian technique (see Chapter 14). The most systematic approach proposed in a very interesting series of papers (Leutwyler 1984. 1985a.101) is free of any subtractiondependent parameters.10.91) does not need subtractions µ and therefore in the chiral limit the relation (9.42). The same Ward identity gave us the relation (9.

4.97) and (9. Our discussion is always in terms of renormalized quantities. For a mass-independent renormalization procedure ratios of the renormalized masses do not depend on µ. Approach based on use of the Ward identity The result (10.46) Working to the first order in m we can now use (9. (10. Finally we would like to stress that the quark masses we talk about are the renormalized quark mass parameters of the lagrangian (the so-called current quark masses).354 10 Spontaneous and explicit global symmetry breaking Here we shall only be concerned with the pseudoscalar meson masses calculated in the first order in the quark mass m. This approach can be used in other . The pion mass has been expressed in terms of the parameters of chiral symmetry breaking: spontaneous 0| ¯ |0 and explicit m. The result (10.47) can be extended to the case of three flavours with m u = m d = m s (for heavy flavours the breaking of chiral symmetry by the mass terms is too strong to be reliably considered as a perturbation). as discussed in Section 10. and operators ¯ γ5 T a and ¯ are also renormalized at the same point. We recall that interactions described by the chirally invariant part L0 of the lagrangian are included to all orders in our discussion. valid in the chiral limit. where µ is some renormalization point. to get 2 m 2 = −(1/ f π )2m 0| ¯ |0 (10. to include the electromagnetic mass corrections which will be considered in Section 10. m ¯ are renormalization-scheme-independent. 0|uu|0 = 0|dd|0 = 0|¯ s|0 To extract reliable values for the quark mass ratios we need. The answer is almost immediate. in SU (2) × SU (2) symmetric QCD with = (u. d) and m u = m d = m.101). However.48)  m 2 0 = m 2 0 = (m d + m s )C + · · ·  ¯  K K    2 ¯ ¯ ¯ s C = −(2/ f π ) 0|uu|0 . We get the following:  2 m 2 0 = m 2 ± = (m u + m d )C + O(m q )  π  π    2  m K+ = (m u + m s )C + · · · (10.1 the products m ¯ γ5 T a . for example.43a) sandwiched between the vacuum and the one-pion state together with the definitions (9. It is the pion mass.98) imply the exact relation f π m 2 = 2m Z π (10.47) can also be derived using a Ward identity directly for the realistic case with the chiral symmetry broken explicitly. Eq. however. Therefore m = m(µ).47) π which is the desired result valid to the lowest order in chiral perturbation theory.

H (0)]]|0 5 5 (10.49) where ab D5 (x) = 0|T ∂ µ Aa (x)∂ ν Ab (0)|0 µ ν The divergence ∂ µ Aµ is soft: the identity (10.s. where Aa (x) is the axial current and ∂ ν Ab stands µ ν µ ν for 2mi ¯ γ5 T b . ab Defining the Fourier transform ab (q) of the D5 (x) 5 ab 5 (q) =i ab d4 x exp(iq x)D5 (x) (10. In general ab (q) is given by the pole and the continuum contribution 5 starting at s = 9m 2 π ab 2 5 (q ) = 2 f π m 4 δ ab π + m2 − q2 π ∞ ds (3m π )2 ρ(s. can also be written as ab 5 (q = 0) = − 0|[Q a (t = 0).50) as the spectral µ π .51) which in a general case. t). ∂ ν Ab (0)]|0 5 ν (10. m 2 ) π s − q2 (10. of (10.52) For us H = m ¯ .52) can be used to calculate the would-be Goldstone boson (pion) masses in the first order in chiral perturbation theory if we approximate the Fourier transform ab (q) by the 5 pion pole.10.53) We have used 0|∂ µ Aa (0)|π b = f π m 2 δ ab and have written (10.49) is zero because in the present case (explicitly broken symmetry) there are no massless bosons in the physical spectrum and therefore no poles at q 2 = 0.2 Quark masses and chiral perturbation theory 355 problems too and its generalizations are used to calculate higher order corrections in the chiral perturbation theory.23) a ∂ µ jµ (x. using (10. ∂ ν Ab (0)]|0 µ ν 0 ν (10.23) The Ward identity reads µ ab ∂(x) 0|T Aa (x)∂ ν Ab (0)|0 = D5 (x) + δ(x0 ) 0|[Aa (x). In the limit q → 0 the l. The starting point is the Ward identity for the Green’s function 0|T Aa (x)∂ ν Ab (0)|0 .h.23). [Q b (t = 0).50) Fourier transforming both sides of (10.49) can then be regarded as an identity for the renormalized Green’s functions. similar considerations hold for any other current corresponding to a generator broken both spontaneously and explicitly when according to (10. Q a (t)] (10. t) = i[H (x.49) and taking the limit q → 0 one gets the following relation: 0= ab 5 (q = 0) + i 0|[Q a (t = 0). The relation (10.

56) can be written as | ( ) = exp(−i a X a )|0 (10. H (0)]]|0 5 5 2 fπ (10. respectively. If the vacuum |0 is left invariant by H then | (g) is left invariant by the equivalent G-rotated subgroup g H g −1 . The set of degenerate vacua is isomorphic to the coset G/H space.3 Dashen’s theorems Formulation of Dashen’s theorems We have learned in Section 9. for example.53) does require subtractions.47).57) Transformations in G/H space correspond to variations of the vacuum expectation values for which the effective action is level at its minimum.55) where f π and |0 should be taken. as in the chiral limit.54) The leading term in m 2 is given by the pion pole because this is the only term where π in the chiral symmetry limit the zero of the numerator in (10.56) where U (g) is a unitary operator corresponding to a group element g. 10. broken and unbroken by the |0 then the transformation (10. it can be seen from the OPE (see Chapter 7) that the product T ∂ µ Aa (x)∂ ν Ab (0) µ ν has a singularity O(m 2 x −6 ).356 10 Spontaneous and explicit global symmetry breaking integral using the standard technique (see. However.90).54) it is O(m 4 ) and does not affect the first order π results. so expansion in m 2 is the chiral perturbation theory.53) partially cancels with the denominator. from commutation relations (9. The pion mass measures the departure from chiral symmetry induced by perturbation H (x). For instance. we easily get (10. [Q b (t = 0). in higher orders the subtraction constants appear as additional free parameters. In our specific problem. ab It is also worth observing that the Green’s function D5 (x) or equivalently the integral in the spectral representation (10. The subtraction ambiguity is O(m 2 ) or according to our discussion following (10. for consistency. π In the order O(m 2 ) we get the so-called Dashen’s relation π δ ab m 2 = − π 1 0|[Q a (t = 0). This set of vacua is generated by the action of the G transformations on the state |0 | (g) = U (g)|0 (10. Bjorken & Drell (1965)). For q 2 = 0 one gets ab 2 5 (q 2 = 0) = f π m 2 δ ab + O(m 4 ) π π (10. Zero mass particles .4 that if a global symmetry G of a theory is spontaneously broken by the vacuum state |0 to a subgroup H then there is a set of degenerate vacua corresponding to different orientations of H in G. If we introduce generators X a and Y a which are.

60) =0 = − 0|[X b .30) we get 0|π|0 = 0 −ε 0|σ |0 ≥ 0 (10.36). to the previous derivation for the fixing of the normalization factor.59) and (10. From the commutation relations (9.60) ensures that in the first order 2 perturbation in H (x) the pseudo-Goldstone boson masses m ab are positive. the Goldstone boson tadpoles vanish.10.60) give for the σ -model with chiral symmetry explicitly broken by the H (x) = +εσ term. however. The criteria for identifying the correct vacuum in the presence of a small perturbation H (x) are called Dashen’s theorems. [X a .3 Dashen’s theorems 357 are quantized excitations – one for each orthogonal direction in G/H (see (9.59) also follows from the requirement of the translational invariance of the vacuum: the vacuum value of a divergence then vanishes. and generally speaking. So we rederive Dashen’s formula (10. Eq.55) we also rederive the result (9. Using the normalization given in (10. ensure the vacuum stability. E( ) should have a minimum (at least a local one) for = 0. Q a ]. We therefore get the following two conditions: a a E( ) = ( )|H (0)| ( ) = 0| exp(i X a )H (0) exp(−i ∂ ∂ ∂2 ∂ a∂ b a E( )| E( )| =0 = i 0|[X a . H (0)]]|0 ≥ 0 a In view of the relation ∂ µ jµ (x) = i[H (x). in agreement with (9. (10.e.51) and Chapter 14). Let us first see what the conditions (10.61) for the correct vacuum.60) has a straightforward interpretation too: up to the renormalization factor the second derivative of the potential with respect to is the Goldstone boson mass matrix (see also Chapter 14) and (10. .39) for the pseudo-Goldstone boson mass.59) (10. condition (10. (10. A small symmetry breaking perturbation H (x) may change the above picture in a qualitative way: it may lift the degeneracy of the vacuum. H (0)]|0 = 0 (10.59) says that π a |H |0 = 0. Since charges X a which do not annihilate the vacuum couple to Goldstone bosons 0|X a |π a = 0. To the leading order in the perturbation H (x) the energy density in each of the degenerate vacua is given by X a )|0 (10.58) To identify |0 as the true vacuum in the presence of the perturbation H (x).55) referring. i.

358 10 Spontaneous and explicit global symmetry breaking For the SUL (2) × SUR (2) symmetric QCD lagrangian with chiral symmetry + broken spontaneously to SU (2) by a condensate UR UL . In particular we recover (9. Some of these ideas will be discussed in Section 11.86) which we have taken as the definitions of the vacuum in the chiral limit and in a specified basis for the chiral fields. Explicit symmetry breaking lifts the degeneracy and leaves (10.62) as the only possibilities. the chiral symmetry which is spontaneously broken. for example.65) . Dashen’s conditions and global symmetry broken by weak gauge interactions A physically very interesting case of an explicit global symmetry breaking by a small perturbation is the following one: consider a strongly interacting theory (a gauge theory) with some exact global symmetry G.62) This follows from the commutation relation (9. operators corresponding to quantum numbers commuting with G are also possible but irrelevant for our discussion.87) and explicitly by the mass term H = m ¯ one gets 0| ¯ γ5 |0 = 0 − 0| ¯ |0 > 0 (10.3. Dashen’s conditions find very interesting applications in models of the dynamical breaking of the gauge symmetry of weak interactions (technicolour and extended technicolour models).61) and (10. Here we give a general technical introduction to such problems which will also be useful in the next section where we calculate the π + − π 0 mass difference. Imagine then that some of the G-symmetry currents couple to some of the gauge bosons of the group G W of weak interactions (in the rest of this section all sums over the group indices are explicitly written down) L1 = − α µ g α Aα (x) jα (x) µ (10. In the first non-vanishing order of perturbation theory in g α the G-symmetry breaking term of such a theory is given by the following effective hamiltonian: H (0) = − α. where is defined by (9. where = v1 as in l (9.84).63) µ where jα (x) are linear combinations of some of the G currents.88) and from the relations ¯ = Tr[ + † ] and ¯ γ5 = Tr[ − † ].β 1 α β g g 2 d4 x µν α β αβ (x)T jµ (x) jν (0) (10.64) where µν αβ (x) is the G W gauge boson propagator µν αβ (x) = i 0|T Aµ (x)Aν (0)|0 ≡ δ αβ α β µν (x) (10.

65) (remember that µν (x) ∼ 1/x 2 for x → 0) that these are operators of dimension ≤ 4.59) and (10.64) converges without subtractions which are non-singlets under the group G. The problem can be studied by means of the operator expansion of the product of the two currents in (10. do not require renormalization and are irrelevant for our discussion. First of all the counterterms affected by H (0) are easily enumerated (by assumption H (0) is used only in the first order perturbation): they involve only those operators of the expansion which are accompanied by coefficients that produce non-integrable singularities.64).67) (α is the G W index. The G W currents are linear combinations of the G currents µ gα jα (x) = a a µ gα ja (x) (10. Because the vacuum is H -invariant c d (G → H spontaneously) only the H -invariant part of the product jµ jν contributes to (10. Among the operators of dimension 4 there may be singlets under G which are again irrelevant. that this does not necessarily happen. We assume here that the expression (10. however.c a gα R ac ( ) jcµ (x) (10. We recall that (10. The non-singlets with d = 4 would affect our discussion but in several interesting cases they do not appear (see. It turns out.1 based on the assumed softness of H apply to this case when the symmetry breaking term H (0) is not soft (d = 6). Dashen’s conditions (10. a is the G index and there is no sum over α on the l.4 for the breaking of the chiral group by H given by the product of two electromagnetic currents). With perturbation H (0) given by (10.66) where U is any unitary operator representing the group G. Preskill 1981).66) expressed in terms of (10. for example. Section 10.h. For any given vacuum from the degenerate set the generators of G split into two groups: the generators Y a such that Y a |0 = 0 .58) in terms of the G currents (Peskin 1981.64).3 Dashen’s theorems 359 The question is whether our considerations in Section 10.64) together give the following expression E( ) = − 1 2 d4 x µν (x) α 2 α α gα 0|T U † jµ (x)UU † jν (0)U |0 (10. It is clear from (10.68) where R( ) is the adjoint representation of G.10. notice the difference in the case if H (0) was considered in any order of perturbation theory.s.58) and (10. The higher dimension operators in the α β expansion of jµ jν do not cause a non-integrable singularity at x = 0.68).) and consequently µ gα U † jα (x)U = a a µ gα U † ja (x)U = a.60) can be rewritten in a more useful form if we express the vacuum energy density E( ) given by (10.64) and (10. and at first glance it appears that it could affect the wave-function renormalization of fields.

by c d Schur’s lemma the only H -invariant term in the product Y jµ Y jν is proportional to δ cd : c d 0|T Y jµ (x)Y jν (0)|0 = δ cd 0|T Y jµ Y jν |0 = Tr[Y c Y d ] 0|T Y jµ Y jν |0 (10.68) can be split into the Y part and the X part µ µ µ U † gα jα (x)U = U † gα jα (x)U |Y + U † gα jα (x)U |x µ µ ≡ Y jαU (x) + X jαU (x) (10.R a b Eqs. It can be shown (Peskin 1981) that this restriction implies the existence of a parity operation P which preserves the Lie algebra of G. there is no summation in T Y jµ Y jν .73) . The decomposition in (10. Since Y a span a single real representation of H (the adjoint representation).70) For chiral symmetry (10. We will limit our discussion to cases in which in the product X a X b also the only H -invariant term is δ ab .360 10 Spontaneous and explicit global symmetry breaking and X a where X a |0 = 0. From the orthogonality of X a and Y b (Tr[X a Y b ] = 0) ˆ ˆ ˆ ˆ Tr[YJαU Y JαU ] = Tr[YJαU JαU ] ˆ ˆ ˆ ˆ Tr[ XJαU X JαU ] = Tr[ XJαU JαU ] Using in addition an obvious relation ˆ ˆ ˆ ˆ ˆ ˆ Tr[YJαU JαU ] = Tr[ JαU JαU ] − Tr[ XJαU JαU ] (10.74) (10.L L. Tr[Y a X b ] = 0).70) also imply that the joint expectation values of one Y jµ and one X jν are zero by parity conservation. each G current carries index Y or X . (10.72) ˆ ˆ + Tr[ XJαU X JαU ] 0|T X jµ (x) X jν (0)|0 } ˆ where JαU is the linear combination of the G generators corresponding to the α current U † gα jµ (x)U with gα included. Correspondingly. of course. R. such that PY a P −1 = Y a . true since P Q a P −1 = Q a .71) (the last line defines our notation) such that only Y (X ) currents from the sum a ac c a.69) (here we normalize the broken and unbroken generators to Tr[Y a Y b ] = a Tr[X a X b ] = δ ab . P X a P −1 = −X a (10. Taking all this into account we obtain for E( ) the following: E( ) = − 1 2 d4 x µν (x) α ˆ ˆ {Tr[YJαU Y JαU ] 0|T Y jµ (x)Y jν (0)|0 (10.70) are.c gα R jµ contribute to the first (second) component. where Y jµ denotes any single current Y jµ .

Assuming that the coefficient of the last trace in (10. One sees that the preferred vacuum orientation corresponds to the minimal number of ˆ broken generators in the decomposition of J α into the generators of G. Writing U = exp(−i a X a ) one has ∂ ∂ a α ˆ Tr( XJαU )2 |U =1 = 2i (10. for projections of transformed currents on the subspace of the original broken generators.e. as before. any single current corresponding to an unbroken . from (10.59) and (10. orthogonality of Y and X and the trace property Tr[[A. one gets the preferred vacuum simply by minimizing the quantity α ˆ ˆ Tr[ XJαU X JαU ] ˆ i. The second Dashen’s condition reads 2 m ab = α ˆ ˆ Tr[ X [X a . X Jα ] = 0 for all αs.78) and Y jµ ( X jµ ) denotes. Jα ] XJα ] = 2i α ˆ ˆ Tr[X a [YJα .75) 2 M 2 = (1/ f π ) d4 x µν (x) 0|T [Y jµ (x)Y jν (0) − X jµ (x) X jν (0)]|0 (10. B]C] = Tr[A[B. We can rewrite Dashen’s conditions (10.3 Dashen’s theorems 361 we finally get E( ) = − 1 2 −1 2 × α d4 x d4 x µν (x) α ˆ ˆ {Tr[ JαU JαU ] 0|T Y jµ (x)Y jν (0)|0 } µν (x) 0|T [ X jµ (x) X jν (0) − Y jµ (x)Y jν (0)]|0 (10. the vacuum orientation is a stationary point of E if ˆ ˆ [YJα . X Jα ]] ∂2 1 2 ∂ a∂ b α ˆ Tr( XJαU )2 |U =1 × M 2 ≥ 0 (10.75) ˆ ˆ Tr[ XJαU X JαU ] The first term in the sum is -independent (without subtraction it may even be infinite) and in the second term -dependence has been factored out from the current matrix elements.60) in the present notation.77) where.76) where the last result follows from the commutation relations [X. C]]. by minimizing the trace for projections of currents J α on the subspace of a a possible equivalent representations of the X and Y or. X ] = iY following from (10.75) is negative as can be argued in some cases. regarding the vacuum |0 as fixed so that the rotation is applied to the perturbation H . Thus.10.70).

[YJα . As mentioned. The expressed as follows ∂2 ∂ a∂ b -dependence is only in the trace which can be α α ˆ T r ( XJαU )2 |U =1 ˆ ˆ ˆ ˆ {Tr[ X [X a .64). The effective hamiltonian is given by (10.79) and of the integral (10. Another source of such a breaking is the electromagnetic interaction of quarks which couples the T 3 generator of the isospin group to the electromagnetic gauge field: Q = T 3 + Y . Jα ] X [X b . including the vacuum. Only scalar operators contribute to the µν integral in (10. This section is devoted to a study of the electromagnetic contribution to the pion masses.2 we have studied an explicit chiral symmetry breaking in QCD caused by non-zero current quark masses. 10. the only d = 4 operator (G α G µν ) is a singlet under µν α the chiral group and our considerations of Sections 10. The latter can often be studied by means of the spectral function sum rules. [ XJα . The quark masses will be neglected in the calculation of the electromagnetic masses in the spirit of the first order perturbation in any explicit symmetry breaking term. X α ]]X b ] Dashen’s condition (10.1 and 10. Several applications of the results of this section will be discussed in the next section and in Section 11.3 apply.3 we notice that the expansion of the operator product T jµ (x) j µ (x) reads T jµ (x) j µ (x) = C1 (x)1 + l q Cq (x)m q qq + C G (x)G α G µν ¯ µν α + operators with d > 4 + non-scalar operators (10.64) with jµ (x) being the electromagnetic current and µν the photon propagator.79) = −2 =2 α ˆ ˆ ˆ ˆ Tr[[YJα . [X b . X a ]]X b − [ XJα . Thus.80) where G α is the gluon field strength. Following our discussion in Section 10.3. Jα ]]} (10. We also see that in the chiral limit m q = 0 all the divergent terms in Hem are singlets under the chiral group and may be disregarded: they induce a universal energy shift for all states.362 10 Spontaneous and explicit global symmetry breaking (broken) generator of G.4 Electromagnetic π + –π 0 mass difference and spectral function sum rules Electromagnetic π + –π 0 mass difference from Dashen’s formula In Section 10.60) or equivalently the positivity of the pseudo-Goldstone bosons mass matrix depends on the signs of the trace (10. in this case the electromagnetic pion mass .78). Jα ]] XJα ] + Tr[ X [X a .

π (m 2 0 )γ = 0 π (10.81) where M 2 can be explicitly rewritten in terms of the vector and axial currents 2 M 2 = (1/ f π ) d4 x µν 3 3 (x) 0|T [Vµ (x)Vν (0) − A3 (x)A3 (0)]|0 µ ν (10. Lo Secco & Weinberg 1975).77) in the previous section). ˆα ˆα Using (10. then J X = 0 and Dashen’s condition for the stationary vacuum is satisfied in the presence of the electromagnetic interactions) we immediately get the following result for the one-photon exchange contribution to the pion mass: (m 2 + )γ = e2 M 2 . Let us consider the K¨ llen–Lehmann spectral representation (Bjorken & Drell 1965) for the timea ordered product of two currents ∞ 0 a b 0|T jµ (x) jν (0)|0 = dµ2 i d4 k exp(−ikx) 2 ρ ab (k.82) was derived by Das et al. originally derived by Weinberg (1967a) for the SU (2) × SU (2) chiral symmetry. Formula (10. possibly infinite.e. i.83) .79) and the fact that JY = eT 3 . k 2 = µ2 ) (2π )4 k − µ2 + iε µν (10. for m q = 0. Non-zero quark masses introduce second order corrections O(αm q ) which need renormalization and are therefore renormalization-scale-dependent. Duncan.6 we define the vacuum of the SUL (2) × SUR (2) ˆα symmetric QCD so that the SUL+R (2) remains unbroken.77). (10. The calculation can be performed by means of Dashen’s formula.81) and (10. (1967). Spectral function sum rules Spectral function sum rules.82) We recall again that all. singlet contributions to the electromagnetic masses have been disregarded in (10.82) (see the derivation of (10.10. We also remember that both equations are valid only in the first order in the symmetry breaking.4 π + –π 0 mass difference 363 difference is finite without renormalization. To calculate the integral (10. J X = 0 for the electromagnetic current (as in Section 9.82) one can use the spectral function sum rules. can be discussed in a general way by invoking the OPE (Bernard.

87) This formal expansion can be meaningful only when the Fourier transform a b d4 x exp(ikx) 0|T jµ (x) jν (0)|0 behaves for large k 2 as O(1) or softer.87) must vanish. If one constructs a linear combination of currents such that the Fourier transform (10. = d x exp(ikx) 0|T 4 a b jµ (x) jν (0)|0 =i 0 ∞ dµ 2 ab ρµν (k. k 2 = µ2 ) k 2 − µ2 (10.364 10 Spontaneous and explicit global symmetry breaking where the spectral function ρ can be decomposed into the spin-one and spin-zero parts: (2π)3 n spin 1 a b 0| jµ (0)|n n| jν (0)|0 δ(k − kn ) = −(gµν − kµ kν /µ2 )ρ (1)ab (µ2 ) (10. Otherwise the coefficients of the expansion must be divergent.86) is for large k 2 softer than O(1).83) F.T. In looking for such products one can use the global symmetry of the lagrangian which is respected by the OPE coefficient functions even if the symmetry is spontaneously broken.88) 0     ∞   (0)  ds sρ (s) = 0  0 ab and these are the spectral function sum rules.86) and expanding it formally in powers of 1/k 2 we get F. = i kµ kν k2 ∞ 0 ∞ 0 dµ2 (1) ρab (µ2 ) gµν (0) + ρab (µ2 ) − i 2 2 µ k ∞ 0 (1) dµ2 ρab (µ2 ) kµ kν +i 2 2 (k ) (1) (0) dµ2 [ρab (µ2 ) + µ2 ρab (µ2 )] + · · · (10. for softer than (1/k 2 ) behaviour  ∞ (1) (0)  ds [ρab (s)/s + ρab (s)] = 0     0   ∞  (1) ds ρab (s) = 0 (10.T. The OPE is useful in finding appropriate combinations of current products with soft high energy behaviour. for example. then the first few terms in the expansion (10.85) Taking the Fourier transform of (10.84) a b 0| jµ (0)|n n| jν (0)|0 δ(k − kn ) = kµ kν ρ (0)ab (µ2 ) (2π)3 n spin 0 (10. In our problem we are interested in .

This operator must be Lorentz-invariant. 3) under SU (2) × SU (2). in the expansion of the two-current product we need one power of the symmetry breaking parameter m for the coefficient function in front of the operator qq to be non-zero. 2) ⊕ (2. gauge-invariant. ab 2 −2 Therefore. Spectral function sum rules can be derived by studying the high momentum limit of G ab (k) = µν a b d4 x exp(ikx) 0|T jµL (x) jνR (0)|0 which transforms as the adjoint representation (3.4 π + –π 0 mass difference 365 the group G = SUL (2) × SUR (2) broken to the H = SUL+R (2). According to (10. i. 3) representation. the symmetry is broken by the mass ¯ term m qq. it must transform as the adjoint representation (3.e. However. H -invariant. Representing the symmetry breaking interaction by a spurion which ¯ ¯ ¯ must also belong to (2. Taking into account that 0|Aa |π b = ikµ f π δ ab µ and therefore (0) 2 ρAab (µ2 ) = δab f π δ(µ2 ) + · · · (0) and neglecting any continuum contribution to ρA which should be small we get from (10. The second sum ¯ rule is then no longer valid: the coefficient of this m qq term scales as x −2 . a singlet under SUL+R (2).10. This can be seen by a spurion analysis (Wilson 1969a): the ¯ current product belongs to (3. because the expansion coefficient functions respect the G symmetry. 2). 2) ⊕ (2. 3) under SU (2) × SU (2).88) hold for the ab spectral functions ρLR . whereas the operator ¯ ¯ qq belongs to (2. The lowest-dimension operator satisfying these requirements is ¯ ¯ a four-fermion operator of dimension (mass)6 ( ¯ transforms as (2. The asymptotic behaviour of G ab (k) is determined by the lowest-dimension µν a b operator in the expansion of jLµ jRν which has a non-zero vacuum expectation value. so its ¯ . 3) of the SU (2) × SU (2). and.88) the following sum rules:  ∞ (1) (1) 2  (ds/s)[ρV ab (s) − ρAab (s)] = f π δab   0 (10.89) ∞   (1) (1)  ds [ρ (s) − ρ (s)] = 0 0 V ab Aab which are valid in the chirally symmetric theory. 2) ⊕ (2. 2)). Thus.82) we are interested in the combination VV − AA = 4 jL jR . If the chiral symmetry is broken by non-zero quark masses the OPE for the difference VV − AA contains a term proportional to m qq. 2) we see that combining one spurion with the qq ¯ operator one can produce the (3. G µν ∼ (k ) up to logarithms and the sum rules (10.

one (ρ) coupled to the vector current and the other (A1 ) to the axial current. Taking 2 ρV (s) = f ρ sδ(s − m 2 ) ρ 2 ρA (s) = f A1 sδ(s − m 2 1 ) A and using both sum rules (10.89) one gets 2 2 2 f ρ = f A1 + f π . µ2 1 1 1 + ln 0 + const. The remaining finite term gives (m 2 + )γ = π 3α 4π( f π )2 ∞ ds ln 0 µ2 (1) (1) 0 [ρV (s) − ρA (s)] s (10.91) However. however. (4π)2 ε (4π )2 µ2 (10. Finally we can calculate (10.89) that the coefficient of the divergent term vanishes. expected to be small because the masses m u and m d are very small in comparison to the scale of QCD.89) it is renormalization point (µ2 ) 0 independent. The corrections are.92) assuming that it is reasonable to saturate the integral (10. using dimensional regularization.366 10 Spontaneous and explicit global symmetry breaking Fourier transform behaves as k −2 .92) with the two lowest-lying narrow resonances. (10. Introducing explicitly µν (x) = gµν d4 k exp(−ik x) (2π )4 k 2 + iε we get M2 = − i 2 fπ ∞ 0 (1) (1) dµ2 3[ρV (µ2 ) − ρA (µ2 )] d4 k 1 4 k 2 (k 2 − µ2 ) (2π ) (10.92) and on account of the second sum rule (10.82). This just reflects the already discussed fact that the first order electromagnetic contribution is finite after the singlet piece under SU (2) × SU (2) has been disregarded.90) The momentum integral is logarithmically divergent and gives. Results We are now ready to calculate M 2 . it follows from the second sum rule (10.

2 and of this section.6 MeV. we have. again using (10.10. corrections of order αm q require renormalization but are also expected to be small (Gasser & Leutwyler 1982). using the first equation (10. Finally. Therefore.93) To calculate the π + –π 0 mass difference we have to combine the results of Section 10.48).94) and (10.48). Although the error due to the resonance saturation is hard to estimate accurately this result should be regarded as at least in qualitative agreement with the experimental value 4.95) taken together can be used to estimate the quark mass ratios (Weinberg 1977).4 π + –π 0 mass difference 367 and 2 2 m 2 f ρ = f A1 m 2 1 ρ A Therefore we can eliminate the A1 parameters to get the final answer in terms of the ρ parameters: (m 2 + )γ = π 3α 2 m 4π ρ fρ fπ 2 ln 2 fρ 2 2 fρ − fπ (10. one gets m π + − m π 0 = (4. as already mentioned. π K m 2 + = C(m u + m s ) + (m 2 + )γ K π m 2 0 = C(m d + m s ) K (10.9 ± 0. Eqs. In the first order in both symmetry breaking perturbations. m 2 + = m 2 0 + (m 2 + )γ π π π where m 2 0 = C(m u + m d ).94) and the K+ –K0 mass difference calculated in the first order of the explicit chiral symmetry breaking is not purely electromagnetic but it also reflects a possible m u – m d difference. Corrections to this result due to the π 0 –η mixing caused by m u = m d can be estimated to be negligibly small (Gasser & Leutwyler 1982). π and therefore m π + − m π 0 ≈ (m 2 + )γ /2m π 0 π Using f π = 93 MeV and f ρ = (145 ± 8) MeV. (10. Completing our discussion given at the end of .95) 2 C = −(2/ f π ) 0|uu|0 ¯ (10.2) MeV. Extending our considerations to the SU (3) × SU (3) case we get (m 2 + )γ = K (m 2 + )γ and (m 2 0 )γ = 0. quark masses and electromagnetic interactions.

8 mu 2m 2 0 + m 2 + − m 2 0 − m 2 + K π π K m2 0 + m2 + − m2 + ms K π = K = 20.368 10 Spontaneous and explicit global symmetry breaking Section 10.2 we get m2 0 − m2 + + m2 + md K π K = = 1.1 md m2 0 − m2 + + m2 + K π K We stress that these results are valid in the first order of the chiral symmetry breaking. .

The lagrangian for a single self-interacting complex field L = ∂µ ∗ µ ∂ − λ( ∗ + m 2 /2λ)2 (11. For the real fields √ = (ϕ + iχ )/ 2 we can choose.11 Higgs mechanism in gauge theories 11. Gauge theories do not obey both requirements simultaneously.2) for m 2 < 0. Goldstone’s theorem does not hold and the so-called Higgs mechanism operates (see. for instance.1) = exp(−i ) . the theory contains states of negative norm. We shall consider the simplest example of the U (1) gauge theory. In a gauge in which the theory has only states of positive norm it is not manifestly covariant. The Higgs mechanism requires the presence of a scalar field coupled to the gauge field. The U (1) is invariant under global U (1) transformation: U symmetry is spontaneously broken (see Section 1. For a U (1) symmetric theory it has to be a complex scalar field. for instance. 0|ϕ |0 = v = (−m 2 /λ)1/2 . Abers & Lee (1973) and references therein). 0|χ |0 = 0 and defining the physical fields ϕ = ϕ − v.1 Higgs mechanism The proof of Goldstone’s theorem given in Chapter 9 for theories with spontaneously broken global symmetry requires Lorentz invariance and Hilbert space with positive-definite scalar products. In consequence. This example illustrates all the essential points. In a covariant gauge. 369 .2) 2 4 This lagrangian describes the interaction of the massive particle ϕ and the massless Goldstone boson χ. χ = χ we get L = 1 (∂µ ϕ∂ µ ϕ + ∂µ χ∂ µ χ) − v 2 λϕ 2 − 1 λ(ϕ 2 + χ 2 )2 − λϕv(ϕ 2 + χ 2 ) (11.

Choosing the gauge function to be −η (x) = −η(x)/v and using (11.2) we can write (x) − 0| (x)|0 = [ρ (x) − v] exp[−iη (x)T ] 0 1 (11. a lagrangian invariant under local gauge transformation is L = − 1 F µν Fµν + (Dµ 4 where Dµ = ∂µ + ig Aµ The gauge freedom has profound implications for the mechanism of spontaneous symmetry breaking. the particle spectrum is evident.6) where ρ =ρ −v In this gauge.370 11 Higgs mechanism in gauge theories It is also useful to introduce ‘angular’ variables and parametrize the field follows (see (9.4) and in terms of the fields ρ = ρ − v and η = vη the lagrangian (11. Now let us introduce a gauge field Aµ .5) Thus. the particle interpretation of the fields ρ and η is the same as that of ϕ and χ .3) Using the freedom in our choice of the vacuum expectation value among the U (1) degenerate set (Section 1.6) is gauge-invariant we can write it in terms of fields U and AU and then µ U L = − 1 F Uµν Fµν + 1 ∂µ ρ∂ µ ρ + 1 g 2 (ρ + v)2 AU AUµ − 1 λ(ρ 2 + 2ρv)2 (11.7)  AU (x) = Aµ (x) + (1/gv)∂µ η(x) µ Since the lagrangian (11.3). There is a scalar meson ρ with the bare mass (2λv 2 )1/2 and a massive vector meson with mass gv and no particle corresponding to the .4) we get (in the real basis)  0  U U (x) − 0| (x)|0 = [ρ (x) − v] 1 (11. T = 0 i −i 0 as (11.8) µ 4 2 2 4 )∗ (Dµ ) − λ( ∗ + m 2 /2λ)2 (11.1) reads L = 1 (∂µ ρ∂ µ ρ + ∂µ η∂ µ η) − v 2 λρ 2 + cubic and higher order terms 2 (11.78)) (x) = ρ (x) exp[−iη (x)T ] 0 1 = ρ sin η ρ cos η . This is easiest to see working with the angular variables. called the unitary gauge (introduced in Section 1.

11) where ( p 2 ) is defined by the gauge-invariant vacuum polarization tensor.9) One can now proceed in several different ways. Let us first take a gauge fixed by the standard gauge-fixing term −(∂µ Aµ )2 /2a. now massive. With the standard methods of Chapters 2 and 3 we get then the following free particle propagators (for example. Such divergences vanish in the S-matrix elements but this we know from the proof of renormalizability of a spontaneously broken gauge theory given in covariant gauges (Abers & Lee (1973) and references therein).12) gets a contribution from the remaining terms in the lagrangian and in particular . The η degree of freedom has been traded for the longitudinal component of the vector field. the gauge symmetry is spontaneously broken and no massless Goldstone boson remains in the physical spectrum. The Higgs mechanism is one of the fundamental ideas for condensed matter physics and it is also the crucial part of unified electroweak theory (see Chapter 12). One possibility is to take S0 as the free lagrangian of the theory with the corresponding definitions of the free particle states (which may not be physical states). Returning to the variables ϕ and χ we get from (11.6) the following lagrangian: L = − 1 F µν Fµν + 1 (∂µ ϕ∂ µ ϕ + ∂µ χ∂ µ χ) − λv 2 ϕ 2 − (1/2a)(∂µ Aµ )2 4 2 + 1 g 2 v 2 Aµ Aµ + gv Aµ ∂ µ χ 2 − g Aµ ∂ µ ϕχ + g Aµ ∂ µ χϕ + 1 g 2 Aµ Aµ (ϕ 2 + 2ϕv) 2 + 1 g 2 Aµ Aµ χ 2 − 1 λ(ϕ 2 + χ 2 )2 − λ(ϕ 2 + χ 2 )ϕv 2 4 (S0 ) ˜ ( S0 ) (SI ) (11. Because of the massive vector meson propagator which for large k grows like kµ kν /m 2 k 2 the Green’s functions contain divergences that cannot be removed by the renormalization counterterms. ≡i µν ( p 2 ) = i(gµν p 2 − pµ pν ) ( p 2 ) (11.1 Higgs mechanism 371 field η which has disappeared from the lagrangian. Quantization in the unitary gauge gives a theory which is manifestly unitary order-by-order in perturbation theory but not manifestly renormalizable. in the Landau gauge a → 0):  Aµ (massless) : −i(gµν − pµ pν / p 2 )/( p 2 + iε) ≡ −iDµν  (11.10) ϕ(massive) : i/( p 2 − 2λv 2 + iε)  2 χ(massless) : i/( p + iε) The full gauge boson propagator µν ( p) = −iDµν 1 1− ( p2 ) (11. Thus.11. Hence it is useful to extend our discussion to covariant gauges.

13a) (11.15) as the free boson propagator.17) 2ξ Then. This so-called Rξ gauge is in wide use. The coupling disappears from the theory as well as its previous interpretation. such an interpretation is certainly procedure˜ and gauge-dependent. ˜ S0 + f (Aµ . Take the gauge-fixing term as 1 f (Aµ . although quadratic in fields too. χ) = 1 g 2 v 2 Aµ Aµ − 1 ξ g 2 v 2 χ 2 − (1/2ξ )(∂µ Aµ )2 2 2 (11. integrating by parts when necessary. Indeed.18) . ˜ In the lowest order the S0 contribution is ≡ −i( pµ pν / p 2 )g 2 v 2 ≡ +ig 2 v 2 gµν Hence ( p2 ) = g2 v 2 / p2 and consequently µν ( p) (11. is counted as an interaction term. One sometimes says that the coupling (11. we can as well take S0 + S0 as our free particle µ lagrangian.16) Note also that the χ propagator remains massless in the Landau gauge.3 where other mechanisms for spontaneous gauge symmetry breaking are discussed.13b) (11. χ) = − (∂ µ Aµ − ξ gvχ)2 (11.14) = −i(gµν − pµ pν / p 2 )/( p 2 − g 2 v 2 ) (11.13a) of a gauge boson to a massless Goldstone boson is the origin of the gauge boson mass and of the spontaneous breaking of gauge invariance.15) which can be rewritten as a combination of the standard propagator for a massive vector boson and of a term corresponding to a scalar negative norm boson coupled to the source of the vector meson gradiently µν ( p) = −i gµν − pµ pν g2v2 p2 1 1 pµ pν −i 2 2 2 2 v 2 + iε −g g v p + iε (11. The presence of the mixing term gv Aµ ∂ χ leads to a system of coupled differential equations for Aµ and χ propagators and in the limit a → 0 we get (11. Secondly.372 11 Higgs mechanism in gauge theories ˜ from S0 which. In this context one should observe that the massless Goldstone boson contributes only to the pµ pν term in the µν but by gauge invariance the gµν term must then be present. Finally we can get rid of the coupling gv Aµ ∂ µ χ in the lagrangian by a clever choice of gauge. This remark will be particularly relevant in Section 11.

11.2 Spontaneous gauge symmetry breaking


and the free propagators are −i pµ pν gµν − (1 − ξ ) 2 2 + iε −M p − ξ M2 gµν − pµ pν /M 2 pµ pν /M 2 = −i 2 −i 2 p − M 2 + iε p − ξ M 2 + iε 2 2 χ : i/( p − ξ M + iε) Aµ : p2 ϕ: i/( p 2 − 2λv 2 + iε) where M = gv. In the limit ξ = 0 we get the Landau gauge and for ξ = 1 the so-called ’t Hooft–Feynman gauge. The unitary gauge is recovered in the limit ξ → ∞. For any finite value of ξ there are unphysical poles at p 2 = ξ M 2 in the gauge boson propagator and in the χ propagator. They cancel, however, in the S-matrix elements which are obtained from the Green’s functions by removing external lines, setting external momenta on the mass-shell and contracting tensor indices with appropriate physical polarization vectors. This can be shown, for instance, by proving that the renormalized S-matrix is independent of ξ (Abers & Lee 1973).

11.2 Spontaneous gauge symmetry breaking by radiative corrections In the model of the previous section gauge symmetry has already been spontaneously broken at the tree level. Here we discuss spontaneous gauge symmetry breakdown by radiative corrections (Coleman & Weinberg 1973). This can occur when elementary scalar fields of zero mass are present in the theory (Georgi & Pais 1977). In the tree approximation there is then no spontaneous symmetry breaking because in this approximation a non-zero vacuum expectation value of the scalar fields can appear only as a result of interplay between the 4 interaction terms and the scalar mass terms. Spontaneous symmetry breaking is, however, produced by non-zero vacuum expectation values induced by the higher order corrections. This is also an example of a dimensional transmutation: a dimensionful parameter 0| |0 is generated in a quantum field theory whose classical limit is described by a scale-invariant lagrangian. We consider first λ 4 theory with a massless meson field and discuss spontaneous breaking of global reflection symmetry → − . The lagrangian density for this theory written in terms of the renormalized quantities L = 1 (∂µ )(∂ µ ) − (λ/4!) 2

+ 1 A(∂µ )(∂ µ ) − 1 B 2 2


− 1/4!C



contains the usual wave-function and coupling-constant renormalization counterterms. Also a mass renormalization counterterm is present, even though we are


11 Higgs mechanism in gauge theories

studying the massless theory, because the theory possesses no symmetry which would guarantee vanishing bare mass in the limit of vanishing renormalized mass (the scale invariance is broken by anomalies). To study spontaneous symmetry breaking in such a theory we will use the effective potential formalism developed in Section 2.6. To the one-loop approximation the effective potential is given by (2.165) and (2.166) (with m 2 = 0 and with the contributions from the mass and coupling-constant counterterms included) V = λ 4 1 2 1 1 ϕ + 2 Bϕ + Cϕ 4 + 4! 4! 2 d4 k λϕ 2 ln 1 + 2 (2π )4 2k (11.20)

In this expression we have performed the usual Wick rotation so the integral is in Euclidean space: d4 k = 1 k 2 dk 2 d 4 , 4 = 2π 2 . Observe also that, even though 2 for m 2 = 0 the series in (2.166) is disastrously IR divergent, its sum exhibits only a logarithmic singularity at ϕ = 0. Cutting off the integral in (11.20) at k 2 = 2 (exceptionally we use the UV cut-off rather than the dimensional regularization method), performing the elementary integration and throwing away terms which vanish as 2 goes to infinity we obtain V = λϕ 2 λ 4 1 2 1 λ 2 2 λ2 ϕ 4 1 ln ϕ + − ϕ + 2 Bϕ + Cϕ 4 + 2 2 2 4! 4! 64π 256π 2 2 (11.21)

Imposing now the definitions of the renormalized mass and coupling constant we determine the value of the counterterms. Defining the renormalized squared mass of the field (x) as the value of the inverse propagator at zero momentum and recalling the interpretation following (2.164) of the derivatives of the effective potential we get d2 V /dϕ 2 |ϕ=0 = 0 (11.22) as the mass renormalization condition. We take the derivative at ϕ = 0 because we are interested in the inverse propagator of the original (not shifted) field (x); note also that the perturbative calculation of V (ϕ) makes use of the Feynman rules for the lagrangian written in terms of the field (x) and therefore it is based on (2.164) even though the theory may be spontaneously broken. From (11.21) and (11.22) we get B = −λ

/32π 2

A possible definition of the renormalized coupling constant would be to identify it with the value of the 1PI four-point function in the original fields (x) at zero external momenta d4 V /dϕ 4 |ϕ=0 = λ Unfortunately, for a massless field we encounter the logarithmic IR singularities.

11.2 Spontaneous gauge symmetry breaking


In the standard renormalization procedure one defines in such a case the coupling constant at some off-mass-shell position in momentum space. Using the effective potential formalism it is much more convenient to adopt an alternative definition, namely d4 V /dϕ 4 |ϕ=M = λ(M) (11.23)

where M is some arbitrary number with the dimension of mass. It is clear from our discussion in Section 2.6 (see (2.156)) that the fourth derivative of V (ϕ) at ϕ = M is the 1PI Green’s function, at zero external momenta, for the shifted fields (x) = (x) − M. In particular if we take M = 0| |0 it is the four-point function taken at zero external momenta for the physical massive field ¯ (x): 0| ¯ (x)|0 = 0. Imposing condition (11.23) on the effective potential V (ϕ) given by (11.21) we finally get V = λ(M) 4 λ2 (M)ϕ 4 25 ϕ2 ϕ + ln 2 − 2 4! 256π M 6 (11.24)

We see that the one-loop radiative corrections have turned the minimum at the origin ϕ = 0 into a maximum and caused a new minimum of V to appear at λ(M) ln( 0| |0 /M) = − 32 π 2 + O(λ) 3 (11.25)

Thus, the one-loop corrections have generated spontaneous breaking of reflection symmetry. However, in our example, the new minimum lies outside the range of validity of perturbation theory as higher orders will give higher powers of λ ln(ϕ 2 /M 2 ) which according to (11.25) is large for ϕ = 0| |0 . Physically interesting theories in which spontaneous symmetry breaking can be studied in perturbation theory are gauge theories with elementary massless scalar fields and possibly fermion fields coupled to the gauge fields. Let us consider the case of a single gauge field coupled to a charged scalar to see qualitatively what happens. In the one-loop approximation the effective potential of the scalar sector gets a contribution from the scalar loops which, aside from irrelevant numerical √ factors (there are now two real scalar fields in the theory: = ( 1 + i 2 )/ 2; ∗ 2 ) ; the effective potential can only depend on the scalar self-coupling is λ( 2 2 2 ϕ = (ϕ1 + ϕ2 ) because the theory is U (1)-invariant), have the same structure as before, and from the gauge field loops in Fig. 11.1 arising from the minimal coupling |(∂µ + ig Aµ ) |2 . The trilinear terms lead to diagrams such as

which vanish if we work in the Landau gauge, where the gauge boson propagator


11 Higgs mechanism in gauge theories

Fig. 11.1.

is Dµν = −i gµν − kµ kν /k 2 k 2 + iε

This is because we calculate diagrams with zero external momenta, and therefore the momentum of the internal scalar meson is the same as that of the internal gauge boson. It is obvious now that the gauge boson loop contribution to the effective potential has the same structure as the scalar loop contribution, so we get V = λ(M)ϕ 4 + [(1/16π 2 )λ2 (M) + bg 4 (M)]ϕ 4 [ln(ϕ 2 /M 2 ) − where b is some numerical factor. If it happens that λ(M) neglect the O(λ2 ) term and then dV /dϕ = 0 for | ϕ | = ( 0|
1 |0 2 25 ] 6

(11.26) 1 we can

g 2 (M)

+ 0|

2 |0

2 1/2


= M exp

1 λ(M) 11 − 6 2b g 4 (M)


Thus, in the one-loop approximation we find in our theory spontaneous breaking of the U (1) symmetry and, this time, the new minimum is in the range of validity of our approximation: due to our choice of the renormalization point such that λ(M) g 2 (M), (11.27) constrains only the value of ln( ϕ 2 /M 2 ) and for small λ and g the terms λ ln( ϕ 2 /M 2 ) and g 2 ln( ϕ 2 /M 2 ) are small. A non-vanishing field vacuum expectation value generated by radiative corrections is an example of the dimensional transmutation phenomenon. This can be seen more clearly if we choose M = ϕ . Then, from (11.27) λ( ϕ ) =
11 bg 4 ( 3

ϕ )


and we recall that M = ϕ corresponds to on-shell renormalization for the shifted fields. We can take dimensionless λ and g or g and dimensionful ϕ as independent parameters of our theory. Spontaneous gauge symmetry breaking leads to the physical consequences discussed in Section 11.1: the would-be Goldstone boson combines with the gauge

11.2 Spontaneous gauge symmetry breaking


boson to make a massive vector meson m 2 (V ) = g 2 ϕ 2 and the remaining scalar field, after shifting, corresponds to a scalar meson of mass m 2 (S) = V ( ϕ ) ∼ g4 ϕ 2. Physics cannot depend on the arbitrary parameter M but one choice of M can be more convenient than the others, particularly since we are working in perturbation theory and have to control higher order corrections. We have already learned that the perturbative calculation of ϕ is reliable when the renormalization point M is such that λ(M) g 2 (M) 1. It is also worth noting that from (11.27) ϕ = M0 e11/6 (11.29)

where λ(M0 ) = 0. Usually, we suppose that λ(M) and g(M) are given at some ˜ scale M = M, for example, at the grand unification scale, and we can then use renormalization group arguments to look for a renormalization scale suitable for our calculation (for example, M0 ). In our renormalization procedure the scalar quartic coupling is defined in terms of the fourth derivative of V . Thus, to study λ(M) one should derive the RGE for this fourth derivative (see Coleman & Weinberg (1973) and Gildener (1976)). We know, however, that the function β(λ) in the one-loop approximation does not depend on the renormalization conventions. Therefore to the lowest order we can use the results of calculations which adopt momentum space renormalization conventions. The one-loop RGEs for λ and g then read 16π 2 M(dλ/dM) = Aλ2 (M) + B λ(M)g 2 (M) + Cg 4 (M) 16π 2 M(dg 2 /dM) = −bg 4 (M) (11.30)

where the dimensionless coefficients A, B and C are given, respectively, by the diagrams shown in Fig. 11.2. To solve (11.30) we introduce α = λ/g 2 . Then we get (t = ln M) 16π 2 dα dα = −bg 2 (t) 2 = Aα 2 (t) + Bα(t) + C 2 (t) dt g dg = A(α − α1 )(α − α2 ), α1 < α2


where B = B + b. For real roots there is a UV stable fixed point at α = α1 < α2 ˜ ˜ ˜ and an IR stable fixed point at α = α2 . Thus, if 0 < α( M) = λ( M)/g 2 ( M) < α1 we expect the RGEs (11.31) to have a solution α(M) = 0 for M = 0. A direct integration gives exp −b
0 ˜ α( M)

dα Aα 2 + Bα + C


˜ 1 2 ˜ g 2 ( M) M =1+ g ( M)b ln (11.32) ˜ g 2 (M) 16π 2 M


11 Higgs mechanism in gauge theories

Fig. 11.2.

or, more explicitly, α(M) = 0 for 16π 2 ˜ ˜ F(α( M)) M = M exp − ˜ g 2 ( M) where ˜ F(α( M)) = 1 1 − exp −b b
˜ α( M) 0


Aα 2

dα + Bα + C


˜ ˜ ˜ ˜ If α( M) is of the order O(1) (i.e. if λ( M) ∼ g 2 ( M)) then F(α( M)) is also of order unity and M O(1) = exp − ˜ ˜ g 2 ( M) M (11.35)

˜ Thus, using (11.29) we conclude that for a reasonably small value of g 2 ( M), the dimensional transmutation can generate a scale M of spontaneous symmetry ˜ breaking which is enormously smaller than the scale M at which as we suppose ˜ is given. This small ratio of mass scales is not put into the theory by hand α( M) but arises automatically from the assumption of a vanishing scalar mass. Thus in order for this to happen some massless scalars which escape getting masses ˜ ˜ of order M must be available in our theory. For instance, if M is the grand unification scale they must remain massless after the spontaneous breakdown of the grand unification group. Such a situation can perhaps arise naturally from ˜ supersymmetries which are unbroken at M. The mechanism described here very much resembles the generation of the scale at which QCD coupling becomes large (see (7.53a)). It can also be generalized to the case of several quartic scalar couplings present in the theory (Gildener & Weinberg 1976). The generation of a very small ratio of mass scales by radiative corrections and ˜ the dimensional transmutation phenomenon is to be contrasted with making M/ M very small in the tree approximation. The latter is highly ‘unnatural’. To be more specific, we have in mind the following: consider a gauge group G which we

11.3 Dynamical breaking of gauge symmetries


˜ want to be spontaneously broken to G 1 at some scale M, and G 1 to be broken to ˜ G 2 at M with M/ M very small. It turns out that one can obtain such a pattern of spontaneous symmetry breaking in the tree approximation by a proper choice of the scalar potential, but only at the expense of tuning the parameters of the potential to an accuracy much higher than the neglected one-loop corrections. This is called the gauge hierarchy problem. The formalism presented in this section can be extended to non-abelian gauge theories with scalars and also fermions (see Coleman & Weinberg (1973) and Problem 11.3).

11.3 Dynamical breaking of gauge symmetries and vacuum alignment Dynamical breaking of gauge symmetry We consider now the possibility of spontaneous gauge symmetry breaking in the absence of elementary scalar fields. The mechanism presented here is often termed dynamical gauge symmetry breaking. Our discussion in this section is closely connected with that in Section 10.3. Imagine a spontaneously broken chiral global symmetry G of some strong gauge interactions with gauge group G H (we shall call it the hypercolour group) also to be broken explicitly by weak interactions with a gauge symmetry group G W ; some of the chiral currents of G couple to gauge bosons of G W . By assumption, it is a good approximation to treat G W interaction to the lowest order (see (10.64)). In Section 10.3 we have learned how to find the preferred vacuum which minimizes the vacuum energy in the presence of the perturbation G W . Assuming the positivity condition for certain current matrix elements the correct vacuum corresponds to the minimal value of the quantity (see (10.75))

ˆα ˆα Tr[ XJU XJU ]


built of the transformed currents projected onto the subspace of the broken ˆ generators in the original vacuum; we recall that J α is the linear combination of the α ˆ ˆ G generators corresponding to the G W current gα jµ (x) and XJ α is that part of J α α which contains only broken generators X . Equivalently, the correct vacuum must satisfy Dashen’s conditions (10.59) and (10.60). Assume that the correct vacuum has been found: it gives a specific orientation of the spontaneously unbroken subgroup H in the G, a set of broken generators X α and the corresponding Goldstone bosons, some of which get masses induced by the perturbation G W . In general one might also single out another subgroup of G: the maximal subgroup S of elements of G which commute with all the generators of G W . Then the coupling of gauge bosons to the currents of G breaks G explicitly to G W × S and we get


11 Higgs mechanism in gauge theories

Fig. 11.3.

Fig. 11.4.

the picture of the G symmetry breaking shown in Fig. 11.3 where the orientation of different subgroups is illustrated in Fig. 11.4. In the zeroth order in the G W couplings there are Goldstone bosons associated with each spontaneously broken generator (regions I + II + III). In presence of the G W interactions the Goldstone bosons split into three different groups which are easy to classify. The G W × S symmetry remains the exact symmetry of the lagrangian and therefore the Goldstone bosons in regions I and II remain massless in any order in the G W interactions, as can be proved following the arguments of Section 9.6. The generators in region III do not correspond to exact symmetries of the full lagrangian and the corresponding Goldstone bosons acquire, in general, masses which in the first order in G W are given by (10.55) or (10.77) and (10.78). From the calculation of the π + –π 0 mass difference we can see that the masses are of the order g (m ρ ∼ ), where g is the coupling constant of the weak interactions and is the confinement scale of the hypercolour interactions. The Goldstone bosons in regions I and II still split into two groups: clearly, those in region II remain in the physical spectrum; however, those in region I are absorbed by the G W gauge bosons which get masses through the Higgs mechanism. One then talks about the dynamical breaking of the G W gauge symmetry. Let us discuss the mass generation for the G W gauge bosons somewhat more explicitly. The α Goldstone bosons in region I couple to the G W currents jµ
α 0| jµ (0)|π a = i pµ f αa


11.3 Dynamical breaking of gauge symmetries

αβ µν

and therefore the gauge-invariant vacuum polarization tensor i
αβ 2 µν ( p )

defined by

= gα gβ

α β d4 x exp(i px) 0|T jµ (x) jν (0)|0 αβ

= i(gµν p 2 − pµ pν ) has a pole at p 2 = 0 (see also Section 11.1)
p 2 →0 αβ 2 µν ( p )

( p2 )



= −(gµν p 2 − pµ pν )

g α g β f αa f βa / p 2


It is important to remember the role of gauge invariance in writing (11.39): the Goldstone bosons contribute only to the pµ pν term of the polarization tensor; the gµν term must be present by gauge invariance and we do not need to specify its origin more explicitly. The full gauge boson propagator αβ reads† µν
αβ µν

= −iDµν [δ αβ −


( p 2 )] = −iDµν (1 +

)−1αβ + higher order terms (11.40)

We are working in the Landau gauge and −iDµν δ αβ is the free propagator in this gauge. Thus, the residue (11.39) gives the vector boson mass matrix m2 = αβ

gα gβ f αa f βa

The Goldstone boson decay constants f αa can be expressed in terms of the appropriate traces. Under assumption (10.70) we have
a 0| jµ (0)|π b = i pµ f π δ ab = i pµ f π Tr[Xa Xb ] a where jµ is a G current and consequently 2 m 2 = fπ αβ a 2 ˆ ˆ ˆ ˆ Tr[ J α X a ] Tr[ J β X a ] = f π Tr[ X J α X J β ]



(where no sum has been taken over α and β). This result has been obtained for the vacuum state |0 . For any other vacuum state related by the unitary transformation U |0 we find 2 ˆα ˆβ m 2 = f π Tr[ X JU X JU ] (11.43) αβ Comparing with (10.75) we see that E( ) ∼ Tr m 2 (11.44)

The conclusion is that the energetically preferred vacuum alignment and the relative orientation of the groups G W and H in the presence of the perturbation

αβ αβ αβ (−i λρ,β γ )(−iD δ γ β )). ρν µν = (−iDµν δ ) + (−iDµλ δ


11 Higgs mechanism in gauge theories

G W are such that the trace of the vector boson mass matrix is the minimal one (the gauge symmetry is broken as little as possible). The weak interactions determine their own pattern of symmetry breaking by the choice of the vacuum orientation. We note also that the vector boson mass matrix obeys, in general, certain relations which are consequences of H invariance. The above ideas may be relevant for the gauge symmetry breaking in the Glashow–Salam–Weinberg theory of the weak interactions (see, for instance, Fahri & Susskind (1981)), however, no convincing model has been constructed yet. We will illustrate them with two simple examples. Examples Consider a strong hypercolour gauge group G H with 2N left-handed fermions and 2N right-handed fermions transforming as the same representation of G H . The theory has G = SUL (2N ) × SUR (2N ) chiral symmetry which, we assume, breaks down spontaneously to H = SU (2N ). The embedding of H in G is characterized by a fermion condensate 0| ¯ Li R j |0 = i j (11.45) As in Section 9.6, assuming i j = vδi j , the chiral group is broken to SUL+R (2N ). Any equivalent orientation of the unbroken SU (2N ) corresponds to a condensate obtained from (11.45) by a unitary transformation
† † UR UL = vURUL


† (remember that U † U = UR UL ; U is an unitary operator, UL(R) are unitary † matrices), i.e. it is specified by a unimodular unitary matrix URUL . We take the weak gauge group to be G W = SUL (2) × U (1) with fermions in N left-handed G W doublets and 2N right-handed singlets, and with the U (1) charge assignments as follows (see Chapter 12):

Y =

Ui Di q


1 2



1 2

Consider first the case N = 1. Denoting the gauge bosons of the Weinberg– α Salam model by Wµ , Bµ the weak coupling is L=−
α α gWµ ¯ L γ µ T α L

− g Bµ ( ¯ R γ µ T 3


+ q ¯ γµ )


where T α is an SU (2) generator normalized to Tr[T α T β ] = 1 δ αβ and q is the mean 2 electric charge of the doublet. Assuming the condensate (11.45) and expressing the

11.3 Dynamical breaking of gauge symmetries


α G W currents jµ in terms of the G currents corresponding to the unbroken subgroup (vector currents) and to the broken generators (axial currents) jL = 1 (V −A), jR = 2 1 (V + A) one gets in the ( L , R ) basis the following expression: 2

Tα ˆ Jα = g +
1 g 2

0 Tα

+g −T

q q+T + 1g 2

= −T 3

1 g 2

Tα T

+ 1g 2

2q + T 3 2q + T 3 (11.48)


Inserting (11.48) into (11.42) gives the familiar vector boson matrix W1 W2 W3 B After diagonalization one gets m W1,2 = 1 g f π , 2 mγ = 0 where

W1 1 2 2 g fπ 4

1 2 2 g fπ 4



1 2 2 g fπ 4 1 2 − 4 gg f π

2 − 1 gg f π 4 1 2 2 g fπ 4

mZ =

1 f (g 2 2 π

+ g 2 )1/2 = m W / cos



the Weinberg angle, is given by cos


g (g + g 2 )1/2

and γ = cos Z = − sin

+ sin + cos

W W3 W W3


In this model all the Goldstone bosons have been used to give masses to the weak gauge bosons. There is no problem of vacuum alignment since all the equivalent vacua (11.46) can be transformed into (11.45) by the gauge transformations in the weak sector. For N > 1 left-handed doublets with the same U (1) charge assignment the situation is more complex. If we introduce the 4N -plet = (U1L D1L . . . D N L D N L U1R D1R . . . U N R D N R ) then the Salam–Weinberg coupling is l Tα ⊗ 1 α L = −gWµ ¯ γ µ 0 − g Bµ ¯ γ µ q q + T3 ⊗ 1 l (11.50)

where the 2N × 2N block T α ⊗ 1 is a direct product of the 2 × 2 weak isospin l and the N × N matrix on the space of N doublets. The weak coupling with


11 Higgs mechanism in gauge theories

the gauge symmetry G W = SUL (2) × U (1) breaks the original chiral symmetry G = SUL (2N ) × SUR (2N ) of the strong interaction sector into G → G W × S where S = SUL (N ) × SUR (N ) × SUR (N ) with the first SU (N ) corresponding to horizontal transformations among N left-handed doublets, and the second (the third) SU (N ) corresponding to transformations among N right-handed singlets UiR (DiR ); UiR and DiR have different charges and therefore the original SUR (2N ) is broken into SUR (N ) × SUR (N ). The spontaneous breakdown of G into SUL+R (2N ) induced by (11.46) leaves the subgroup SUL+R (N ) of S unbroken. The spontaneous breakdown of the original SU (2N ) × SU (2N ) chiral symmetry produces 4N 2 − 1 Goldstone bosons. Three of them are used to give masses to the gauge bosons. Again, there is no vacuum alignment problem. The breaking of S into SU (N ) which remains the exact symmetry of the full theory leaves 2(N 2 − 1) exactly massless Goldstone bosons. They are electrically neutral because they correspond to generators of transformations among fermions with the same charge. There remain 2(N 2 − 1) charged Goldstone bosons which do not correspond to generators of exact symmetries of the full lagrangian and they should, therefore, receive masses from the G W interactions. Splitting the weak currents into the ˆ ˆ unbroken YJ α and the broken parts XJ α as in (11.48)  Tα ⊗ 1 2q + T 3 ⊗ 1 l l  1 1 Y ˆα   J ≡ 2g + 2g α 3 T ⊗1 l 2q + T ⊗ 1  l  (11.51)   −T α ⊗ 1 −T 3 ⊗ 1 l l  1 1 X ˆα  + 2g J ≡ −2g  Tα ⊗ 1 l T3 ⊗ 1 l it is evident from (10.79) that the Goldstone bosons receive no mass in the first order in G W . The broken generators X a are of the block-diagonal form −A A where the As are 4N 2 − 1 matrices which can be easily constructed for any N , and ˆ ˆ therefore for YJ and XJ given by (11.51) the two contributions to the trace in (10.79) cancel each other. Actually, one can see that if G W commutes with one of the chiral groups SUL (2N ) or SUR (2N ) then there will be pseudo-Goldstone bosons which remain massless to the lowest order in G W . Indeed, one sees directly from (10.66), using H invariance as in (10.69), that the -dependent (U -dependent) part of the vacuum energy E( ) relevant for the pseudo-Goldstone boson masses (10.79) comes from the term E( ) = − 1 2 d4 x


† α † α 2 gα 0|T UL jµL (x)ULUR jνR (0)UR |0


α α α α whereas jµL jνL and jµR jνR terms contribute only the

-independent constants.

D2R (− R L (11. 11. U1R L D1R ( − 1 ). In our example the U (1) current also does not contribute to E( ) (although it has both the left-handed and the right-handed parts) because the left-handed U (1) current is an SUL (2N ) singlet. This can be avoided by giving the left-handed doublets different U (1) charge. 11.5. To be specific we consider a model with four left-handed and four right-handed fermions with the G W representation content given by U1 D1 Y = U2 D2 − .3 Dynamical breaking of gauge symmetries 385 Fig.5. In the model discussed above there remain massless neutral Goldstone bosons in the physical spectrum because the SU (N ) group of transformations between doublets remains the exact symmetry of the theory.11. Such a model is very instructive because then the U (1) boson exchange contributes the -dependent term to the E( ) and the pattern of the G W breakdown depends on the U (1) charge assignments. for example. so it is SUL (2N )× SUR (2N )-invariant and hence -independent. ( + 1 ). The U (1) contribution is therefore SUL (2N )invariant and also SUL+R (2N )-invariant because the vacuum is.55) .54) where in the product A ⊗ B the first matrix is in the 2 × 2 weak space and the second one is on the space of doublets.50) the G W coupling in the (U1L D1L U2L D2L U1R D1R U2R D2R )) is Tα ⊗ 1 l α L = −gWµ ¯ γ µ + 1 ⊗ T3 l 1 ⊗ T3 l 0 − g Bµ ¯ γ µ basis ( = 0 T3 ⊗ 1 l (11. 2 L. 2 U2R (− + 1 ).53) 1 ) 2 . Let us first assume that the SU (4) × SU (4) symmetry of the G H lagrangian is broken by the ¯ ¯ 0|ULi UR j |0 = 0|DLi DR j |0 = vδi j (11. as in Fig. 2 − Similarly to (11. The charged pseudo-Goldstone bosons receive masses in the next order in G W due to the mass splitting of the γ and Z which has to be taken into account in that order.

Therefore.57) gives the following masses for the remaining pseudo-Goldstone bosons which are denoted by the flavours linked by the corresponding generators: non-diagonal terms of the mass matrix vanish and bosons corresponding to the two generators like (11. Assuming that the integral M 2 is dominated by the lowest-lying vector mesons of the correct quantum numbers we find. D1 ) with one of (U2 . as for the π + –π 0 mass difference in Section 10.56) tells l. as before the contribution of the SU (2) current to the two terms under the trace in (10.53).58)    2 m D1 U2 ∼ g 2 ( − 1)M 2 where M 2 is the integral given by (10. we find an interesting result: for | | > 1 Dashen’s condition is satisfied by masses (11.55) turns out to be unstable. However.79) cancels. A look at (10.     1 0 −i 0 0 0 1  1  0 0  0 0    or (11. M 2 > 0. for l.79) and (11.56) and (11. us that Goldstone bosons corresponding to broken generators T a ⊗ 1 T a ⊗ T 3 and 1 ⊗ T 3 receive no mass in the first order in G W .386 11 Higgs mechanism in gauge theories condensates.4. D2 ).56) There are 15 Goldstone bosons corresponding to the broken generators X= 1 −A A 2 where A represents T a ⊗ 1 2T a ⊗ T 3 .57) √  √   1 0  i 0   2 2 0 0 0 0 0 0 linking one of (U1 . therefore.55). We must.55) is the preferred vacuum in the presence of perturbation G W . for the U (1) charge assignments for doublets such that | | < 1 the vacuum (11. (11. l example. We can now split the G W currents into unbroken and broken parts in the vacuum (11. An explicit calculation l based on (10.57) are degenerate in mass and  m 2 1 U2 = m 2 1 D2 ∼ g 2 2 M 2  U D   2 2 2 m U1 D2 ∼ g ( + 1)M (11. find the correct .58) and the vacuum (11.79). For the U (1) current we have ˆ J = 1g 2 T3 ⊗ 1 + 2 1 ⊗ T3 l l T3 ⊗ 1 + 2 1 ⊗ T3 l l + 1g 2 −T 3 ⊗ 1 l T3 ⊗ 1 l (11.78). It is clear from our previous discussion that only the U (1) current is interesting to us. 1 ⊗ T 3 and eight matrices of the form. This defines the vacuum and the breaking SUL (4) × SUR (4) → SUL+R (4) in the basis (11.

which breaks the SUL+R (4)-invariance of the term. respectively. However.s. simultaneously transforms the other term of the U (1) current to such a form that the full transformed current has ‘better’ SUL+R (4) symmetry than the original one.11. is the U (1) current in the original basis (11. split into unbroken and broken parts. The right transformation can be found by noting that the transformed U (1) current. Before we find this transformation let us notice that in the original basis (11. for | | < 1 one can imagine ˆ that a transformation U † J U .36) the right transformation is the one which ˆ2 ˆ2 minimizes the Tr XJU or maximizes the Tr YJU of the projection of the transformed current on the subspace of the original broken (X a ) and unbroken (Y a ) generators. It is then clear that for | | > 1 that basis is the right one: the L given by (11. Hence. According to (11. Since the trace of B 2 should be the minimal one.55) by all unitary transformations U of SU (4) × SU (4). is as follows:        U†       −        U      −1 − − 1+ −1 + 1− = A A + B −B where the matrix on the l.54) is ‘as far as possible’ SUL+R (4)-invariant. the transformed current should have both 4 × 4 blocks as similar as possible (A + B ≈ A − B ≈ A).54).54) the term in the lagrangian proportional to is SUL+R (4)-invariant. for | | < 1 the right transformation is the one which gives        ˆ U† JU =       −              −1 − − 1+ 1− −1 + .h.3 Dynamical breaking of gauge symmetries 387 vacuum in the set of vacua obtained from (11.

The new vacuum is then defined by the condensate ¯ ¯ ¯ ¯ D1L U2L D2L U   1L 1 0 0 0 † UL = v  0 0 1 0     0 1 0 0  0 0 0 1 U1R D1R U2R D2R UR ¯ so the electrically charged composite field U2 D1 acquires a non-zero vacuum expectation value. Thus. Problems 11. Of course.1 Derive the remaining Feynman rules for the U (1) theory of Section 11. one can easily find a set of broken and unbroken generators corresponding to the new vacuum as well as all the gauge boson masses and pseudo-Goldstone boson masses. respectively.54) one sees that this is the transformation D1R ↔ U2R .2 Check by explicit calculation that at the tree level the unphysical ξ -dependent poles cancel in the S-matrix for the process ϕ1 + ϕ2 → A1 + A2 where ϕ and A are the physical scalar particle and the massive vector meson of the U (1) theory in Section 11. 11.2 to non-abelian gauge theories including fermions. An interesting conclusion of the above exercise is that the coupling of a weak gauge group to the chiral currents of a spontaneously broken chiral symmetry of strong interactions may. for dynamical reasons. Calculate the Higgs particle (physical scalar) mass in terms of the W and Z masses in the SU (2) × U (1) electroweak theory with one massless scalar complex doublet and taking fermions as massless.388 11 Higgs mechanism in gauge theories From (11. lead to different theories depending on the specific assignment of the chiral multiplets to the weak group representations.1.1 in the Rξ gauge. the U (1)em subgroup of the gauge group SU (2) × U (1) is broken and the photon has a mass when | | < 1. 11.3 Extend the discussion of Section 11. .

2. 389 . 1.12 Standard electroweak theory The basic concepts of the electroweak theory (Glashow 1961. 2. 1.1)  1  c  3 . We recall that the hypercharge is defined as Y = Q − T 3 . 2. The complete SUc (3) × SUL (2) × UY (1) theory is called the Standard Model of elementary interactions. +1)   A  1 3. It is just used to denote the left-handed Weyl fields which are singlets of SUL (2). Salam 1968) have been outlined in the Introduction. 3 is the generation index. supplemented by the idea of minimal unification with electromagnetic forces. (Any reader unfamiliar with the two-component spinor notation should consult Chapter 1 and Appendix A. and T 3 is the third component of its weak isospin SUL (2). At this stage the superscript ‘c’ does not have the same meaning as in Chapter 1. + 3  dA    c uA 3 . where Q is the electric charge of the field. respectively. In terms of the left-handed Weyl fields each generation consists of:  lA 1. The last entry shows the hypercharge of each field. 1.) The first and second entries show the transformation properties under the colour SU (3) and the weak SUL (2) groups. The matter fields consist of quarks and leptons forming three generations. The symmetry is assumed to be local gauge symmetry. The underlying symmetry group SUL (2) × UY (1) emerges from study of the charged weak currents. in accordance with general remarks in the Introduction and as required by unification with electromagnetism. It will acquire the proper meaning of charge conjugation (as in Chapter 1) after the electroweak symmetry breaking to the electromagnetic U (1). we have the following gauge vector fields:† Wµ for SUL (2) and Bµ for UY (1). Weinberg 1967b. + 6 qA (12. − 1  2    ec (1. a Thus. The hermitean conjugates of the fields carrying it will then combine with appropriate † The SU (3) gauge group of QCD was extensively discussed in Chapter 8. − 2 3 where A = 1.

l2 ≡ e c e1 ≡ ec . l3 ≡ µ τ c c c e2 ≡ µ . ( l A )L ≡   ec ¯A 0 (12. We shall also use the notation ( l1A )L ≡ ( ν A )L . We can also introduce the four-component chiral fermion fields. u2 ≡ c . Writing explicitly the SUL (2) doublet components and using the conventional terminology we have l1 ≡ νe . 2.  dA uc ¯A 0 where i = 1. c d3 ≡ b c for quarks. The left-(right-)handed chiral fields L(R) ≡ 1 (1 ∓ 2 γ5 ) L(R) are doublets (singlets) under SUL (2) and the electric and colour charges of the left.390 12 Standard electroweak theory components of the doublets to form Dirac fields. One Higgs doublet is the minimal set of scalar fields which allows for the required symmetry breaking.1) must be invoked. To this end one introduces a doublet of scalar fields H . νµ ντ . the one-doublet theory is what is called the Standard Model. with the quantum numbers (1. q2 ≡ . with charge conjugation acting only on their electric and colour (i. q3 ≡ d s b c c c c c u1 ≡ u . as in (1. Since the presence of more doublets (for constraints on the presence of other representations of scalars see Problem 12. We recall that the unbarred (barred) two-component spinors carry lower (upper) Lorentz indices.1) has no particular theoretical motivation (apart from the supersymmetric models which require an even number of Higgs doublets). 1 ) 2 with respect to the full SU (3) × SUL (2) × UY (1) gauge group. . ( e A )R ≡ . u3 ≡ t c c d1 ≡ d c .e. It does not destroy the renormalizability of the theory (’t Hooft 1971a. conserved) charges. which are strictly massless in the symmetry limit. ( d A )R ≡ ¯ c .and right-handed fields are pairwise the same. The Higgs mechanism is the origin of the masses of the W ± and Z 0 vector a bosons (which are linear combinations of the original Wµ and Bµ fields) and of the masses of elementary fermions. the so-called Higgs mechanism (discussed in Section 11. e3 ≡ τ c for leptons and q1 ≡ u c t . more precisely. c d2 ≡ s c . The gauge symmetry SUL (2) × UY (1) must be spontaneously broken to the electromagnetic UEM (1) or. called the Higgs doublet.2) q iA 0 0   i  ( q A )L ≡ ( u A )R ≡ .72):  l iA 0  i  . 2. b).

and are eliminated by normal ordering (a redefinition of the vacuum energy) which is understood also to be present in (12. Therefore.1 The lagrangian 391 Moreover. By proper redefinition of the group generator basis the remaining U (1) can always be interpreted as the electromagnetic UEM (1). . the quantum one. 12. the Higgs boson mass range can extend up to O(1 TeV). As discussed in Chapters 1 and 9.and partly two-loop corrections to the tree level predictions have been tested (see later). as the correct theory of elementary interactions for the energy range up to O(100 GeV). after symmetry breaking. mainly by the LEP experiments.4). the pattern SU (2) × U (1) → U (1) is unique if the Higgs mechanism is generated by vacuum expectation values of the components of a complex scalar SU (2) doublet field H (x). Weinberg 1967b.3). The electroweak theory has been confirmed experimentally with the precision of one per mille.3) and the components H + and H 0 are complex fields. Thus. the Higgs scalar has not yet (1999) been discovered. However. As discussed in Chapter 9.1 The lagrangian With the ideas and the formalism introduced so far in this book we are well equipped to formulate the electroweak theory (Glashow 1961. and is extensively searched for in present and planned experiments. In particular. full experimental clarification of the Higgs sector is expected to provide an important bridge to physics beyond the Standard Model. This is different from quantum vacuum fluctuations which are singlets with respect to all symmetries.and right-handed chiral matter fields. the vacuum carries quantum numbers of the component H 0 . It may be useful to stress that.1) the desired pattern is obtained with H (x) = where 1 0|H (x)|0 = √ 2 0 v (12. It can be elementary or composite or even only the effective description of a strongly interacting (at the scale O(1 TeV)) gauge boson sector (Section 11.4) H + (x) H 0 (x) (12. interestingly enough. With the basis fixed by the charge assignment (12. spontaneously broken symmetries can be intuitively depicted as the property of the vacuum which is a big reservoir of the scalars H 0 .4). From the point of view of the electroweak theory considered in the range up to O(100 GeV). they can generate the Dirac mass terms for those particles. according to (12. Higgs doublets and only Higgs doublets can have renormalizable Yukawa couplings to the left.12.

’t Hooft 1971a. b) as a gauge theory. The generation index A is summed from 1 to 3.7) can be rewritten in the more familiar form i a Lmatter = i( l A )L γ µ ∂µ + igWµ T a − g Bµ ( l A )L 2 + i( + i( + i( + i( e A )R γ q A )L γ d A )R γ u A )R γ µ µ ∂µ − ig Bµ ( e A )R q A )L µ µ i a ∂µ + igWµ T a + g Bµ ( 6 i ∂µ − g Bµ ( d A )R 3 2 ∂µ + i g Bµ ( u A )R 3 (12. a a Lkin = 1 Wµ g µν ∂ 2 − ∂ µ ∂ ν Wν + 1 Bµ g µν ∂ 2 − ∂ µ ∂ ν Bν gauge 2 2 b c a b b b Lint = gεabc ∂ µ W aν Wµ Wν − 1 g 2 W aµ W bν Wµ Wν − W aµ W aν Wµ Wν gauge 4 The matter field lagrangian in the two-component spinor notation reads i a Lmatter = il¯A σ µ ∂µ + igWµ T a − g Bµ l A ¯ 2 c µ c + ie A σ ∂µ + ig Bµ e A ¯ ¯ i a + iq A σ µ ∂µ + igWµ T a + g Bµ q A ¯ ¯ 6 i c ¯c ¯ + id A σ µ ∂µ + g Bµ d A 3 2 + iu c σ µ ∂µ − i g Bµ u c ¯A ¯ A 3 (12. respectively. In terms of four-component chiral fermions defined in (12.2) the lagrangian (12.8) . Its kinetic and interaction parts are. g are the two coupling constants associated with SUL (2) and UY (1) groups.7) where T a are the SUL (2) generators in the fundamental representation and g.392 12 Standard electroweak theory Salam 1968.fix + Lghosts The gauge boson part has the standard form: a Lgauge = − 1 Wµν W aµν − 1 Bµν B µν 4 4 (12. The full classical gauge-invariant lagrangian consists of several parts: LSM = Lgauge + Lmatter + LHiggs + LYukawa + Lg.6) a a a b c where Wµν = ∂µ Wν − ∂ν Wµ − gεabc Wµ Wν and Bµν = ∂µ Bν − ∂ν Bµ are the field a strengths of the Wµ and Bµ gauge fields and εabc are the SUL (2) group structure constants.5) (12. respectively.

12.e. the minimum of the potential is for H†H = − v2 m2 ≡ λ 2 (12.c. if they exist.c. for a = 1. (iτ2 H ) j q j = εi j Hi q j is also an invariant of SUL (2). i. The neutrino Yukawa terms and explicit Majorana mass terms for the righthanded neutrinos can then be added to the electroweak lagrangian without destroying the renormalizability of the theory.13) √ √ 2 2 where Y is the hypercharge generator. as 2 of SUL (2). d B )R ( i q A )L − YuB A εi j Hi ( u B )R ( j q A )L (12. The SUL (2) × UY (1) symmetry is then broken down to U (1) (which is identified with the electromagnetic U (1) with Q = T 3 + Y ) because     0 0 T a  v  = 0.9) the Higgs doublet acquires the vacuum expectation value. Note.† Using the four-component chiral fields this can be rewritten as LYukawa = −YlB A Hi ( e B )R ( i l A )L − YdB A Hi ( + h. j and over the fermion generation indices A and B is understood.8) compared to (12.10) we have not included the Yukawa couplings for neutrinos since our list (12. We have also used the fact that the two-dimensional representation of SU (2) is real (see Appendix E) and iτ2 H transforms as H . (12. Such an extension may be necessary in view of the accumulating evidence for neutrino masses.12) By SUL (2) rotation we can always redefine the vacuum so that only the vacuum expectation value of the lower component of the Higgs doublet is non-vanishing. . † In the lagrangian (12. however.5): c LYukawa = −YlB A Hi li A ec − YdB A Hi qi A d B − YuB A εi j Hi q j A u c + h.2) of physical fields does not contain the right-handed neutrinos. (T 3 + Y )  v  = 0 (12.11) When m 2 < 0 is chosen in (12. Y AB are 3 × 3 complex Yukawa coupling matrices and summation over the SUL (2) indices i.1 The lagrangian 393 (Note the change of signs of the U (1) charges of the SUL (2) singlets in (12.7). that the right-handed neutrinos. Indeed. 3. 2.9) Interactions of the Higgs doublet with the matter fields are described by the Yukawa part of the full lagrangian (12.) The lagrangian for the Higgs field has the form i ← a LHiggs = H † ∂ µ − igT a Wµ − g Bµ 2 λ 2 H†H − m2 H † H − 2 i → a ∂ µ + igT a Wµ + g Bµ H 2 (12. Therefore.10) B B where ε12 = −ε21 = −1. are sterile (T3 = Y = 0) with respect to the electroweak interactions and their inclusion would be a no-cost extension of the theory.

i.8)) J aµ = f µ JY = f f ¯ L γ µY f f L f ¯ L γ µT a f L f R (12.19) The orthogonal combination defines another electrically neutral field of the theory: 0 Zµ = 1 3 3 (gWµ − g Bµ ) ≡ cos θW Wµ − sin θW Bµ (g 2 + g 2 )1/2 (12.16) + f f ¯ R γ µY f The SUL (2) current has. explicit V –A structure. This combination can be easily found either from the requirement that the photon does not couple to the neutral component of the Higgs field. or from the requirement that its coupling to fermions is vector-like with the strength given by their electric charges. We find Aµ = (g 2 1 3 3 (g Bµ + g Wµ ) ≡ cos θW Bµ + sin θW Wµ + g 2 )1/2 (12.14) where (see (12. from the tree level structure of the lagrangian.17) where the Weinberg angle is defined as sin2 θW ≡ g2 . g2 + g 2 tan θW ≡ g g (12. Both left. we shall discuss the predictions of the theory which follow from the underlying gauge symmetry and the Higgs mechanism. too.20) .and right-handed fermions contribute to it.e. It follows from the unification hypothesis that the photon field Aµ is a com3 bination of the Yang–Mills fields Wµ and Bµ . of course.2 Electroweak currents and physical gauge boson fields The gauge boson couplings to matter follow from Lmatter and have the form µ a Lmatter ⊃ −g J aµ Wµ − g JY Bµ (12. This is a potential source of chiral anomalies (see Chapter 13). 12.5) will be discussed later. First.394 12 Standard electroweak theory The gauge-fixing term and the Fadeev–Popov ghost term of the lagrangian (12. but with different U (1) charges. Note that the U (1) current is chiral.18) and the electric charge is (e > 0) e≡ (g 2 gg = g sin θW + g 2 )1/2 (12.15) (12.

three of the four degrees of freedom present in the Higgs doublet H (x) are 1 ± 1 exchanged for the longitudinal components of the gauge bosons Wµ ≡ √2 (Wµ ∓ 2 iWµ ) and Z 0 .12. the reader is invited to check this by explicit calculation). and it depends crucially on the SUL (2)×UY (1) representation content of the Higgs fields. Moreover. in accord with the fact that the U (1)EM gauge symmetry remains exact.25) (actually the same result holds with any number of Higgs doublets.9) we see that the fourth degree of freedom originally present in the field H (x) remains in the physical 2 spectrum as an electrically neutral massive Higgs scalar with the mass Mh = λv 2 . which obtain the masses 2 MW = 1 g 2 v 2 . 4 2 M Z = 1 (g 2 + g 2 )v 2 4 (12. however.9) L that the minimum of the scalar potential occurs for v 2 = −2m 2 /λ. the value of the Weinberg angle has a priori nothing to do with the vector boson masses. and together with the electric charge can be used as independent input parameter fixing the couplings g and g . . in atomic parity violation experiments (for other measurements of sin2 θW – see Section 12. The physical gauge and Higgs boson spectrum of the electroweak theory is most easily seen in the so-called unitary gauge introduced in Chapter 11. obtained after the W ± and Z 0 bosons are given masses by the Higgs mechanism. but is just a measure of this departure.22) µ The mixing angle θW measures the departure of the weak neutral current J Z from pure V –A structure! It is measured experimentally.21) + f f ¯ Rγ µQ f f R (12. Using the ‘angular’ variables for the Higgs field 1 ηa (x) a H (x) = √ exp i T v 2 0 v + h 0 (x) (12. The W ± and Z 0 boson masses are related by MW = cos θW MZ (12.14): g g µ 0 µ 0 JZ Z µ = (J 3µ − sin2 θW JEM )Z µ cos θW cos θW with µ JEM = f f ¯ L γ µQ f f L (12. Also from (12. It is worth stressing that the departure in the neutral current from pure V –A Lorentz structure is the basic prediction of the unified theory. Furthermore.24) The photon remains massless.23) and choosing the gauge rotations a (x) = −ηa (x)/v.2 Electroweak currents 395 Its couplings to fermions then follow from (12.4). Such a relation is. for example. we again see from (12.

Inserting (12.fix part of the lagrangian. These terms are straightforward to obtain and their derivation is left as an exercise for the reader. cancelled by the appropriate choice of the Lg. contribute to the S-matrix elements. G 0 and G + fields as well as their self-interactions 0 ± and interactions with the gauge fields Aµ . setting external momenta on the mass-shell.28) The fields G 0 and G ± are massless at this stage but they will be given masses by the gauge-fixing term. In lagrangian (12. Finally.26) where h 0 is assumed to have a vanishing vacuum expectation value. Before fixing the gauge we write† in general H= G+ 1 √ (v + h 0 + iG 0 ) 2 (12. Also note that the neutral field is split into two real fields. the unphysical poles of the Green’s functions do not.3) we have renamed the Goldstone boson fields which in the unitary gauge are ‘eaten up’ by the longitudinal components of the physical massive gauge boson fields.9) one finds (with v 2 = −2m 2 /λ fixed as usual in its minimum) that the fields G 0 and G ± remain in the lagrangian (12.27) which mix W ± and Z 0 vector bosons with the Goldstone † Compared to (12. We recall that S-matrix elements are obtained from the Green’s functions by removing external lines.27) In the so-called Rξ gauge these terms are. as discussed in Section 11. the unphysical Higgs fields (the three Goldstone bosons ‘eaten up’ in the unitary gauge by the longitudinal components of the gauge boson fields) remain in the lagrangian (and in the Feynman rules). and contracting tensor indices with appropriate physical polarization vectors.9) we also have the canonical kinetic terms for h 0 . . However.1. in the following we choose to work with the covariant gauge-fixing condition.26) in the scalar potential in (12.9) and mix with the gauge bosons (v can always be taken as a real parameter): i + − 0 LHiggs = · · · + gv(Wµ ∂ µ G − − Wµ ∂ µ G + ) − 1 (g 2 + g 2 )1/2 v Z µ ∂ µ G 0 2 2 (12. of course.396 12 Standard electroweak theory Since in the unitary gauge the theory is not manifestly renormalizable. Z µ and Wµ . We complete this section by discussing some aspects of the lagrangian in covariant (renormalizable) gauges. The terms (12. In this class of gauges. however. one finds the mass terms for the gauge bosons 1 2 3 LHiggs = · · · + 1 g 2 v 2 (W 1µ Wµ + W 2µ Wµ ) + 1 v 2 (gW 3µ − g B µ )(gWµ − g Bµ ) 8 8 2 − 2 0 ≡ MW W +µ Wµ + 1 M Z Z 0µ Z µ 2 (12.

more explicitly. one gets Lg.2 Electroweak currents 397 fields are cancelled by the following gauge-fixing lagrangian† Lg. Finally. The first three terms are gauge-dependent corrections to the kinetic terms for the gauge bosons.31) The last two terms indeed cancel the terms in (12. . the ghost lagrangian can be obtained using the general procedure described in Chapter 3. We follow here the simplest approach and introduce only one such parameter. δL H = ig H δ L T (12.fix = − with F 0 ≡ ∂ µ Bµ − iξ g H † Y H − H † Y H a F a ≡ ∂ µ Wµ − iξ g H † T a H − H † T a H 1 2ξ 3 FaFa a=0 (12.12. 1 F 0 ≡ ∂ µ Bµ − ξ g vG 0 2 1 3 µ 3 F ≡ ∂ Wµ + ξ gvG 0 2 i 1 F 1 ≡ ∂ µ Wµ − √ ξ gv(G − − G + ) 2 2 1 2 F 2 ≡ ∂ µ Wµ − √ ξ gv(G − + G + ) 2 2 Integrating these terms by parts. δ W a = −gεabc W b δ c + ∂ δ a  L µ L µ µ L µ L † In general. Performing infinitesimal gauge transformations on the fields in F a in (12.29) (a = 1. 3) (12.30) or. δY W µ = 0    δ B = 0. 2. one can introduce different gauge-fixing parameters ξ for different components of the W a field and for the B field. and the remaining two terms are mass terms for the unphysical Goldstone bosons. δL H = −igδ a T a H  L    a a  † † † † δY H = ig δ Y H Y.fix = 1 0 µ ν 0 1 1 + − Z µ∂ ∂ Z ν + A µ ∂ µ ∂ ν A ν + Wµ ∂ µ ∂ ν Wν 2ξ 2ξ ξ 1 2 2 − ξ M Z G 0 G 0 − ξ MW G + G − 2 1 i 0 + − + (g 2 + g 2 )1/2 v Z µ ∂ µ G 0 − gv Wµ ∂ µ G − − Wµ ∂ µ G + 2 2 (12.27).30):  δY H = −ig δ Y Y H.32) a  δY Bµ = ∂µ δ Y .

36) ± and expressing the functional derivatives (12. one arrives at Lghost generating the Feynman rules collected in Appendix C. With . y)ηY (y) + ηL (x)MLL (x.33) through the physical fields Wµ . y)ηL (y) + ηL (x)MLY (x. y) ≡ = ξ gg H † Y T a H + H † T a Y H δ (4) (x − y)    δ Y (y)     0  δF (x)  † a † a (4) a  MYL (x. y) ≡ = ∂ + ξ g H Y Y H + H Y Y H δ (x − y)    δ Y (y)    a   δF (x) ←µ  2 abc b 2 † c a ac  MLL (x. Aµ and h 0 . Clearly. y)ηY (y) ¯ (12.398 12 Standard electroweak theory we compute the corresponding Fadeev–Popov determinants (the variations are taken always at = 0):  δF 0 (x)  2 2 † † (4) MYY (x.33) (the symbol ∂ means that the derivative acts on the object standing to its left in the full expression for the ghost action). 0 Z µ .   1 1 2 η± = √2 ηL ∓ iηL . We shall work in the so-called ’t Hooft–Feynman gauge with ξ = 1. y) ≡ = ξ gg H T Y H + H Y T H δ (x − y)  a δ L (y) (12. The ghost action then reads Sghost = dx ab a dy ηY (x)MYY (x. all those fields have well-defined symmetry properties under SUL (2) × UY (1) and we call them to be in an ‘electroweak basis’. defining dx Lghost (x) (12.34) Since the variations of F a contain factors δ (4) (x − y). 12. this can be written as an integral of a local lagrangian density Sghost = Finally. y) ≡ = ∂ + gc ∂ Wµ + ξ g H T T H  c  δ L (y)  † a c (4) + H T T H δ (x − y)   δF a (x)  a  MLY (x. (12.35)  3 η Z = cos θW ηL − sin θW ηY . y)ηL (y) ¯ ¯a ←µ a a a ¯a + ηY (x)MYL (x.   3 η A = sin θW ηL + cos θW ηY .3 Fermion masses and mixing The Yukawa terms (12.10) in the electroweak lagrangian couple the left-handed Weyl fermion doublets of SUL (2) with the left-handed Weyl fermion singlets and the Higgs boson.

37) B B 2 2 2 The physical interpretation of these terms becomes more transparent when we rewrite the lagrangian (12. µ. to diagonalize the mass matrices we perform unitary transformations on the right. more fundamental theory will fix those parameters in some electroweak basis. 2 v [m u ] = √ [Yu ] 2 (12.c. ec . Fisher. this basis is defined up to rotations in three-dimensional c flavour space. Yukawa couplings are free parameters of the present theory. 2. They can be diagonalized by two unitary transformations (check it): V † [m]U → [m]diagonal with real diagonal entries. l A .and left-handed fields. ( ( u A )R d A )R = (UR ) AB ( = (DR ) AB ( u B )R d B )R (12. 2 v [m d ] = √ [Yd ].10) is that neutrinos remain massless. A non-zero vacuum expectation value of the Higgs field (12.26) generates fermion mass terms: v v v c Lmass = − √ YlB A e A ec − √ YdB A d A d B − √ YuB A u A u c + h. τ ). 3). u c . independently for each category of fields q A . e. d A (A = A A 1.3 Fermion masses and mixing 399 three families of fermions. for example. As mentioned in Section 12.1 one consequence of the lagrangian (12. d B )L .12. c strictly speaking.39) In general the Yukawa matrices can be the most general complex 3 × 3 matrices.37) in terms of the four-component chiral fields. The accumulating experimental evidence for neutrino masses and oscillations (for a review see. An attractive and straightforward extension is the inclusion of the right-handed neutrinos or.1) .e. One may hope that the future. The primed fields describe physical particles: after diagonalization we can combine chiral fields into Dirac fields = L + R . Kayser & McFarland (1999)) may call for some extension of the theory.c. of a set of new left-handed Weyl fields νL in the list (12.40) and analogously for the down quarks and charged leptons e A (i. which are mass eigenstates. (12. (12. independently for up and down quarks and charged leptons: ( ( u A )L d A )L = (UL ) AB ( = (DL ) AB ( u B )L . Thus.38) 2 The mass matrices are: v [m e ] = √ [Ye ]. defined in (12.2): v v Lmass = − √ YlB A ( e B )R ( e A )L − √ YdB A ( d B )R ( d A )L 2 2 v BA − √ Yu ( u B )R ( u A )L + h. the latter follows from the fact that the transformations V and U can include phase rotations independently for the leftand right-handed fields.

43).2). that due to the doublet structure of all left-handed quark multiplets. after the Higgs field aquires a nonzero vacuum expectation value.40) in terms of the left-handed Weyl fields) − Jµ = A u A σµ d A + ν A σµ e A ¯ ¯ ¯ ¯ u A σµ (VCKM ) AB d B + ¯ ¯ AB A = ≡ AB ν A σµ e A ¯ ¯ dB ¯ u γµ PL (VCKM ) AB A + A ¯ ν γµ PL A eA . in particular. The neutral weak current and the electromagnetic current have the same form in the mass eigenstate and in the electroweak basis. the neutral weak current remains diagonal in flavour in the mass eigenstate basis. they are determined by the same Yukawa coupling . This see-saw mechanism (Gell-Mann. the Yukawa couplings of the physical scalar h 0 and of the unphysical pseudoscalar Goldstone boson G 0 (present in the Rξ gauge) become diagonal in flavour in the mass eigenstate basis. In (12. Finally.400 12 Standard electroweak theory with four-component ( ν A )R given by the relation analagous to (12. Ramond & Slansky 1979) can naturally explain the neutrino masses in the experimentally suggested range provided the Majorana mass M is of the order of the Grand Unification scale. (12.14) to the gauge bosons in terms of the physical fermion fields: g g µ 0 µ − + Lmatter ⊃ − √ (Jµ W +µ + Jµ W −µ ) − J Z Z µ − e JEM Aµ cos θW 2 (12. One then obtains the standard electroweak theory supplemented by the effective dimension five operator ˆ ˆ ˆ Yν M −1 YνT ( i j Hi l j )( km Hk l m ) (12. the neutrino fields can always be rotated in such a way that there is no lepton mixing in (12. Indeed. namely.c. New renormalizable terms can then be added to the electroweak lagrangian.41) c (from tree diagrams with νL exchange) which.42) where (returning first to the Weyl spinor notation – rewrite (12. Note.43) † and VCKM ≡ UL DL is called the Cabibbo–Kobayashi–Maskawa (CKM) mixing matrix. We can now rewrite the matter couplings (12. because the right-handed fields only have diagonal couplings to the Bµ gauge field. It is also important that the matrices UR and DR drop out completely from the final lagrangian. cB neutrino Yukawa terms YνB A i j Hi l jA νL and the right-handed neutrino Majorana c cB ˆ mass matrix M B A νLA νL + h.37) neutrinos are massless and. therefore. Assuming that the entries of the matrix M are much larger than the electroweak scale one can integrate out the heavy right-handed neutrino fields. gives masses to the almost pure Majorana lefthanded neutrinos.

For n = 3 there is one physical phase. So only 2n − 1 phases can be removed. and are diagonalized simultaneously with the latter. In the world with three generations of quarks the C P symmetry is explicitly broken if this phase is different from zero. the n × n unitary mixing matrix has 2n 2 − n 2 = n 2 real parameters. We end this section with a few more remarks. γ and the phase δ fixed by experiment.45) The second one (Wolfenstein 1983) is only approximate (in this parametrization VCKM is unitary only up to order O(λ4 )). The Cabibbo–Kobayashi– Maskawa quark mixing matrix is a 3 × 3 unitary matrix. We know from experiment that λ ≈ 0. 2n − 1 phases can be removed by phase redefinitions of the physical quark mass eigenstates. This completes the discussion of the electroweak lagrangian in the Rξ gauge.12. The two most often used parametrizations of VCKM read  VCKM =  −sθ cγ − cθ sβ sγ exp (iδ) sθ sγ − cθ sβ cγ exp (iδ) c θ cβ sθ cβ cθ cγ − sθ sβ sγ exp (iδ) −cθ sγ − sθ sβ cγ exp (iδ) sβ exp (−iδ)   c β sγ c β cγ (12. but is sufficient for the present level of accuracy of the measurements. and are summarized in Appendix C. † This is no longer true if the Higgs sector consists of several Higgs doublets. unless for some symmetry reasons only one doublet couples to the up-type quarks and one to the down-type. the matrix VCKM remains invariant when we change all quarks by the same phase. and  VCKM ≈  3 1 − 1 λ2 2 −λ − iAλ5 η Aλ (1 − ρ − iη) 2 λ 1 − 1 λ2 2 −Aλ − iAλ η 4 Aλ3 (ρ − iη)   Aλ2 1 (12. Moreover. β. Thus. The Feynman rules for the full lagrangian can be derived by the standard methods of Chapter 2. The number of real parameters is therefore n 2 −2n +1. This is another reason (apart from economy) to consider one-doublet theory. .† The coupling of the unphysical charged Goldstone boson G + to physical quarks involves the CKM matrix as for the charged weak currents. with several Higgs doublets one generically has phenomenologically dangerous flavour changing neutral-current transitions mediated by Higgs boson exchange. Therefore. In the general case of n quark doublets.45) we clearly see the order of magnitude of the elements of the VCKM matrix. and from (12. n 2 −2n +1− 1 n(n −1) = 1 (n −1)(n −2) is the 2 2 number of phases. Notice that the phase transformations on the left.and right-handed quarks have to be the same to preserve real diagonal mass terms.44) with the angles θ. of which 1 n(n −1) are angles of 2 n ×n orthogonal matrix. Indeed. but not all of them are observable.3 Fermion masses and mixing 401 matrices which give the masses. For n = 2 no physical phase in the CKM matrix is present.23 (the Cabibbo angle) and A ≈ 1.

One should also remember that weak decays of elementary particles played a crucial role in the development of the electroweak theory. Moreover.402 12 Standard electroweak theory 12. g2 + g 2 α= g 2 sin2 θW .50)) and M Z = 91. energy of 200 GeV. we need to calculate them to the required accuracy in the electroweak theory parametrized by the set (12. We shall use these observables as the three measurements fixing the parameters of the lagrangian order by order in perturbation theory. These are: (a) very small momentum.e. the Higgs boson mass and the strong coupling constant. G µ = 1. as we 2 ˆ Z. where αEM = 1/137.0359895(61) is the fine structure constant. G µ and M Z .002 GeV is the physical Z 0 -boson mass (we distinguish it by the hat from the mass introduced as the parameter of the lagrangian. (b) the region of the Z 0 resonance at s 1/2 ≈ 91 GeV studied experimentally with very high precision in e+ e− collisions at LEP1. This will illustrate the basic logic of such calculations. sin2 θW = g2 . with non-renormalizable four-fermion lagrangian introduced on purely phenomenological grounds in 1933 (Fermi theory). sin θW is fixed in terms of G µ . which will reach the c.m. and a convenient choice is. The parameters of the electroweak gauge sector of the Standard Model lagrangian are the gauge couplings g and g and the vacuum expectation value v of the Higgs field.46). for example. Of course. . The fermion masses are known with good (or very good) accuracy for leptons and less precisely † At present (1999) the second phase of the LEP accelerator. For processes induced by the weak forces there are two regions of energy scale which have been studied experimentally particularly thoroughly. 4π 1 2 M Z = (g 2 + g 2 )v 2 4 (12.16639(2)×10−5 GeV−2 is the so-called Fermi constant directly related to the muon decay width (see ˆ (12.1867 ± 0. M Z = M Z . CKM angles. the so-called LEP2 is operating. they can be replaced by some combinations of them. q ≈ 0. Any further measurement is already a test of its predictions. At present the ˆ best measured electroweak quantities are αEM . It is clear from ˆ Section 12.1 that at the tree level we get α = αEM .46) Three independent measurements are thus necessary to fix these input parameters of the theory (beyond tree level they are renormalization-scale-dependent). shall see. processes (mainly weak decays of leptons and hadrons). in a search for some departures from the Standard Model at those high energies.4 Phenomenology of the tree level lagrangian The electroweak theory discussed in this chapter correctly describes all known phenomena induced by the electromagnetic and/or weak forces up to the energy scale O(100 GeV). αEM and M The complete set of the electroweak parameters also includes fermion masses. i. this distinction becomes important beyond the tree level).† We are going to concentrate on these two regions and to provide sample calculations in the standard electroweak theory.

for the decay µ− → eνµ νe . The currents are the same as in (12. At the tree level. Effective four-fermion interactions We proceed with the discussion of the tree level predictions. first we have to calculate the µ− decay widths. except for the top quark. in its full sense introduced in Chapter 7. and used relation (12. c ≡ cos θW .132). One should also remember that the predictions for the electroweak processes in the hadronic sector generically have in addition uncertainties coming from our limited knowledge of the quark current matrix elements taken between the hadronic states (i. All known elementary fermions.4 Phenomenology of the tree level lagrangian 403 for quarks. Charge current processes with four external light fermions. The V –A current is always built of a pair of fermions which form an SUL (2) doublet.12. will be discussed in Section 12.47).48) We have abbreviated our notation. It is important to stress that we introduce the effective four-fermion lagrangian as the four-point part of the integrand in the effective action (2. 12. The concept of the effective low energy theory. for example. and then we shall be able to discuss the predictions of the theory.43). can be described in the tree level approximation by the effective four-fermion lagrangian √ − LCC (x) = −2 2 G F Jµ (x)J µ+ (x) eff where √ 2 GF = πα 2 s 2 c2 M Z (12.25) (this relation is actually strictly valid for the parameters of the lagrangian in any order of perturbation theory in the MS scheme). since we can discuss the whole class of processes at once.1. the coefficient G F is universal for all four-fermion operators present in the lagrangian (12. Calculating the muon decay (µ → eνµ νe ) width in the tree approximation ¯ . and with their momenta q MW. Z 0 and Higgs bosons and the top quark). which is approximately local for the considered range of phase space.Z . The lagrangian (12.e. s ≡ sin θW . from the lack of precise theory of hadrons as bound states of quarks and gluons). are much lighter than the W ± and Z 0 bosons.47) accounts for the amplitude that. ¯ corresponds to the diagram in Fig.47) (12. It can be determined in terms of the muon decay widths. The origin of this approximate locality is in the decoupling of the heavy particles (W ± . and includes the appropriate CKM matrix element. This is a convenient way to organize certain calculations in the complete electroweak theory.6. As advocated above.

Historically. .00004) × 10−6 s the value G µ = 1.3 cancel out.1. 12.49) where τµ and µ are the muon life time and its full decay width.2. we get 1/τµ = µ = G 2 m 5 /192π 3 F µ (12. f (x) = 1 − 8x + 8x 3 − x 4 − 12x 2 ln x and G F has been renamed G µ . Kinoshita & Sirlin 1959). in a four-dimensional regularization scheme and with the on-shell renormalization (Berman 1958. Fig. Tree level diagram for muon decay in the Fermi theory. 12.48) is expressed in terms of the muon lifetime after correcting the theoretical prediction for O(αEM ) QED radiative effects. which will be denoted by G µ .2 and 12. From this formula and the measured muon lifetime τµ = (2. the so-called Fermi constant. its comparison with † This result is finite. i. the standard attitude is that (12. Once G F is determined by its identification with G µ . Actually.19703 ± 0.3 and integrating over the whole phase space. Tree level diagram for the charged currents interactions in the Standard Model. taking into account muon decays with one photon in the final state. 12. Computing the full set of diagrams shown in Figs. this calculation has been performed in the Fermi theory. one gets the following result:† 1/τµ = µ = G2 m5 µ µ 192π 3 × f m2 e m2 µ 1+ αEM 2π 25 − π2 4 (12.16639(2) × 10−5 GeV−2 is extracted. Traditionally (but not quite consistently for a tree level calculation).50) defines a new fundamental constant. The absence of the IR and mass singularities follows from the Lee–Nauenberg–Kinoshita theorem (Chapter 5). the G F in (12. In four-dimensional regularization schemes the counterterms shown in Fig.50) where αEM is the fine structure constant.404 12 Standard electroweak theory Fig. respectively. 12.e.

For instance. QED corrections to the muon decay in the Fermi theory. The relation (12. any other experimentally measured decay constant G ll (defined by the analogue of (12. The crossed circles denote counterterms.12.94 GeV (the superscript (0) means the tree level prediction for the physical W -mass). The recent (1997) combination of different measurements at LEP and Tevatron colliders gives ˆ MW = 80. Numerically.25)) 1/2 ˆ (0) ˆ MW = M Z 1 1 + (1 − 4A)1/2 (12. can be described by the effective four-fermion lagrangian √ µ (12. for vanishing fermion masses and in the tree approximation. 12. we get MW = 80. can then be used to exchange the third parameter sin2 θW for the accurately measured G µ .52) LNC = −ρ f f 2 2 G µ J Z J Z µ eff .39 ± 0.08. As said earlier.4 Phenomenology of the tree level lagrangian 405 Fig. The discrepancy signals the need of including radiative corrections (see next section). at the tree level the parameters α and M Z of the lagrangian have to be identified ˆ with the fine structure constant αEM and with the physical Z -boson mass M Z .48).50) for beta decay of lepton l into lepton l ) is a tree level test of the electroweak theory (with the O(αEM ) QED corrections included). for the W ± -boson mass one gets (using again (12. e− νµ → e− νµ ) in the limit q ≈ 0. With the three input parameters taken from experiment we can predict at the tree level the values of all other electroweak observables. The neutral-current transitions (like. for example.3.51) 2 √ ˆ (0) ˆ2 where A ≡ παEM / 2 G µ M Z . with G F = G µ .

the parameter ρ measured in the leptonic sector of the neutral-current transitions is ρEXP = 1.3). For instance. f f gR = −Q f sin2 θ eff f f (12.53) Calculating the Feynman amplitude corresponding to Fig.21) and (12. Z 0 couplings For the on-shell Z 0 boson coupled to leptons and quarks we have (by analogy with (12. 1994). J Z is given by (12. 12. . where the neutral current is µ JZ = f ¯ f γ µ PL T f3 f − f Q f sin2 θ eff ¯ f γ µ f (low) f f ≡ f f f ¯ f γ µ (gL PL + gR PR ) (12.52).406 12 Standard electroweak theory Fig. 12. interpreted as a low energy phenomenological lagrangian.48)). These predictions can be compared with experimental determination of the parameters ρ (strictly speaking ρ f f ) and sin2 θ eff f (low) defined by the lagrangian (12.25) (used in the derivation of (12.006 ± 0.25) we see that at the tree level 2 sin2 θ eff f (low) = sin θW (12.54) and.55) This follows from the relation (12.4 and using (12. we obtain the remarkable tree level prediction of the SM† ρff ≡ ρ = 1 (12. the differences are easy to understand) √ ˆ µ 0 L Z f f = −(ρ f 2 G µ )1/2 2 M Z J Z Z µ (12.56) µ where. moreover.53) with gL = T f3 − Q f sin2 θ eff . In this case the Higgs sector has the so-called custodial SUV (2) symmetry (see Problem 12.52).034 (Vilain et al. Tree level diagram for neutral-current interactions in the Standard Model. It follows from the identity of the two currents.4. which in turn is the consequence of using only Higgs doublets to break SUL (2) × UY (1) symmetry.57) √ 2 † Note the factor of 2 in the denominator in the effective coupling 2 2 G µ ρ = g 2 /2M Z cos2 θW obtained from the tree level diagram. as before.

A formal proof of this fact can be found in ’t Hooft (1971a. where A is defined below (12.21215. At the tree level one gets 2 sin2 θ eff = sin2 θ eff f f (low) = sin θW (12. part of the renormalization scheme is the choice of the input observables that are used to fix the parameters of the lagrangian. The complexity of the theory together with arbitrariness in the choice of the ‘best’ renormalization scheme often make it difficult to follow and to compare different approaches. The experimental values for the parameters ρ f and sin2 θ eff can f be obtained by measuring the widths and various asymmetries in the Z 0 → f¯ f decays.12.23151±0. then ρ f = ρ = 1. At present. at the tree level we get (Z 0 → e+ e− ) = 84. there is a consensus at least with regard to that latter point.5 Beyond tree level Renormalization and counterterms The electroweak theory is a renormalizable field theory. Strictly speaking. 12. The commonly . partially.58) Also.51).10 MeV.82 MeV. which should be compared with the LEP measurement 83.91 ± 0.59) where Nc = 1(3) for leptons (quarks). Thus. b).00022 with the tree level 2 eff 2 1 prediction sin θ f = sin θW = 2 [1 − (1 − 4A)1/2 ] = 0. which suggests the existence of a still deeper (more unified?) theory of elementary interactions. The latter ensures the absence of chiral anomalies (Chapter 13).5 Beyond tree level 407 (one also often defines vector and axial couplings gV = gL + gR . Crucial for renormalization is spontaneous (and not explicit) breaking of the gauge symmetry and also the matter content. higher order contributions. after the experimental results of LEP1. The importance of higher order effects can be judged from the comparison eff of the experimental world average sin2 θe = 0. In this book we shall be mainly concerned with practical aspects of the renormalization programme to provide a basis for systematic higher order calculations. A vast literature exists on precision calculations and precision tests of the electroweak theory which include all one-loop effects and also. The interplay of quarks and leptons in obtaining the anomaly-free theory is a big puzzle of the Standard Model. gA = gL − gR ). The decay width of Z 0 into a fermion–antifermion pair is √ ˆ3 4m 2f 1/2 Nc ρ f 2 G µ M Z (Z 0 → f¯ f ) = 1− ˆ2 12π MZ × (gL + gR ) 1 − f2 f2 m 2f ˆ2 MZ + 6gL gR f f m 2f ˆ2 MZ (12.

Therefore. the often used on-shell OS scheme) is obtained by finite renormalization. in the one-loop calculations considered in this book. s 2 (µ) and M Z (µ) defined in (12. (4) Obtain finite predictions for the computed observables in terms of the chosen physical ˆ input observables αEM . G µ . Complete fixing of the renormalization scheme then requires only a choice of the renormalization scale µ. G µ and 2(1) ˆ2 ˆ M Z are then identified with experimentally measured values αEM . and 2 (µ) and M (µ). Once we know the theory is renormalizable. at the tree the equations can then be solved for α(µ). µ 2 (1 − s 2 )M 2 2s Z and in this order one gets c2 = 1 − s 2 = 1 4παEM 1+ 1− √ ˆ2 2 2G µ M Z 1/2 2 ≡ c0 (12. the most pragmatic attitude to the renormalization programme (useful in a theory as complex as the electroweak theory) is to use lagrangian (12. We get (1) (0) (1) (0) ˆ 2(1) ˆ 2(0) ˆ2 αEM = αEM + δαEM .46)). s Z level we have πα (0) ˆ (0) MZ = MZ . Thus.61) (3) Compute in the one-loop approximation any other observable quantity of interest in terms of the same set of the renormalized parameters of the lagrangian. M Z ). G µ and M Z by inserting into the result of step (3) the 2 expressions for the parameters α(µ). M Z = M Z + δ M Z . G µ and the mass M Z of the Z 0 boson. s 2 (µ) and M Z (µ) obtained in step (2).60) αEM = α. in one-loop calculations we can proceed as follows: (1) Choose the set of the input observables in terms of which the others will be calculated. G µ = G µ + δG µ . The quantities αEM . dimensionally regularized. Since we know the infinities cancel out. Of course. The remaining freedom in the renormalization scheme is the choice of the renormalization conditions that fix the counterterms in the intermediate steps of the calculation. Any other renormalization scheme (for example. we are assured that calculation of any physical observable in terms ˆ of the input physical parameters (such as αEM . As in the rest of the book. G (0) = √ (12.5) as it stands simply without explicitly listing the counterterms. The Minimal Subtraction scheme is a convenient reference frame for comparison of different approaches. M Z ) must give a finite answer. we choose to work in the Minimal Subtraction scheme (actually in the MS scheme). G µ . keep only the terms which are finite in the limit ε → 0. ˆ We use the set ai = (αEM . we can as well subtract them in the dimensionally regularized result. without keeping track of the necessary counterterms. keep only the terms which are finite in the limit ε → 0. in terms of the (1) (1) same set of the renormalized lagrangian parameters.408 12 Standard electroweak theory ˆ used input parameters are αEM . .5) interpreted as the renormalization scale-dependent renormalized parameters in the MS (we shall use the set α(µ). (2) Compute all input observables ai . G µ and M Z . in terms of the parameters of lagrangian (12.

5). M Z (µ). We summarize here the renormalization constants of the complete Standard Model. The gauge invariance of the theory is easily taken care of if we introduce the renormalization constants at the level of the original lagrangian (12. One way of organizing such calculations is as follows (Barbieri 1993): calculate a renormalized matrix element (or any observable quantity) 2 2 M(1) = M(0) (α(µ).12.62) 2 ˆ 2(0) ˆ2 ˆ2 ˆ Using the relations M Z ≡ M Z = M Z −δ M Z .63) 2 ∂αEM ∂G µ ∂ MZ For higher order calculations it is necessary to go beyond the ‘short cut’ approach described above and to introduce the counterterms explicitly (see the end of Section 4. G µ . It is convenient to define the abelian Z g in the same way.4). G µ . s 2 (µ)) + δM(α(µ). s 2 (µ)) (12. Ward identities) and is necessary in studying the renormalization group equations for the parameters of the theory (Problem 12. where M Z is the measured physical (0) mass. G a denote the gluon fields. For the gauge fields and couplings we follow the definitions introduced in Chapter 8 for QCD but we change the notation in the QCD sector for a more convenient one in the complete Standard Model framework (gs is the QCD coupling constant. we can rewrite the last equation as follows: ˆ ˆ ˆ M(1) (αEM . M Z ) = M(0) (αEM . M Z ) + δM(αEM . M Z ) (0) (0) (0) ∂M ∂M ∂M ˆ2 − δαEM − δG µ − δ M Z (12. It is helpful at this point to recall the notation for QED in Chapter 5: there eB = .64) H = ZH H   −3/2 B  gs = Z G Z s gs      −3/2 g B = Z B Zg g      −3/2 B g = ZW Zgg The renormalization constants for the non-abelian couplings g and gs are defined as in Chapter 8.5 Beyond tree level 409 The above prescriptions define the one-loop perturbative calculation up to some terms of the second order which may appear depending on the particular way of realizing this programme. This is also helpful in getting a more profound flavour of the renormalization programme (for example. respectively. α(µ) ≡ αEM = αEM − δαEM and similarly replacing s 2 (µ) by the Fermi constant. the counterterms Z 3 and Z 1Y M are µ renamed as Z G and Z S .3). the superscript B means ‘bare’):  1/2 G aB = Z G G a  µ µ   1/2 a  aB  Wµ = Z W Wµ      1/2 B Bµ = Z B Bµ    1/2 B (12. M Z (µ). G µ .

Yu ¯ R L H B = Z Yu Yu ¯ R L H . † With massive neutrinos it is still convenient to work in the basis with diagonal matrix YiA and non-diagonal effective neutrino mass operator m AB ( i j l iA H j )( i j l i H j ). In Appendix C the vertex counterterms are then expressed in terms of δY s. For fermion fields we use the chiral notation of (12. for the Higgs boson mass parameter m 2 . If for QED we redefine as in (12. The Yukawa coupling renormalization is determined by the vertex counterterms Z Y s B B B defined by the matrix equations. B ‡ In the Rξ gauge it is convenient but not necessary to introduce the counterterm δv for the Higgs vacuum expectation value. Replacing the unrenormalized fields and parameters in the original lagrangian (12.and right-handed chiral parts of the UY (1) lagrangian. ˜ we get Z e = Z 3 . electron and photon selfenergy renormalization constants. the Yukawa matrices Yl A are diagonal. its self-coupling λ and the vacuum expectation value v we write‡  1/2  v B = Z H (v − δv)   −2 B (12.65) B 1/2 AC   d A R = (Z d ) dC R     −1/2 −1/2 A −1/2 A A B A A   Yl = Z H (Z e ) (Z Yl Yl )(Z l )    B −1/2 −1/2 AX −1/2 Y C  AC XY   Yu = Z H (Z u ) (Z Y u Y ) (Z q )    B −1/2 −1/2 AX −1/2 Y C  AC XY Yd = Z H (Z d ) (Z Y d Y ) (Z q ) Since there is no flavour mixing in the lepton sector.† The quark wave-function renormalization constants are hermitean. for example. Z 2 .5) with the help of the above relations and splitting all Z s: Z i = 1 + δ Z i . .64). for example. In the present case the same argument applies separately to the left. Z Yu Yu = Yu + δYu . and it is convenient to define the renormalization constant matrices δY s. The U (1) Ward identity ensures −1/2 −3/2 ˜ Z 1 = Z 2 and Z e = Z 3 . Finally. in particular since λ is additively renormalized (for example. eB = Z 3 Z e e. respectively. by fermionic loops proportional to the Yukawa couplings).410 −1/2 12 Standard electroweak theory (Z 1 /Z 2 )Z 3 e ≡ Z e e. where Z 1 . and it follows that the relation Z B = Z g can be imposed. Z 3 are the vertex.66) λ = Z H (λ + δλ)   −1 2 B 2 2  (m ) = Z (m + δm ) H The additive notation is again useful.2):  B 1/2 A  l A L = (Z l ) lA L     B 1/2 A   = (Z e ) eA R eA R     B 1/2 AC   q A L = (Z q ) qC L     B 1/2 AC  u A R = (Z u ) uC R (12.

ˆ ˆ MW and M Z must be distinguished from the µ-dependent lagrangian parameters MW and M Z (the same remark applies to the other masses too) which are given by (12.24). and then consider a sample of the most relevant physical applications. if the same physical quantity is calculated in terms of the same input physical observables to the same order in perturbation theory. Likewise. we recall that.4 can be conveniently divided into two classes: universal corrections to the gauge boson propagators (called sometimes the ‘oblique’ corrections) which in most cases give the dominant contribution. as we shall see. it will be convenient also to recast one-loop corrections to the selected processes in the form of corrections to the effective lagrangians introduced in Section 12.18) should be distinguished from several quantities directly related to experimental measurements and often also called in the literature the Weinberg angles.) We are already familiar with the fact that the counterterms which have been introduced at the level of the original lagrangian (12. radiative corrections to the processes described in Section 12.4. and in the four-component Dirac notation for fermions (appropriate after spontaneous gauge symmetry breaking) are presented in Appendix C. Although this dependence is. we must carefully distinguish parameters of the lagrangian which depend on the renormalization scale (and may also be gauge-dependent) and the physical quantities. the lagrangian parameter s 2 (µ) ≡ sin2 θW defined by (12. to estimate the uncertainty one would have to perform full next order calculation. In the following we discuss certain building blocks of the one-loop quantum corrections in the electroweak theory. (We do not explicitly write down the renormalization constants for the ghost fields but they are included in the Feynman rules given in Appendix C. the result does not depend on the remaining freedom in the choice of the renormalization scheme.5) are also sufficient to renormalize this theory after spontaneous breaking of the gauge symmetry. a given observable is calculated in terms of two different sets of input observables. For instance. of higher order. in principle.12. If.4 and the ˆ2 ˆ2 ratio of the physical gauge boson masses 1 − MW / M Z . The Feynman rules derived from the lagrangian including the counterterms. and process-dependent corrections consisting of corrections to the external fermion lines. It is worth stressing once again that. Finally. 2 2 c2 (µ) = MW (µ)/M Z (µ). in this book we shall never call the latter a sine of the Weinberg angle. corrections to the vertices and corrections . In particular. To avoid any confusion. We recall that in the MS scheme. Important examples are various sin2 θ eff introduced in Section 12. however. going beyond the tree level. Corrections to gauge boson propagators In general.5 Beyond tree level 411 we can generate the necessary counterterms. in any order of perturbation theory. finite order results can numerically depend on the choice of the input parameters.

Thus. much smaller than the ‘oblique’ corrections. for example. They cancel out. and are therefore frequently neglected. the dependence on the renormalization scale µ of the loop corrections to gauge boson propagators alone does not disappear. however. ‡ The gauge dependence of the gauge-dependent parts of the tree level propagators cancel with the gauge-dependent Goldstone boson propagators. The net result is equivalent to using the propagators in the unitary gauge. Note also that. The tadpole diagrams must be included if one wants to verify gauge invariance of the mass counterterms. in general.67) where Pµν and L µν are transverse and longitudinal projection operators. the corrections to the gauge boson propagators are not separately renormalization scheme-independent (in a given order) since they share the counterterms. 12. keep in mind. gauge-independent results one should include both types of correction.5. Moreover. The full vector-boson self-energy amplitude. in which only 1/ε terms are subtracted. after summing up one2 particle reducible diagrams. for example.‡ † Note the sign convention which is different from the one used for fermion self-energies.5. 12. in physical amplitudes and will be omitted in the following. in some schemes (like the MS scheme) inclusion of at least part of the process-dependent corrections may be necessary to avoid spurious singularities in the amplitudes in some kinematical limits (see below). the self-energy amplitudes at k 2 = MV are corrections to the physical masses. will be denoted by† i µν V1 V2 ( p) = iP µν T 2 V1 V2 ( p ) + iL µν L 2 V1 V2 ( p ) (12. Moreover. T s is cancelled by the vertex corrections. This is immediately seen for processes with external vector bosons (εµ k µ = 0) and with external fermions in the limit m 2f → 0 (conserved currents). We begin with the process-independent corrections to the gauge boson propagators. in the calculation in the MS scheme. like. In the general case these terms cancel (order by order in perturbation theory) Goldstone boson self-energy amplitudes as a consequence of the Slavnov–Taylor identities. with the vertex corrections. however. Process-dependent corrections are usually (but not always) in the commonly used gauges. generated by box diagrams. the longitudinal terms L1 V2 of the V gauge boson self-energies do not contribute to the physical matrix elements. that the splitting into universal and process-dependent parts is gauge-dependent and to get physical. . for on-shell amplitudes. which consists in general of two parts corresponding to diagrams shown in Fig. Contributions to gauge boson self-energy. This is convenient since the free fermion and boson propagators have opposite signs and in both cases. Also.412 12 Standard electroweak theory Fig. One should. Feynman (ξ = 1) or Landau (ξ = 0) gauge.

The diagram to be evaluated is shown in Fig. with the dimensionless couplings g → gµε/2 ) (ig1 )(ig2 )µε dn k −ig σ σ −ig λλ 2 2 (2π)n (k + p)2 − M1 k 2 − M2 × g µσ (k + 2 p)λ + g µλ (k − p)σ − g σ λ (2k + p)µ × g νσ (k + 2 p)λ + g νλ (k − p)σ − g σ λ (2k + p)ν where the couplings are. using our experience from the chapters on the scalar field theory and on QED. M2 ) + (4 p 2 + M1 + M2 )b0 ( p 2 .12. M1 . we explicitly compute in the ’t Hooft–Feynman gauge the contribution of the gauge boson loop to the gauge boson self-energies. For instance. The formulae for other contributions (from fermion. We recall in particular that. g1 g2 T 2 2 2 10A( p 2 .6.6 (in n = 4 − ε dimensions. M1 . As an illustration.11). b0 . Using the trilinear vector boson couplings given in Appendix C we get then for Fig. 12.6.5 Beyond tree level 413 Fig. due to UEM (1) .1)–(D. 12. renormalization scale-dependent but to this order of calculation they can be replaced by their tree level values. Vector boson contribution to the vector boson self-energy. 12. Higgs boson and mixed Higgs–vector boson loops) are given in Appendix D. in principle. M2 ) V1 V2 ( p ) = − (4π)2 2 2 + a(M1 ) + a(M2 ) + 2M1 + 2M2 − 2 p 2 3 With some labour the other contributions to the vector boson self-energies can be obtained in a fairly straightforward manner. With some algebra (and remembering that g µν gνµ = n) we find −ig1 g2 µε dn k i g µν (2k 2 + 2kp + 5 p 2 ) n [(k + p)2 − M 2 ][k 2 − M 2 ] (2π) 1 2 + (4n − 6)k µ k ν + (n − 6) p µ p ν + (2n − 3)(k µ p ν + p µ k ν ) T 2 V1 V2 ( p ) and L1 V2 ( p 2 ) can now be expressed in terms of the functions a. A V and B given by (D.

Comparing with the T γ case and isolating the combination of the two point functions which γ does vanish at p 2 = 0. moreover. Z Z ( p ).1(a)–(c) at p 2 = 0. the contributions of the diagrams shown in Figs D. finite by including counterterms shown in Appendix C. only the transverse part of the photon propagator is renormalized and. which depend on five independent renormalization constants δ Z W . δ Z g . which is useful for us one gets the following result: ˜ γ γ (0) = 23 1 5 M2 α 1 M2 − η + − ln W + ln W + 3π 4 2 2 µ2 4 µ2 Nc f Q 2f ln f m 2f µ2 (12.2(a)–(c) and D. in the limit p 2 = 0. MW = c2 M Z and all the parameters are renormalized at the scale µ. . one easily finds T γ Z (0) = −2 M2 α c 2 M Z −η + ln W 4π s µ2 (12. the two MW -dependent terms are the W ± and the Goldstone boson contributions. for instance. For the γ –Z 0 mixing amplitude diagram D.2 we can include γ γ the fermion loop corrections to the photon propagator and. γ T(L) T(L) T(L) 2 2 2 T(L) 2 Eight self-energies: γ γ ( p ) are made W W ( p ).2) as the corrections to the photon self-energy with appropriate changes in one of the vertices (see Appendix C). respectively. but to ˆ2 this order of calculation they can be replaced by the physical parameters αEM .1 and D.71) It is important that T Z (0) does not contain a fermionic contribution. This is a non-trivial and practical test of the computation.70) In the case of the photon self-energy.2(d) (with W + G − and W − G + circulating in the loop) does not cancel the contribution of D. At one loop it is given by the same set of Feynman diagrams (Figs D. L γ ( p 2 ) ≡ 0 and T γ (0) = 0. γ Z ( p ). M Z and the appropriate function of G µ .68) i.e. 3 for quarks) and charge of the fermion.2(d) separately vanish at p 2 = 0.414 12 Standard electroweak theory gauge invariance. Using the results of Section 5.1(a)–(c) plus D.2) i µν γ γ ( p) = i g µν − pµ pν p2 p2 ˜ γ γ ( p2 ) (12. δ Z B . δ Z H and δv.69) Here Nc f and Q f are the colour factor (1 for leptons. Of 2 2 course. the photon self-energy can be written as (see Section 5. Another self-energy which requires a comment is the γ –Z 0 mixing amplitude. µ is the MS renormalization scale and the divergence η ≡ 2/(4 − n) + ln(4π) − γ E may be removed by adding appropriate counterterms. It will prove useful to decompose it as follows T 2 γ Z(p ) = T γ Z (0) + p2 ˜ γ Z ( p2 ) (12.

LCC (Born) ≡ − eff 2π α 2 s 2 c2 M Z − Jµ J µ+ . Fig. for example. Corrections to elastic neutrino scattering which are singular at p2 = 0. 12.8(a). as it is clear from (12. numerically small and can be neglected.8(b).72) (12. 12. 12. 12.12. ‘Oblique’ correction to muon decay and elastic neutrino scattering.8(a) cancels against the similar singularity contained in the vertex correction shown in Fig. it cannot be incorporated as it stands into the neutral-current effective lagrangian.7. 12. 2 In the limit p 2 ≈ 0 and neglecting terms O(m 2f /MW ).47) and (12.5 Beyond tree level 415 Fig. The sum of these two graphs is finite at p 2 = 0.48). However.73) where.7). Being singular at p 2 = 0. .8. inserted into four-fermion Green’s functions (Fig. It is important to remark that an additional contribution which should in principle be included is the one shown in Fig. can be summarized in the form of corrections to the effective fourfermion lagrangians: LCC (oblique) = 1 − eff LNC (oblique) = 1 − eff T W W (0) 2 MW T Z Z (0) 2 MZ × LCC (Born) eff × LNC (Born) eff (12. 12. the singularity at p 2 = 0 of the diagram shown in Fig. the one-loop corrections to the gauge boson propagators.

M Z .75) T(t. m t . the dominant contribution to ρ is readily calculable. the Higgs sector and the fermion mass splittings within each generation. However. This symmetry is explicitly broken by the g coupling. we have already noted that ρ f = 0 in the limit of exact custodial SUV (2) symmetry. as any other physical observable. Indeed.416 12 Standard electroweak theory The first interesting physical result which we can already get at this point is that. The latter contribution is finite by itself since it vanishes when the splitting vanishes. G µ and the Higgs boson mass. as mentioned earlier. therefore. as a generic situation in the case of some truncations in this programme. m b .b) W W (0) 1 1 1 1 1 − a(m t ) − a(m b ) − (m 2 + m 2 )b0 (0. m t ) + − s 2 + s 4 m 2 b0 (0. The largest violation of the custodial SUV (2) originates from the large top quark–bottom quark mass splitting. Still.74) Of course. renormalization scaleindependent) result for ρ f in each order of perturbation theory.t) Z Z (0) = Nc α 4π 2s 2 c2 32 8 1 − s2 + s4 3 9 2 2 − a(m t ) − m 2 3 3 t 1 8 32 − m 2 b0 (0. m b ) b 3 b 3 9 .e. the dominant contribution to (12.74) is finite by itself. with one-loop accuracy and neglecting corrections other than the ones to the W ± and Z 0 gauge boson propagators. m t . is calculable (finite) in terms ˆ of αEM . m b . Thus. i. m b ) + − s 2 + s 4 m 2 b0 (0. With the help of the formulae for the fermionic contribution to the gauge boson self-energies collected in Appendix D one finds Nc α 1 1 m2m2 m2 t b t (m 2 + m 2 ) − = ln 2 b 4π s 2 12 t 6 m2 − m2 mb t b                                                      (12. we have T W W (0) 2 MW T Z Z (0) 2 MZ ρf ≡ ρf − 1 ≈ − (12. m b ) − m 2 − m 2 t b t 3 3 6 3 3 b T(t. It comes from the third generation fermion loops.b) Z Z (0) = Nc α 4π 2s 2 c2 8 4 1 − s2 + s4 3 9 2 2 − a(m b ) − m 2 3 3 b 1 4 8 − m 2 b0 (0. m t ) t t 3 3 9 T(b. a systematic application of the renormalization programme gives a finite (and. ρ f . the above combination of the self-energy amplitudes is not separately renormalization-scale-independent. m t .

. T 2 W W (k ) 2 = k 2 − MW − T 2 W W (k ) (12. At each order the physical mass of the W ± boson is given 2 in terms of the running mass MW (µ) by the requirement that the real part of the two-point non-local vertex of the effective action (also called 1PI two-point Green’s function). the e+ e− scattering close to M Z (in the resonance region).† This condition applies to the transverse part of W W . where a systematic perturbation theory result is insufficient.6 ± 5. and since m t m b ≈ 5 GeV) it can be approximated as ρ f ≈ Nc αEM m 2 t ˆ2 16πs 2 c2 M Z (12.76) This result is indeed renormalization-scale-independent (as for a finite contribum b (experimentally m t = 175. we see that it indeed has the Breit–Wigner form with the decay width of W ± given by 2 Im T W (k 2 = MW ) (we ignore here the subtleties related to the definition of the W mass of unstable particles and to the gauge-independent definition of their width).5 Beyond tree level 417 where Nc = 3 is the colour factor.6. we need the one-loop relation between MW (µ) and the above three measured parameters. which will be discussed later in this section. and should be replaced by the Breit–Wigner form of the full propagator. For that. equivalently. for example. The imaginary part of T W (k 2 ) can be obtained by applying the Landau–Cutkosky W rule to the Feynman integrals or by taking the imaginary part of the final result (see Appendix D). There are. we obtain in our approximation the first one-loop prediction of the electroweak theory: ρ f = 0. important physical situations such as the phenomena at energies in the vicinity of the physical masses of unstable particles.b) Z Z (0) ˆ2 MZ = α Nc ˆ2 4π 4s 2 c2 M Z m2 + m2 − t b 2m 2 m 2 m2 t b t ln 2 2 2 mt − mb mb (12. however.152) (or.5 GeV whereas tion). So far we have discussed the self-energies of the vector bosons in a systematic perturbation theory.b) W W (0) ˆ2 MW − T(t.78) ˆ2 vanishes at p 2 = MW .t) Z Z (0) ˆ2 MZ − T(b. G µ and M Z . † We have not yet given all the calculations necessary for solving this equation with one-loop accuracy in terms 2 ˆ2 of αEM .12. can be obtained by summing all 1PI diagrams). introduced in Section 2. It is easy to check that T(t. Since the transverse part of the full W ± propagator is given by (2.0096.77) Expressing the Weinberg angle in terms of G µ (in a one-loop calculation the tree level formula can be used).

Of course.9 as −i ≡ −i ( / L p PL + / R p PR + m) ≡ −i( / Vp + / A γ5 p + m) (12.80) Fermion self-energies We compute also the gauge boson and scalar contributions to the fermion selfenergy. The same procedure is necessary for the Z 0 boson propagator because of the 0 Z –γ mixing. We begin with the 2 × 2 matrix for the inverse propagator T 2 Z γ (k ) = k2 − − T 2 γ γ (k ) T 2 γ Z (k ) − T 2 γ Z (k ) T 2 Z Z (k ) 2 k2 − MZ − (12.1) and (D. the physical mass and the width of W obtained from this procedure are the same (due to Ward identities) as those defined for the transverse part.82) Performing the Dirac algebra and using the formulae (D.8) we obtain the . 12. The twopoint vertex function is now a 2 × 2 matrix and the longitudinal part of the full propagator is obtained by inverting this matrix. 12.79) Using 2 × 2 matrix inversion we find for the transverse parts of the Z 0 –γ propagators  1 1  T 2  G Z Z (k ) = ∼ 2  2  [ T Z (k 2 )]2 k − M Z − T Z (k 2 )  γ Z  T 2 − M2 − 2) −  k  Z Z Z (k  2− 2)  k (k γγ       1 1 T 2 G γ γ (k ) = ∼ 2 T 2 2 k − T γ (k 2 ) [ γ Z (k )]  γ   k 2 − T γ (k 2 ) − 2  γ 2 T 2)  k − M Z − Z Z (k       2 2 (k ) (k )   γZ γZ T   Gγ Z = 2 ∼ 2 2 [k − (k 2 )][k 2 − M 2 − T (k 2 )] − [ T (k 2 )]2 k (k − M 2 )  γγ Z ZZ γZ Z (12. Denoting the amplitude corresponding to the general self-energy diagrams shown in Fig.81) we have for Fig.9(a): −i = (i)2 µε −igµν dn k i (2π )n [(k + p)2 − m 2 ] [k 2 − M 2 ] µ k / ×γ (cV − cA γ 5 ) (/ + p + m) γ ν (cV − cA γ 5 ) (12.418 12 Standard electroweak theory For the longitudinal part of the W ± propagator the situation is more complicated as it receives a contribution from the W ± –Goldstone boson mixing.

M) 2 p2 × p cL cL PL + cR cR PR / + m cL cR PL + cR cL PR b0 ( p 2 . applying the general formula (12. m. m. m. which is the low energy effective theory obtained after integrating out the weak . M Z and G F . M) Similarly. It is also important to stress that the renormalization-scale-dependent coupling α(µ) in the complete electroweak theory is different from the analogous coupling in QED. the running parameters. for a scalar contribution (Fig. we now have to calculate loop corrections to the input observables to relate the renormalization scale-dependent lagrangian paramˆ eters to the measured αEM . These corrections enter into the calculation of any other (predicted) observable if accuracy beyond the tree level is needed.5 Beyond tree level 419 Fig. 12. M) 2 p2 × p |cV + cA |2 PL + |cV − cA |2 PR / − m(|cV |2 − |cA |2 ) n b0 ( p 2 . as usual.84) (12.83) For example.9.12.85) 2  MW α s2 1    + ln 2 e R (0) = (4π) c2 2 µ All the parameters are. result (4π)2 = n−2 a(M) − a(m) + ( p 2 + m 2 − M 2 )b0 ( p 2 .83) to the W ± and Z 0 corrections to the electron self-energy we get in the limit m e MW. Running α(µ) in the electroweak theory Following our general strategy.9(b)) we find (4π)2 = 1 a(M) − a(m) + ( p 2 + m 2 − M 2 )b0 ( p 2 . 12. General vector boson and scalar contributions to a fermion self-energy.Z :  2 2 MW 1 1 MZ α (1 − 2s 2 )2 1   + ln 2 + + ln 2  eL (0) =  2 2 c2 (4π) 2s 2 µ 4s 2 µ (12. M) (12. m.

12. Here we want to find the relation between αSM (µ) and the fine structure constant αEM = 1/137.86).10(a) but with the photon replacing Z 0 .10 (the black blobs denote one-loop diagrams) and one-loop W ± and Z 0 contributions to the electron self-energies.86) In any order of perturbation theory the lagrangian parameter α(µ) can be expressed ˆ2 in terms of αEM and M Z by calculating the Thomson scattering in the limit of k → 0 (k is the photon four-momentum) in the complete electroweak theory renormalized at some scale µ.87) † Box diagrams with a gauge boson (γ or Z 0 ) attached to the external electron lines vanish in the limit k = 0. and the subscript ‘SM’ will be omitted. we shall keep the distinct notation αSM (µ) and αEM (µ). At the one-loop level one has to compute the diagrams shown in Fig. The external photon is real with the four-momentum k ≈ 0. in the limit k = 0. The fine structure constant can be defined by the classical cross section for Thomson scattering (see. When necessary. Bjorken & Drell 1964) σ = 2 8π αEM 3 m2 e (12. for example.6. is reproduced by the following tree level lagrangian: L R Llow = − 1 (1 + δz γ )Fρκ F ρκ + (1 + δz 2 ) ¯ e i/ PL e + (1 + δz 2 ) ¯ e i/ PR ∂ ∂ eff 4 L / − e(µ)(1 + δz 1 ) ¯ e A PL e − (δm + mδz L ) ¯ e PL e R / − e(µ)(1 + δz 1 ) ¯ e A PR e e − (δm + mδz R ) ¯ e PR e (12. at k = 0 and with the electrons on-shell is exactly cancelled by the photon contribution to the external line renormalization (as demonstrated in Chapter 5). in a form such that it will then be easy to compare it with (12. sector. 12. This point will be discussed in detail in Section 12. By explicit calculation one finds that the result obtained in the complete electroweak theory. It is convenient to summarize the result of this calculation.† The diagram similar to Fig.10.420 12 Standard electroweak theory Fig. performed in the MS scheme. . 12. Corrections to the photon–electron interaction.

In the MS scheme and for k = 0 we get δz γ = − ˜ γ γ (0) L e(µ)δz 1 = − e(µ)(1 − 2s 2 ) 2sc 3 2s 2 1 + ln 6 s c + T γ Z (0) 2 MZ 2 MW µ2 T γ Z (0) 2 MZ − e3 (µ) (1 − 2s 2 )2 (4π )2 4s 2 c2 2 MZ 1 + ln 2 2 µ (12. moreover. To deal † Two facts about (12. . In this regime strong interaction corrections become large and non-perturbative.88) e3 (µ) s 2 (4π )2 c2 R e(µ)δz 1 = e(µ) − 2 MZ 1 + ln 2 2 µ (12. of course.89) where ˜ γ γ (0) is given by (12. and we get the one-loop relation αEM = α(µ) 1 + ˜ γ γ (0) + 2 = α(µ) 1 + s c T γ Z (0) 2 MZ 2 MW 4 + 2 µ 3 α 2 − 7 ln 4π 3 Nc f Q 2f ln f m 2f µ2 (12. and the result obtained for it in perturbation theory is not reliable.71).90) 2  MZ e2 (µ) s 2 1    δz R = − + ln 2 (4π)2 c2 2 µ Comparison with (12.69) and γ Z by (12.91) can be identified† with αEM .85).91) are worth noting: it is an equivalence relation (as we checked by explicit calculation) L L R R due to the unbroken UEM (1) symmetry and. One-loop W ± and Z 0 contributions to the electron self-energies have already been computed and are given by (12.92) We remind the reader once again that α(µ) ≡ e2 (µ)/4π ≡ αSM (µ).5 Beyond tree level 421 All the fields and parameters are. They give  2 2 MW (1 − 2s 2 )2 1 MZ e2 (µ) 1 1   δz L = − + ln 2 + + ln 2   (4π)2 2s 2 2 µ 4s 2 c2 2 µ (12.86) is immediate if we rescale the fields in the lagrangian (12. The effective charge L L R R e(µ) 1 + δz 1 − δz 2 − 1 δz γ ≡ e(µ) 1 + δz 1 − δz 2 − 1 δz γ 2 2 (12. those of the complete theory. One complication in using (12.92) is that ˜ γ γ (0) contains light quark contribution.12.87) to get the canonical form of the kinetic terms. δz 1 − δz 2 = δz 1 − δz 2 = 0 because of the weak contributions.

3. and the finite γγ quantity 2 2 ˆ had (M Z ) = ˜ had (M Z ) − ˜ had (0) γγ γγ γγ (12.50) as a fundamental constant. in practice. This can be achieved if we succeed in splitting the full result into QED corrections to the tree level effective lagrangian and the genuine one-loop electroweak corrections to the coupling G F . now. to write down the result of the one-loop calculation in the electroweak theory in the same form and to express the lagrangian parameters in terms of the constant G µ .0007 (Eidelman & Jegerlehner 1995). in the limit q ∼ 0 and neglecting fermion masses. we have to subtract from the full one-loop Green’s function the Green’s function obtained by calculating the diagrams in Figs. the remaining piece can indeed be summarized in the form of the effective fourfermion lagrangian (in fact this is true in any order of perturbation theory). 12. much more elegant to consider the constant G µ defined by (12.2 and 12. It can be computed via the dispersion relation with one subtraction at p 2 = 0 (since ˆ (0) = 0). It is.94) and Im ˆ ( p 2 ) can be expressed in terms of the measured ratio R(s) = σ (e+ e− → hadrons)/σ (e+ e− → µ+ µ− ) (where s is the centre of mass energy of the process): 2 Im ˆ (s) = 1 αEM R(s).0280 ± 0. however.422 12 Standard electroweak theory with this problem one writes ˜ γ γ (0) = ˜ had (0) + ˜ remainder (0) γγ γγ 2 2 = − ˆ had (M Z ) + ˜ had (M Z ) + ˜ remainder (0) γγ γγ γγ 2 where ˜ had (M Z ) can already be computed in perturbation theory. As we shall see. the effective four-fermion lagrangian has to be written as a sum of flavour-dependent . The dispersion relation reads ˆ had (s) = γγ 1 s π ∞ ds 4m 2 π Im ˆ (s ) s (s − s) (12.93) is the photon self-energy scalar function in the on-shell scheme. Evaluation of the integral gives the value ˆ had (M Z ) = γγ 3 0. Thus. Muon decay in the one-loop approximation The most straightforward approach to using the muon decay width for fixing the parameters of the electroweak theory at the one-loop level is simply to express them in terms of the width itself. This is true for all charged current processes with four external fermions but.

47) has the structure α GF √ π 2 1 + const. for example. for instance. 12. These predictions can. For muon decay. The subtle point of our subtraction procedure is that the result (12. As mentioned earlier.50)) will depend on the renormalization scale µ: αEM in (12. Another consequence is more delicate.12. This is. more precisely. Thus. τ → µ¯ µ ντ and τ → eνe ντ has its own Fermi ¯ ν ¯ constant. The subscript f of the couplings G denotes the dependence on all the fermion flavours of a given four-fermion operator. Due to the process-dependent corrections. the lepton universality is no longer strictly true and each decay: µ → eνe νµ .50) will be replaced by α(µ). with four-dimensional regularization prescriptions. ¯ e γµ γν γρ PL νe ¯ νµ γρ γν γµ PL µ ε α GF 1 α GF 1 + const. In the following we shall use G µ (best measured experimentally) as the input observable and the others will be the predictions of the theory at each order of perturbation.50) has been obtained with a four-dimensional regularization and in the on-shell renormalization scheme. O2 ε .95) or. they are no longer universal and become fermion quantum number dependent. One obvious consequence is that our final result (suppose it has the form (12. in general.5 Beyond tree level 423 pieces: LCC = − eff A.D √ 2 2 G f ¯ u A γ µ PL (VCKM ) AB dB † ¯ dC γµ PL (VCKM )C D uD (12. whereas now we work with the dimensional regularization and in the MS scheme. [4O2 + (O1 − 4O2 )] ≡ √ π 2 ε π 2 ε νe where O2 ≡ ¯ e γλ PL energy correction give ¯ νµ γ λ PL α GF √ π 2 µ . to the coefficient of the four-fermion operator ¯ e γµ PL νe ¯ νµ γ µ PL µ .3 cancel out.95) with similar terms (with obvious simplifications) for the leptonic and semi-leptonic processes. The diagrams with the fermion self- −4 1 + const. This difference is a higher order effect and can be neglected at one-loop accuracy. no longer true in dimensional regularization. which describes muon decay. be expressed as the values of the parameters κ f defined by the equations: G f = κ f G µ . By explicit calculation one can check that the vertex correction to the lagrangian (12.B.C. the vertex counterterm and the wave-function counterterms in Fig. we have to complete the calculation of one-loop corrections to the relevant piece of the effective lagrangian (12. O1 = √ + const.

which imply (γµ γρ γν PL )i j (γ ν γ ρ γ µ PL )kl = 4(γµ PL )i j (γ µ PL )kl .3 is in the S M replaced by the box diagram shown in Fig. The vertex graph in Fig. The finite parts of the last four graphs shown in Fig. Thus.97) depends on the choice of it.e.41). therefore. and the finite part in (12. The finite result would then be a finite correction to (12.97) Our calculation requires.424 12 Standard electroweak theory In four dimensions there is the identity† ¯ e γ λ γ κ γ ρ PL νe ¯ νµ γ ρ γ κ γ λ PL µ = 4 ¯ e γ κ PL νe ¯ νµ γ κ PL µ (12. the identity (12. in addition. i. With the above background we can proceed with the subtraction procedure. Another way of reaching the same result is to perform the one-loop QED calculations starting from the outset with the tree level effective lagrangian written in the ‘charge retention’ form (12. There is then no need to calculate matrix elements of the (so-called evanescent) operator (O1 − 4O2 ). 12. The finite result is the same as if (12. after the Fierz rearrangement ¯ e γ λ PL νe ¯ νµ γλ PL µ = ¯ e γ λ PL µ ¯ νµ γλ PL νe (12.96) and the two divergent contributions would cancel each other.98).98) which again follows easily from (A.47). However.41). an additional counterterm in the form of the operator (O1 − 4O2 ) and an additional renormalization condition. these finite pieces (Dugan & Grinstein 1991). (O1 − 4O2 ) ε (12. 12.11. † In the Weyl spinor notation it has the form e σ λ σ κ σ ρ νe ¯¯ ¯ ν µ σ ρ σ κ σ λ µ = 4 e σ κ νe ¯ ¯ ¯ ¯¯ νµ σ κ µ ¯ ¯ and follows directly from the completeness relation for the σ matrices summarized in Appendix A (A. 12.96) is not valid in 4 − ε dimensions and instead for the sum of the vertex and the self-energy corrections we get an additional piece 1 + const.96) was used in n dimensions.3 are equal to their obvious analogue in the full amplitude and the latter should be simply omitted. the original MS scheme must be supplemented by this renormalization condition. In the four-component notation it follows from the identity γµ γρ γν = γµ gρν + γν gµρ − γρ gµν − iεµρνσ γ σ γ5 εαβγ µ ε αβγ ν = −6gµν . It is possible to chose the subtraction scheme so that one subtracts all the 1/ε poles and.

We calculate the contribution to the effective lagrangian from the box diagram shown in Fig. Computing in the same limit (again introducing the photon mass) the effective lagrangian generated by the first diagram shown in Fig.96) we get (see (D. 12.5 Beyond tree level 425 Fig. the result for that diagram should be subtracted from the one for the box diagram. 12. 12. this result is obtained either by starting with the charge retention form of the four-fermion effective . As explained earlier.11 in the limit of zero external momenta. λγ ) ¯ νµ γ λ PL 2s 2 2(4π) 2 MW α = ln 2 × LCC (Born) eff 4π λγ µ ¯ e γλ PL e (12.11. W –γ box in the Standard Model.47) which is relevant for the muon decay (we have retained only terms singular in the limit λγ → 0).99) where LCC (Born) now denotes only that part of the tree level effective lagrangian eff (12. MW . Hence.15) and (D.3 one gets − α 4π λ2 1 γ + ln 2 2 µ × LCC eff (12.12.100) where µ is the renormalization scale. The corresponding amplitude is finite and is reproduced by the following effective lagrangian: LCC (γ W ) = i eff e4 2s 2 k k d4 k ¯ νµ γ λ PL (/ + m µ )γ ρ µ ¯ e γρ (/ + m e )γλ PL 4 2 − m 2 ][k 2 − m 2 ][k 2 − M 2 ]k 2 (2π) [k µ e W e In the limit of zero fermion masses and introducing the photon mass λγ to regularize the integral in the infrared one gets LCC (γ W ) = eff d4 k ik µ k ν 2 (2π )4 k 4 [k 2 − MW ][k 2 − λ2 ] γ ¯ νµ γ λ γµ γ ρ PL µ ¯ e γρ γν γλ PL × e4 2s 2 e Replacing k µ k ν → (1/n)k 2 g µν → 1 k 2 g µν under the integral and using the identity 4 (12.18)) LCC (γ W ) = eff e4 F(0.

The remaining box diagrams which have to be evaluated are shown in Fig.96). these four box diagrams generate the effective lagrangian LCC (W Z boxes) = − eff α 4π 1− M2 5 5 + 4 ln W × LCC (Born) eff 2 s 2 2s MZ (12.12.102) 2 2 each give −(e2 /16π 2 )(2/s 4 )(1 − 2s 2 ) ln(MW /M Z ) as the coefficient of the effective Born lagrangian. 12. The computation of the first two makes use of identity (12.101) Note that the dependence on the IR cut-off has disappeared. 12. Subtracting (12. when inserted in the full diagram for the muon decay. The other two diagrams give amplitudes which are equal to each other and. respectively.104) . lagrangian. or by imposing the particular renormalization condition for the new counterterm which is necessary in n dimensions if we work with the lagrangian (12. Remaining box diagrams for the muon decay in the Standard Model. after using the identity ¯ e γ λ γ κ γ ρ PL ¯ νµ γ λ γ κ γ ρ PL = 16 ¯ e γ κ PL ¯ νµ γ κ PL νe µ νe µ (12.103) ≡ BW Z × LCC (Born) eff Vertex corrections to be computed are shown in Fig. 12.426 12 Standard electroweak theory Fig. Together.13. In the limit of vanishing external momenta and neglecting the fermion masses. these corrections give LCC (vertex) = eff × LCC (Born) eff (12.100) from (12. The result is again the Born effective 2 2 lagrangian multiplied by the coefficients −(e2 /16π 2 )(1/4s 4 ) ln(MW /M Z ) and 2 2 2 2 2 2 4 −(e /16π )[(1 − 2s ) /4s ] ln(MW /M Z ).47).100)) we finally get eff LCC (γ W ) = eff M2 α 1 + ln W × LCC (Born) eff 4π 2 µ2 ≡ Bγ W × LCC (Born) eff (12.12.99) (and identifying to this order LCC with eff LCC (Born) in (12.

5 Beyond tree level 427 Fig. vertex and external wave-function renormalization corrections we also neglect the contributions of the Higgs and Goldstone bosons whose couplings are proportional to the masses of fermions to which they couple. the contribution of the external line renormalization to the effective lagrangian for muon decay reads LCC (ext lines) = eff 1 2 µL + νµ L + eL + νe L × LCC (Born) eff (12.83).13.12. The W ± and Z 0 gauge boson contributions to the muon and electron self-energies are given by (12. In the calculation of the box. Vertex corrections. 12. up to terms O(m 2f /MW ).105) (a) α 1 − 2s 2 = 4π 4s 2 c2 =− 2 MZ 1 + ln 2 2 µ (b) 2 MZ α 3 c2 M 2 5 − + ln 2 − 2 ln W 2 4π 2s 2 6 µ s MZ (c) 2 MZ 5 α 3 − 6s 2 c2 M 2 − + ln 2 − 2 ln W =− 2 4π 2s 2 6 µ s MZ (d) =− 3α 4π 2 M2 5 MZ − + ln 2 + ln W 2 6 µ MZ Diagrams renormalizing external fermion lines are shown in Fig. using the general formula (12. 12.106) where we have again neglected the charged lepton masses. We conclude that. one gets νµ (νe )L (0) = α 4π 1 2s 2 M2 1 + ln W 2 µ2 + 1 4s 2 c2 2 MZ 1 + ln 2 2 µ (12. For the neutrino self-energy.85). where ≡ (a) + (b) + (c) + (d) and                              (12. Due to the presence of the projections PL in the µW νµ and eW νe vertices only L s contribute to the muon 2 decay amplitude.107) .9.

109) After (12. For c2 (µ) = 1 − s 2 (µ) = . adding the correction (12.92).111) + −2 s0 c0 T γ Z (0) ˆ2 MZ + µL − ˜ γ γ (0) + Bγ W + BW Z + 2 + νµ L (12.111) can be used to express s 2 (µ) in terms of G µ . M Z and G µ .79) and (12.72) to the W ± boson propagator computed earlier we get T (0) Gµ πα(µ) 1 − 2 WW + Bγ W + BW.Z + 2 √ = 2 2 2 2 2 2s c M Z (µ) c M Z (µ) + µL + νµ L (12. we can replace the running parameters by the measured ones in all one-loop corrections. to the order we are working. A ≡ παEM / 2 G µ M Z = s0 c0 .108) Finally.110) can immediately be inverted.92) and (12. 2 Inserting the obtained α(µ) and M Z (µ) into (12.51). We get 1 2A A 2 = c0 − 2 1 + (1 − 4A)1/2 − 2 1/2 2 (1 − 4A) c0 − s 0 (12. (12. With these results we can predict any other electroweak observable. Eqs.80) we have 2 ˆ2 M Z = M Z (µ) + T 2 Z Z (M Z ) (12. as in (12. (12.61). Eq.e → 0) Bγ W + BW Z + 2 × α = −4 ln 4π s 2 +2× 2 MZ µ2 1 2 µL + νµ L 2 7 − 12s 2 MW +6+ ln 2 2s 2 MZ (12.428 12 Standard electroweak theory Collecting together all the corrections we get (note that corrections to the electron and muon ‘sides’ are equal in the limit m µ.113) √ 2 2 ˆ2 and.110) 2 The three equations can be solved for the parameters α(µ). this is the second equation which relates the measured quantities to the running parameters.109) we get π αEM Gµ (1 + √ = ˆ2 2 2s 2 c2 M Z where with one-loop precision (µ) = − T W W (0) 2 ˆ2 c0 M Z T ˆ2 Z Z (MZ ) ˆ2 MZ ) (12. From (12. s 2 (µ) and M Z (µ) in 2 ˆ terms of αEM .112) 2 2 and c0 and s0 are given by (12. This task is simplified if we notice that. The third one is the one-loop correction to the physical mass of the Z 0 boson.

we can obtain the one-loop prediction for the W ± mass. of course. Those relations contain logarithms. It should be remembered that the important step was the use of the strictly one-loop relations (12. We shall show there that it is possible to calculate αSM (µ ≈ M Z ) in terms of αEM with all large logarithms summed up to all orders.117) . the numerical accuracy is improved since these are the largest corrections to MW . Then using that αSM (M Z ) in (12. µ dependence in (12. Thus.12. Although it introduces only partial summation of higher order effects and. as such.115)) which can make all of them simultaneously small.109) is an important improvement. the most important contributions to r are contained in factors 2 2 2 2 ˜ γ γ (0) and [ T W (MW )/MW ] − [ T Z (M Z )/M Z ].5 Beyond tree level 429 instance.69) expressing αSM (µ) in terms of αEM .116) W Z can be approximated as r = ··· − 2 c0 ρ + ··· 2 s0 (12.115) This result is. (12. the result (12. From the previous equations we get r =− T W W (0) 2 ˆ2 c0 M Z + − ˜ γ γ (0) − 2 T 2 ˆ2 W W (c0 M Z ) 2 ˆ2 c0 M Z T s0 γ Z (0) − 2 c0 2 s0 T 2 ˆ2 W W (c0 M Z ) 2 ˆ2 c0 M Z − µL T ˆ2 Z Z (MZ ) ˆ2 MZ c0 ˆ2 MZ + Bγ W + BW Z + 2 + + νµ L (12.116) Numerically.115).6.115) we have defined the correction r . one gets the final one-loop result: 2 ˆ2 ˆ2 M W = c0 M Z 1 − 2 ˆ2 ≡ c0 M Z 1 − T ˆ2 Z Z (MZ ) ˆ2 MZ 2 s0 r 2 2 c0 − s0 + T 2 ˆ2 W W (c0 M Z ) 2 ˆ2 c0 M Z − 2 s0 2 2 c0 − s0 (12.114) and expressing c2 (µ) in terms of G µ .113).115) suffers from considerable inaccuracy introduced when expressing αSM (µ) in terms of αEM . renormalization-scale-independent. In the last step of (12. The second term in (12. Using the equation 2 ˆ2 MW = MW (µ) + T 2 W W (M W ) ˆ2 = c2 M Z + T 2 ˆ2 W W (c0 M Z ) 2 − c0 T ˆ2 Z Z (MZ ) (12. We shall return to this point in Section 12. as it should be when a physical observable is expressed in a given order of perturbation theory in terms of other physical observables.92) and (12. which is the common notation in the literature. and there is no choice of the scale µ (which is arbitrary from the point of the final result (12.

118) With a bit more work. Below we will compute radiative corrections to the effective Z 0 f f¯ lagrangian. i. higher order corrections can also be summarized in terms of that effective lagrangian. They become flavour-dependent.54) and (12. using the formulae for the b0 function.119) For the dominant contribution of the Higgs boson to r≈ 2 Mh αEM 11 ln 2 2 ˆ2 16πs0 3 c0 M Z r one finds (12.56).52) and relations (12. Indeed. The custodial SUV (2) symmetry is broken by the Yukawa couplings and by the UY (1) current. Two-body Z 0 decays can be described by a local effective lagrangian since they depend on kinematical invariants whose values are fixed in terms of the external particle masses.430 12 Standard electroweak theory because T 2 W W (M W ) 2 MW − T 2 Z Z (M Z ) 2 MZ ≈ T W W (0) 2 MW − T Z Z (0) 2 MZ ≈ ρ (12.e. All these observables are calculable from the effective lagrangian for the on-shell Z 0 coupled to a pair of fermions. . which are different for different fermions and different from the f analogous parameters present in the low energy neutral-current lagrangian. loop corrections to flavour conserving neutral-current processes can be summarized in terms of the parameters ρ f G µ and sin2 θ eff f (low) already introduced in the lagrangian (12.120) 2 2 The terms ∼Mh /MW cancel out.† It measures the partial decay widths of the Z 0 boson and various asymmetries of the decay products. Corrections to the Z 0 partial decay widths A very precise test of the electroweak theory comes from the measurements of the Z 0 boson properties at CERN and SLAC e+ e− colliders. This is an illustration of the important screening of heavy Higgs boson effects (Veltman 1994). Higher order effects can be parametrized by the effective parameters ρ f G µ and sin2 θ eff . introduced in (12.55) are modified. it is possible to show that the leading m t dependence of r is: r ≈− 2 c0 3αEM m 2 αEM mt t + ln 2 2 2 ˆ2 2 ˆ s0 16πs0 c0 M Z 12πs0 c0 M Z (12.52). † For several years CERN LEP and linear SLAC electron–positron colliders operated at the centre of mass √ energy s = M Z producing millions of Z 0 s at rest and allowing for detailed study of its couplings to light fermions. In the limit of q ∼ 0 and vanishing fermion masses. different for different flavour pieces of the lagrangian (12.

Diagrams for Z 0 boson decay.14.121) are the external fermion wave-function renormalization factors (see (12. and δZL. ελ (q) is the polarization vector of the Z 0 boson. we have for these diagrams the following amplitudes (the factor T f3 − Q f s 2 in the couplings can be written as ± 1 (1 − 2|Q f |s 2 ) with the sign + 2 (−) for up (down) type fermions) (a) = ∓ ie u( p1 )γ λ (1 − 2|Q f |s 2 )PL + (−2|Q f |s 2 )PR v( p2 )ελ ¯ 2sc T 2 ie γ Z (M Z ) v( p2 )ελ (b) = ∓ u( p1 )γ λ 2sc|Q f | ¯ 2 2sc MZ (c) = ∓ ie 1 u( p1 )γ λ (1 − 2|Q f |s 2 ) ¯ 2sc 2 T 2 Z Z (M Z )PL + (−2|Q f |s 2 ) (d) = ∓ 1 2 T 2 Z Z (M Z )PR v( p2 )ελ ie u( p1 )γ λ [FL PL + FR PR ] v( p2 )ελ ¯ 2sc ie u( p1 )γ λ (1 − 2|Q f |s 2 )δZL PL + (−2|Q f |s 2 )δZR PR v( p2 )ελ ¯ (e) + ( f ) = ∓ 2sc The factors FL and FR denote one-loop vertex amplitudes (note that e/2sc has 2 been factored out).R (m f ) + 2m 2f 2 V (m f ) + 2m f 2 m (m f ) (12. T Z (M Z ) is Z 2 2 the derivative at q = M Z of its transverse self-energy.12.14. Denoting the momentum of the Z 0 boson by q and the momenta of the outgoing fermion and antifermion by p1 and p2 . respectively. The relevant diagrams are shown in Fig. 12.81) and .5 Beyond tree level 431 Fig. 12.R = 2 L.

(Z 0 → f f¯) increases with increasing top-quark mass. . and which originates solely from expressing the t parameters s and c through s0 and c0 . the final one-loop result can be written in the form (12.123) are the ‘renormalized’ (scale-independent) vertex corrections.118) – which contains a term 2 proportional to m 2 /MW (12.125) The final result for the partial decay width is still subject to QED and (for f = q) QCD radiative corrections. one finds precisely the combinations (12. Collecting all the factors together. With O(αEM αs ) accuracy (including virtual and real emissions) we get (Z 0 → f f¯) = 0 (Z 0 3αEM 2 Qf → f f¯) 1 + 4π 1− αs . Since the effective coefficient of ρ is positive. † Indeed. expressing the MS parameters s. For all fermions f except for the b quark‡ the vertex corrections are small. in the popular on-shell renormalization scheme.42).432 12 Standard electroweak theory compare with (5.124) where 2 s 2 = s0 1 + 2 c0 2 2 c0 − s0 (12.77).R ). where the counterterms also contain finite parts (fixed by some physical conditions). For the effective mixing angle we find sin2 θ eff = s 2 1 − f c0 s0 T 2 γ Z (M Z ) 2 MZ − c0 s0 T γ Z (0) 2 MZ ˆ − FL + 1 − 1 2 2|Q f |s0 ˆ FR (12.† and is given by (12.122) ˆ2 MZ (12.123) for the renormalized vertex corrections.126) with 0 given by (12. each external wave-function renormalization factor is 1 δZL.56) with ρf = 1 − Here c0 2 ˆ FL ≡ FL + (1 − 2|Q f |s0 )δZL − 2 s0 ˆ FR ≡ FR − 2 2|Q f |s0 δZR T γ Z (0) + T ˆ2 Z Z (MZ ) +2 c0 s0 T γ Z (0) ˆ2 MZ ˆ ˆ + 2( FL − FR )      (12. ¯ ‡ For kinematical reasons Z 0 cannot decay into a t t quark pair. The corrections to ρ f and sin2 θ eff are then dominated by f the factor ρ present in – see (12.112). π (12.59). Its presence originates from expressing with one-loop accuracy the factor e(µ)/s(µ)c(µ) through measurable quantities.74) and (12. c and 2 e through the measurable quantities (as we did in the case of r ) and replacing the amplitude by the corresponding contribution to the effective action.

It comes from FL (the vertex corrections) and from the renormalization of the external b-quark line (diagrams (d)–( f ) in Fig.16 and Appendix C reveals that only diagrams (e) and ( f ) contain 2 the factor m 2 /MW . 12. therefore. We will compute the leading term. Inspection of the Feynman rules in Fig.16. Diagrams (c) and (d) are finite. 12. 12. Charged Goldstone boson couplings to the (t. Finally. b) quark pair. e mt +i √ 2s MW PR − mb PL mt e mt +i √ 2s MW PL − mb PR mt Fig. larger mass requires larger Yukawa coupling.15. 12. .14).5 Beyond tree level 433 ¯ Fig. 2 m 2 /MW t ¯ For the bb pair in the final state there is an additional contribution of order eff ˆ to the parameters ρb and sin2 θb . The relevant vertex corrections are shown in Fig. and the logarithmically divergent loop integrals can at most give ln m t .15. m t -dependent vertex corrections to the Z 0 bb decay.12. 12. The reason is that the top-quark mass is generated by spontaneous breaking of the gauge symmetry and. The loop integral goes as 1/m t . compensating factors m t /MW coming from the G + tb vertices. This is an interesting example of when the Appelquist–Carrazzone decoupling theorem does not work. coming from the Yukawa coupling G + tb of the Goldstone field t G + . diagrams (a) and (b) do not have any m t /MW factors in their vertices.

77). we find that for b quarks the new negative contribution (12. the full contribution of order m 2 /MW to the renormalization factor δZL ≈ t ± L can be obtained by calculating the contribution of G to the self-energy of the b quark.Z m t .15(e) gives† i e (e) e3 1 − 2s 2 m 2 t FL = 2 2 2sc 2s 2sc MW = −i dn k 1 2k λ PR PL 2 2 (2π )n / − m t (k − MW )2 k γ λ PL m2 3 e3 1 − 2s 2 m 2 1 t t −η − + ln 2 + O 2 8s 3 c M 2 (4π) 2 µ mt W dn k 1 1 PR γλ n k2 − M 2 (2π ) / − mt k W Similarly. for the Fig.128) 2 ˆ There is no similar contribution of order m 2 /MW to FR .128) of order † We also approximate the CKM matrix element Vtb by 1.127) Collecting all contributions together (and replacing s. i.15( f ) we find i e (f) e e2 m 2 t FL = 2 2sc 2sc 2s 2 MW × = 4 4 1 1 − s 2 PL − s 2 PR PL 3 3 / − mt k dn k k2 2 (2π )n (k 2 − MW )(k 2 − m 2 )2 t e3 m 2 4 2−n t − s2 3c M 2 4s 3 n W 4 + 1 − s2 3 dn k m2 t γ λ PL 2 (2π )n (k 2 − MW )(k 2 − m 2 )2 t 1 dn k n (k 2 − M 2 )(k 2 − m 2 ) (2π ) t W = 4 2−n e3 m 2 t − s2 3c M 2 4s 3 n W 2 + 1 − s2 3 dn k m2 t γ λ PL 2 (2π )n (k 2 − MW )(k 2 − m 2 )2 t m2 4 2 1 t s −η − + ln 2 3 2 µ 2 + 2 1 − s2 3 γ λ PL = −i e e2 m 2 t 2 (4π)2 8s 3 c MW 2 Finally. One obtains G± L (0) = 1 e2 m 2 t 2 (4π )2 4s 2 MW −η − m2 3 + ln t2 2 Q +O 1 mt (12. for vanishing external momenta. t (12.e. 12.434 12 Standard electroweak theory In the limit MW. c by s0 . . Comparing with ρ. 12. and neglecting the b quark mass Fig. c0 ) we get 2 2 ˆ 2 FL = 2 FL + 1 − s0 3 G± L + ··· = − α m2 t 2 2 ˆ2 4π s0 c0 M Z (12.

In general. namely. to introduce the effective low energy theory in its full sense defined in Section 7. ln(E/M). the subsequent decoupling of heavy particles gives field theories which are valid in a smaller and smaller energy range. As a result. in the MS scheme).6 Effective low energy theory 435 2 m 2 /MW to the parameter ρb overcompensates the positive universal contribution t ¯ of the same order. They can be considered as given in terms of the parameters of the original theory (matching conditions) or as free.7. they are described by lagrangians which consist of dimension four or less (renormalizable part) and higher dimension operators. It is also very interesting to look at the previous results from another perspective. which one encounters when calculating low energy. We have used the language of effective lagrangians. compared to effective eff Weinberg angles for other fermions. renormalizable parameters of the local effective theory. decrease of sin2 θb with m t .109) and (12. An effective theory for light particles can be obtained by calculating the Green’s functions with only light field external legs at small momenta (compared to heavy masses) and with all heavy and mixed light–heavy particle loops. apart from those present in the running coupling α(µ). but by inspection of (12.5 we have presented a sample of systematic tree level and one-loop calculations in the standard electroweak theory.6 Effective low energy theory for electroweak processes In Sections 12. µ . with the muon decay amplitude as an example.4 and 12. In the final step one redefines the light fields and parameters so that the whole dependence on the heavy fields is absorbed into those new. but that was merely as a local approximation to the effective action for the considered range of phase space. 12. parameters. In a renormalizable theory with several mass scales. E M. This result has not been explicitly written down. effective. Of course. but it is interesting as an illustration of the method. and to compare it with the systematic perturbative calculation performed in the previous section.50) one can convince oneself that it does not contain large 2 logarithms. The same correction (12. The UV divergences arising in these graphs must be renormalized by appropriate counterterms (for example. It is interesting to discuss this technique in the electroweak theory. processes in the full theory containing the large mass scale M. (Z 0 → bb) decreases as m t grows. ln(MW /m 2 ). Suppose we calculate at one-loop level the muon decay amplitude in the complete theory.128) is also responsible for the slightly slower.12. The main virtue of such a separation of mass scales is the possibility of easy resummation to all orders of large logarithms. summing up the terms O(α ln(MW /m f )) is less important than the terms O(αs ln(MW /m f )) (QCD corrections) discussed in the next section.

129). let us proceed with the construction of the effective theory. with one-loop QED corrections included. for instance. by decoupling heavy particles: vector and Goldstone bosons. i. The basic idea is then to apply the RGE in the effective theory to evolve G F (µ) down to the scale µ ∼ m µ . Higgs bosons and the top quark. Clearly. The Wilson coefficients of the effective four-fermion operators can be calculated in each order of perturbation by comparing the leading term in the expansion in powers of 2 (1/MW )n of the amplitudes calculated in the complete electroweak theory with the amplitude generated by (12. from the point of view of minimizing higher order corrections to the Wilson coefficient G F (µ). and for the light fermions are negligible compared to the f corrections O(αG µ ) to the Wilson coefficients G f .109)! And here comes an important observation: the correction to the Wilson coefficient depends on the renormalization scale in the effective theory (remember that in the calculation of the previous section the scale µ is the renormalization scale for QED corrections to the four-fermion operator) and involves a logarithm which can be large.129). In consequence. Putting aside (for a moment) the pathology of our example. . such a procedure gives the effective theory described by the lagrangian √ − L = LQED − 2 2 G F J +µ Jµ (12. example G F (µ) is renormalization-scale-independent at the one-loop level (a fact which is directly connected to the absence of logarithms of two separate scales in complete one-loop calculation).129) are the lowest dimension operators describing weak transitions. after decoupling heavy particles. to be able to identify the Wilson coefficient at one-loop accuracy. as well as terms with different chirality structure (V + A currents and scalar currents). It is minimized for µ ∼ MW . The local. we know already the one-loop corrections to the Wilson coefficient G F : they are given by (12. with QED corrections included. and has now the interpretation of the Wilson coefficient. pathological. and these are the logarithms that one would like to resum to all orders. we have to require equality of the full SM one-loop amplitude and the amplitude generated by lagrangian (12. 2 J − J µ+ . In the present. With 2 † A systematic expansion in inverse powers of MW would also give higher dimension operators. which generically give such logarithms of two different mass scales. for example. and to calculate the muon-decay matrix element at that scale.49). describing muon decay. Compared mf µ 2 to (12. operators present in lagrangian (12. obtained for four-fermion processes in the limit q → 0. In fact. and (12.129). since the counterterm diagrams in Fig 12. dimension six.† Thus.436 12 Standard electroweak theory This is pathological from the point of view of perturbative calculations in the full theory.3 cancel out. they are suppressed by O(m 2 /MW ).129) where G F is given by (12.e. at the tree level. the tree level effective lagrangian must be interpreted as the lagrangian at the decoupling scale µ ∼ MW .48) is the equation for the Wilson coefficient G F (µ) at µ ∼ MW .

Finally we calculate the muon decay matrix element by calculating finite parts of the one-loop QED corrections in the effective theory. also do not contain any logarithms of the renormalization scale. in terms of the low energy observable G µ . Choosing µ ∼ M Z . at the tree level that matrix element does not depend on the renormalization scale but it should be understood as taken at the scale µ ∼ m µ .98). One can see it by the following argument. for consistency. The power of the effective theory approach resides in the truncation of this series. that our example is pathological). the W ± mass. Next. does not undergo renormalization). is conserved (and. This is a product of the neutrino current (which does not interact with photons) and the current which. be taken as µ ∼ m µ to minimize two-loop corrections to the matrix element (remember. to minimize higher order corrections to it. Again. Suppose we calculate higher order QED corrections to the four-fermion operator written in the charge retention form ¯ e γ λ PL µ ¯ νµ γ λ PL νe . in the limit of vanishing fermion masses.51). (12. Thus. and of course each new coupling requires new renormalization constant (the theory is not renormalizable with a finite number of counterterms). and in the present case we use LCC only in the Born approximation. we expect to minimize two-loop corrections. is now α(M Z ). (12. Although those corrections do not contain logarithms of the renormalization scale. the latter should.) The main difference of this approach compared to the tree level calculation in the full theory is that α in (12. obtained by the Fierz rearrangement.6 Effective low energy theory 437 G F (MW ) = G F (m µ ). if they had been present. we include one-loop corrections to G F . with new dimensionful coupling constants ≤ O(G 2f ). In our pathological example G F is renormalization-scale-independent at any order. we should evolve G F (MW ) to G F (m µ ) with two-loop RGE in the effective theory.12. in principle. In this way we would resum all terms α(α ln(MW /m f ))n ) of the full theory. we can finally calculate the muon-decay matrix element at the tree level (to match the overall accuracy).109). and in consequence in (12. however. Such contributions are effective dimension > 6 operators. of course.48). all infinities generated by this four-fermion vertex can be renormalized by the field wave-function renormalization constants and the Wilson coefficient is renormalization-scale-independent (see Section 7. The perturbation theory based on lagrangian (12. .7).129) would include higher order contributions generated by perturbation in the four-fermion operator itself. Eventually we can express G F (MW ) in terms of G µ with the next-to-leading logarithm accuracy. eff consistent with the fact that operators with dimension > 6 have already been neglected in the construction of the effective theory. (Here we encounter the same pathology: one-loop corrections to the matrix element. This is an important improvement for calculating at the tree level electroweak observables at the W ± mass scale. for example. In the next order of perturbation theory. therefore.

This is the usual bottom-up attitude in which we study physics in a certain energy range without even knowing the more complete theory. The result is . the corresponding ‘minimal’ effective renormalizable theory is not only sufficient but also best suited to describe physics in that range (up to corrections O(q 2 /m 2 ). even at the level of pure QED we can discuss several steps of ‘effectiveness’: in the top-down approach. d. µ. τ ) and five quark flavours (u.n d . In the one-loop approximation we get: (n e αEM = αEM. b). and finally arriving at the QED of photons and electrons valid in the range 0 < E < m µ .n d . obtaining each time a new renormalizable effective theory valid in a smaller and smaller energy range. the numbers of the charged leptons. n u are. This effective QED is valid for ˆ the energy range 0 < E < M Z . namely the QED of Chapter 5.129). by its own coupling constant αEM. c. where n e . Since the theory is renormalizable. Actually.n u Nc f Q 2f ln m 2f µ2 +O m2 n m2 (n+1) (12.n d .n u ) (µ) 1 + α 3π n e .438 12 Standard electroweak theory QED as the effective low energy theory It is now worthwhile to discuss in more detail the connection between the complete electroweak theory and the renormalizable part of the effective theory defined by lagrangian (12. down.n u ) (µ). Each effective QED is described by its proper scale-dependent lagrangian (n e parameters and.e. three lepton flavours (e. and we are able to prove by the decoupling procedure that the usual bottom-up attitude is correct (Appelquist & Carrazzone 1975). respectively. the main virtue of effective theories (which is easy resummation of large logarithms) is achieved if this theory is used only in the ˆ energy range m b < E < M Z . s. Eventually. Thus. In the energy range of interest.and up-type quarks. in particular. its structure does not contain any information about the energy range of its applicability. n d . by repeating in these simpler cases the procedure used for calculating αEM in the complete electroweak theory). present in the effective theory. our goal is to use each effective QED only in the energy range m n < E < m n+1 . However. it still contains hugely different mass scales (fermion masses). where m 2 is the next mass threshold). In the next step. In the present case we are fortunate to know both the complete theory and the effective theory. starting with the decoupling of the weak sector and the even heavier top quark we obtain QED describing the interaction of photons.130) These equations are obtained by calculating in each effective QED the k → 0 limit of the Thomson scattering (i. The first point to stress is that the renormalized (in the M S scheme) coupling constant for each effective QED can be fixed in terms of αEM . we can integrate out subsequent quark and lepton flavours (in the order of their heaviness).

i.0) αEM (µ) = b1 (1. n d . and the µ scale in (12. 0)(αEM )2 dµ 2π (12.0. 0 < µ < m µ .0.0. n u ) with n e .0) of αEM (µ2 ≈ m 2 ) in terms of αEM is O(α 2 ) since no large logarithm appears e in this case in the one-loop (12. In order to accomplish this programme. for example. (In fact.0) running αEM (µ) (with β(2.131) the result already known from Chapters 5 and 6.130) is valid up to non-perturbative corrections to the photon propagator discussed in Section 12. n d .0) αEM (µ) = 1 (1. with all terms [α ln(m µ /m e )]n summed up. where m h is the mass of the lightest decoupled fermion.12.0. we have µ where (in general) b1 (n e . It is clear from (12. 0. we obtain αEM at the scale µ = m µ . between αEM (µ) and αEM (µ) for µ ≈ m 2 (for the sake of µ . Of course. we have to know the relations between the couplings of two different effective QEDs in the range of µ where both are (1.0) Thus. in an effective QED with several flavours.0. up to the next threshold etc.130).0) 1 (1.130) reflects quite clearly the problem of large logarithm resummation. The accuracy of the determination (1. Summing up leading logarithms. However.0) valid.130) contains large logarithms for any value of µ. 0)).) (1. 0.0) (2. αEM . 0) ln 2 4π µ0 (12. We can then switch to the next effective QED parametrized by the (2.0. When we calculate the low energy observable.0. For any µ in the range of the validity of the effective theory. and in this theory improved accuracy can be achieved only by calculating higher order corrections to this equation.2) αEM (µ).131) at some µ0 ≈ m e . n u ) = 4 (n e + 1 n d + 4 n u ) 3 3 3 (12. they are also absent in higher order corrections to this relation used in the effective QED of only photons and electrons. n u the renormalized coupling is defined only in the range 0 < µ < m h . until we reach (3.e. n u ‘active’ fermions.133) d (1.134) (1.0) αEM = αEM (µ ≈ m e ) + O(α 2 ).6 Effective low energy theory 439 given by the corrections to the photon propagator.0. in the one-loop approximation for the β-functions.0) Once αEM (µ) is fixed by (12.130) that (1.0) αEM (µ0 ) − µ2 1 b1 (1.0. n d . 0.5. for fixed n e .132) is the first coefficient of the β-function β(n e . (12. Eq. the solution reads: 1 (1. we can proceed in a different way.. (12. its µ dependence in the MS is determined by the RGE. (12.0.0. n d .3. We also remark that (12.130) can take any value in this range.

2) 2) 3π αEM (µ) αEM (m e 1 1 Nc f Q 2f f =t ln m 2f µ2 (12.) It is very easy to see that our procedure finally gives 1 = (1. e The procedure which we have outlined here is important in the following context. contains large logarithms (there is no scale µ for which all of them can be small).0) αEM (m 2 ) = αEM ). which are free.2) = αEM (µ) + O(α 2 ).137) (3.† (Another way to get (12.e. with one-loop accuracy we get from (12. we want to fix αSM (µ) in terms of αEM . (12. Since at µ ≈ M Z the logarithms in (12. .0.136) i. We see that in this simple example.0. as in Section 12. One way to get these relations is to calculate the same physical quantity in both theories. Eq.92) and to express αSM (µ) in terms of αEM with all the large logarithms summed up to all orders. Such corrections can be neglected as long as we work with one-loop RGE.135) illustrates the general feature of decoupling in the MS scheme: the dependence on the heavy states is absorbed into effective parameters of the lower energy theory.3. These are the so-called matching conditions.0.0) .0) (2.2) αSM (µ) = αEM (µ) 1 − α 4π M2 2 16 m 2 t − 7 ln W + ln 2 3 µ2 9 µ (12.0 decoupling procedure. This is indeed achieved by using our effective theory approach.3. therefore.135) is to begin with two-lepton theory and to follow the 1. and (12. but this is not a necessary choice of µ.0. the leading logarithms are summed up to all orders (we remember that (1. The final step is to use the matching condition (3. m2 µ (12.0) (2. for µ = m µ we get αEM = αEM be used for any µ such that (α/3π) ln(m 2 /µ2 ) µ (1. It is.0.135) for µ ≈ m µ .0) αEM (µ) = αEM (µ) 1 + m2 α µ ln 2 3π µ +O m2 e . Taking it as αEM .130): (1.0. (12. The one-loop approximation.3.5.0) + (3. renormalizable parameters from the point of view of the latter theory.0.135) can 1.137) are small we can rewrite † Of course. the two couplings are equal up to corrections O(α 2 ). αEM (µ) is the effective coupling constant of the ‘effective’ one-lepton theory obtained from the full two-lepton theory. for µ ≈ M Z . Then. Suppose that.440 12 Standard electroweak theory our examples we assume that all quarks are heavier than leptons). very desirable to go beyond the accuracy of (12. for µ ≈ m µ .92).

b → sγ ( B = 1). If both logarithms are small enough the whole procedure is self-consistent. αSM (M Z ) is used as input parameter in the renormalization group evolution up to scales µ ≈ 1016 GeV. Then. calculable without any new counterterms. it is interesting to discuss the predictions for them in the electroweak theory. for example. in 0 ¯0 particular. bd → bd ( B = 2) and s → ¯ dγ ( S = 1). Our final comment is that (12.138) requires for consistency the two-loop renormalization group evolution in that large µ range. In some very interesting physical questions. only the global baryon number is conserved.5).7 Flavour changing neutral-current processes We have learned already in Section 12. At the quark level. the lepton number conservation (separately for each lepton flavour) is exact.7 Flavour changing neutral-current processes 441 our result in the compact form: 1 1 1 + = αSM (M Z ) αEM (4π) M2 2 4 − 7 ln W + 2 3 3 MZ Nc f Q 2f ln f m 2f 2 MZ (12. In such cases the two-loop evolution of coupling constants is discontinuous. 12.92). for radiative flavour changing . for C P violation in the K –K system). If one wishes. see for example (Chankowski. One is that often several mass scales must be decoupled at the same time. renormalizability and the particle content of the theory. with a real or virtual photon. Płuciennik & Pokorski 1995). Such processes are. while the quark mixing explicitly breaks the quark flavour U (1)s. Moreover. Such transitions ¯ ¯ are responsible for physical processes like K0 –K0 and B0 –B0 mixing (and. of course. the flavour changing neutral-current processes involving quarks are generated in higher orders in the electroweak interactions. Since they are strongly suppressed in Nature. For quarks. Both neutral gauge boson and Higgs boson couplings are diagonal in the flavour mass eigenstate basis. the generic examples of flavour changing neutral-current ¯ ¯ ¯ transitions are the reactions d s → ds ( S = 2).137) illustrates two generic points. one can avoid the constant ( 2 ) in (12.12. It reflects the Ue (1) × Uµ (1) × Uτ (1) global symmetry of the lagrangian (12.138) where the sum extends to all fermions including the top quark. in theories of grand unification. Thus.3 that the structure of the electroweak theory is such that it ensures the absence of the tree level flavour changing neutral currents. Another general fact is the presence of some constant factors in the one-loop matching conditions. ln(µ/M Z ) ln(m f /M Z ) and the accuracy of (12.137) (the so-called finite threshold correction) 3 by using Dimensional Reduction (DR) scheme instead of the MS. This is the desired improvement over the one-loop formula (12. This follows from the SUL (2) × UY (1) gauge invariance.

it is approximately local and can be interpreted as an effective lagrangian. 12. 12. but are less interesting phenomenologically.17(a) with the W + W − exchange. for example.t d] q q d4 q [ ¯ s γµ PL (/ + m qi )γν PL d ][ ¯ s γ ν PL (/ + m q j )γ µ PL 2 2 2 (2π)4 (q 2 − m qi )(q 2 − m q j )(q 2 − MW )2 (12. in the ’t Hooft–Feynman gauge. due to Fermi statistics of the quark fields. Similar processes occur in the up-quark sector. for example.139) The colour indices are contracted within square brackets. (2. Predictions for physical processes are obtained by convoluting the quark offshell amplitudes with the hadron wave-functions or (in practice – as we shall see) by calculating the hadronic matrix elements of operators present in the effective lagrangian. (b) penguin diagram contribution to a S = 1 process. For the time being. we remain at the quark level and estimate qualitatively the expected order of magnitude for the S(B) = 2 and S(B) = 1 amplitudes. (a) Leading Standard Model contribution to K0 –K0 mixing. for the s d → s d transition. Suppose we want to . We shall be interested in the off-shell 1PI Green’s functions which build the effective action. In the limit of small (compared to the gauge boson mass) external momenta which interests us. as can be checked by general techniques developed in Chapter 2 for obtaining the Feynman rules from a lagrangian. we have Leff = 1 2 × e √ 2s 4 Vis Vid V js V jd i.17.c. we have ¯ ± the contribution from the diagrams shown in Fig. too. The two sets of diagrams enter with a relative minus sign. decays of strange and bottom mesons (like.442 12 Standard electroweak theory ¯ Fig. the celebrated b → sγ decay) and for decays like K → π e+ e− or B → K e+ e− . For the sum of the two diagrams 12. ¯ For the S(B) = 2 transitions. The corresponding contribution to the effective action is finite and has dimension [mass]−2 .132).17(a) (G denote the charged unphysical Goldstone bosons). j=u.

Such a mechanism of suppression of the flavour changing neutral-current amplitudes is known as the Glashow–Iliopoulos–Maiani (GIM) mechanism. the leading αG F term is suppressed by the same very small CKM angles as the double top quark exchange contribution. in fact. There are four possible ways of contracting the fields in Leff (12. the physics which determines the quark Yukawa couplings). (As we shall see. j=u. However. 2 † The dimensionless coefficient of the term 1/MW could also include the term ln(m q /MW ).202). As far as the u and c exchange is concerned. u. We see that the strong suppression of the flavour changing neutral-current transitions is indeed a prediction of the Standard Model.139) with the external fields that contribute to the 1PI four point Green’s function. because couplings of G ± to light quarks are suppressed by m q /MW .e.7 Flavour changing neutral-current processes 443 recover the contribution of the effective vertex (12. are in turn suppressed by light quark 2 2 masses. the successful predictions for the flavour changing neutral-current processes should be considered as fairly accidental and.139) into (2. We insert the effective lagrangian (12. For the box diagram there is no logarithmic term. ¯ On dimensional grounds.139).c Vis Vid = −Vts Vtd . to recover the result of the calculation in the original theory.† The same refers to contributions from G ± exchange. The other two contractions give the same result. unitarity of the CKM matrix has been used: i. j=u. they are O αG F m q /MW . which are proportional to larger CKM angles.c Vis Vid V js V jd (12.17(a). Therefore. We shall return to this point shortly. j=u.c 2 2 m qi 2 2 MW MW 2 mq 2 MW . 12. . we can take the limit m qi MW . from the point of view of the Standard Model. The remaining terms. it is probably the (unknown) physics beyond the Standard Model which is responsible for it (namely.139) is strongly suppressed by its very small mixing with the first two generations of quarks.139) to the Green’s function 0|T s ¯ d s ¯ d |0 . and make use of unitarity of the Cabibbo–Kobayashi–Maskawa (CKM) matrix. The contribution from the top quark exchange in (12.and/or c-quark exchange: Auc ∼ e √ 2s 4 1 2 MW Vis Vid V js V jd 1 + O i. Consequently. this prediction follows not only from the structure of the theory but also depends on the empirical pattern of the quark masses and mixing angles. Thus.12.140) In the last line. Two contractions give the two sets of Fig. we then get the following estimate for the s d → s d ¯ transition amplitude with double W -boson. we need an extra combinatorial factor of 1 in 2 front of (12. 2 mq j ∼ αG F Vts Vtd +O i. with the relative minus sign. i. this suppression is just the right one for accounting for the measurements).

the latter diagram is proportional to 1/MW . They are often important phenomenologically.444 12 Standard electroweak theory It is also interesting to comment on the S(B) = 1 transitions at the qualitative level.141) This time. the dimensionless coefficient of the αG F term† contains logarithms of light quark masses. In processes where the GIM mechanism is only logarithmic. one of the most interesting field theoretical aspects are the QCD corrections to them. (In practice.) This qualitative difference between penguin and box diagrams can be seen either by explicit calculation or by the following heuristic argument. The corresponding amplitude behaves as m2 u + O(Vtd Vts ). they receive contributions from box diagrams and also from the so-called penguin diagrams like. no additional suppression except for the usual αG F factor occurs in this case. In the first case we get either a tree level (UV-finite) diagram or a one-loop diagram with two four-quark vertices. 2 m2 MW c i. respectively. On 4 dimensional grounds.17(b). and their calculation strongly relies on the effective theory approach.141) can be considered only as an indication that non-perturbative strong interactions play an important role in the quark-level s → de+ e− transition amplitude. for example. (12. 2 m qi .141) could be used in a quantitative calculation only if the up-quark mass was larger than ∼1 GeV. The MW dependence of the amplitude can be traced back to the behaviour of the effective field theory diagrams which are obtained by contracting the heavy internal lines in the box and penguin diagrams. 12. but only logarithmic in the case of certain penguin diagrams. unlike the previously considered box diagrams. Thus (12. since the masses of the up and charm quarks are quite different. j=u. In the final subsection of this chapter we describe these techniques using the concrete example of C P violation in the neutral kaon system. which means that the power-like GIM mechanism occurs. we obtain a UV-divergent one-loop diagram with a single four-quark interaction. perturbative calculations can be trustworthy only if all the quark masses under logarithms are larger than the QCD confinement scale. The 1/k 2 from the photon propagator is cancelled by the k 2 /MW factor coming from the vertex function. Since the flavour changing neutral-current transitions involve strongly interacting quarks. Since this is not the case in Nature. We often say that the GIM mechanism is power-like in the case of box diagrams. the one shown in Fig. Auc ∼ αG F Vid Vis ln + O(Vtd Vts ) = αG F Vud Vus ln 2 2 † The leading term behaves like 1/MW . At one-loop. In the penguin case.c (12. The logarithmic part of the vertex function is independent of quark masses and vanishes due to unitarity of the CKM matrix. This signals the possibility of only logarithmic GIM mechanism.

17(a). The complete result for the Leff given by (12. ¯ At the quark level.139) is easy to obtain. 2 (1 − xi ) (1 − xi )(1 − x j ) (12. t is understood in (12. j = u. Summation over indices i. 12. x j ) = 1 + 1 xi x j A(xi .145) Thus. the diagrams shown in Fig. The remaining box diagrams (remember that (12.146) . x j ) = 1 xi − x j xi2 ln xi 1 − (i → j) + . we shall turn to the explicit form of their relation to the C P violation parameter εK .17(a) can be reproduced by the following effective four-quark interaction lagrangian Leff (box) = − where ˜ S(xi .c. we shall first discuss the calculation of these diagrams and QCD corrections to them.147) 2 G 2 MW F 4π 2 ˜ λi λ j S(xi . In the following. c. Next.144) where B(xi . j=u. x j ) 4 i j − 8B(xi . Using the formulae from Appendix D and the identity (12. this amplitude is generated by the s d → s d amplitude ¯ which receives the leading-order contribution from the box diagrams presented in Fig.t d ¯ s γ µ PL d (12.143) 2 G 2 MW λi λ j F A(xi . x j ) ¯ s γµ PL i. x j ) replaced by 1 x x [A(xi .142). x j ) = 1 xi − x j 1 xi ln xi − (i ↔ j) + 2 (1 − xi ) (1 − xi )(1 − x j ) (12.48) and xi = m qi /MW .12.139) accounts only for the W – W exchange diagrams) with either two unphysical Goldstone scalars in the loop or one such scalar and one W ± -boson are calculated in an analogous way. G F is given by (12. x j ) 4 (12.5).7 Flavour changing neutral-current processes 445 QCD corrections to C P violation in the neutral kaon system We are interested in C P violation in the kaon–antikaon transition amplitude (for the general discussion of C P violation in the kaon system consult Section 1. The sum of these diagrams is reproduced by an effective lagrangian which has the same form as Leff (W W ) with A(xi . The products of the CKM matrix elements ∗ have been denoted shortly by λi = Vis Vid and colour indices are contracted within brackets. 12. up to O(external momenta/MW ). x j ) − 2xi x j B(xi .142) 2 2 To this order.96). x j ) ¯ s γµ PL 4π 2 d ¯ s γ µ PL d (12. it can be written in the following form: Leff (W W ) = − with A(xi . x j )] (12.

12. gluons and quarks with momenta close to MW or m t (i. Here. 12.e. Sample QCD correction to K0 –K0 mixing. In order to find it. In other words. One could then ask why we intend to perform the perturbative logarithm resummation instead of applying non-perturbative methods immediately. This procedure will be described in some detail below. like that shown in Fig. Since MW 80 GeV and m t 175 GeV. ¯ The s d → s d amplitude presented in Fig. since we can then easily apply the (perturbative) renormalization group methods for resummation of large logarithms.e. for example. for instance. Q 2 stands for all possible Lorentz low invariants made out of momenta and masses of external particles in the considered amplitude.18. An important property of these corrections is that their ordinary perturbative ex2 pansion is actually an expansion in powers of αs ln(MW /Q 2 ) and αs ln(m 2 /Q 2 ) t low low rather then in powers of αs itself.5 GeV squared. 12. lattice calculations.446 12 Standard electroweak theory ¯ Fig.18. and the ordinary expansion in αs breaks down. We need to resum these large logarithms from all orders of the perturbation series.17(a) by connecting the quark lines by an arbitrary number of gluon lines. ¯ the purely perturbative resummation of large logarithms in the s d → s d amplitude ¯ will automatically provide us with quantities suitable for applying non-perturbative methods. As we shall see.17(a) is subject to QCD ¯ corrections. i. it is at most of order of the neutral kaon mass m K 0. The answer to this question is that the accessible non-perturbative methods are not suited for calculating quantities which receive significant contributions from rapidly varying field configurations. It is here where the main virtue of using the effective theory is. 12. . very much larger than QCD ) cannot be accounted for by non-perturbative methods in the case when external states are light hadrons at rest. The kaon–antikaon transition amplitude we look for is not perturbatively calculable even when large QCD logarithms are resummed. They are described by diagrams obtainable from those shown in Fig. the logarithms are large. one needs to use non-perturbative methods like.

12.20). The numerical value of C(µ) can be found perturbatively in QCD by requiring equality of the effective theory and full Standard Model amplitudes up to O(external momenta/MW ). and by all the QCD corrections to this diagram.17(a) (with only top quarks in the loop) and the tree level d s → s d ¯ amplitude generated by Leff (see Fig. On the effective theory side we would have to include one-loop QCD corrections to the vertex generated by Leff (see Fig. the result we have already obtained in (12.7). At the leading order in QCD.146) immediately implies that ˜ C(µ0 ) = S(xt . For simplicity. 12. At the next-to leading order in QCD.12. At the first glance. one needs to keep in mind that we intend to effectively resum large .18 on the Standard Model side.7): Leff = LQCD (gluons and light quarks) G2 M 2 − F 2W λ2 C(µ) ¯ s γµ PL d t 4π ¯ s γ µ PL d (12. 12. i.149) Here.148) 2 up to non-leading corrections in Q 2 /(MW or m 2 ) and higher order electroweak t low corrections.7 Flavour changing neutral-current processes 447 Fig. 12. 12. 12. Each of the two diagrams ¯ ¯ has a combinatoric factor of 2. At the leading order in QCD. The fact of basic importance here is that this amplitude can be reproduced to all orders in QCD by the following effective lagrangian (see Section 7. for the amplitude generated by the diagrams in Fig.19 obviously requires no renormalization.19. it may seem strange that we specify this renormalization scale even though the tree level diagram in Fig. we would need to take into account diagrams like the one shown in Fig. we shall discuss the procedure of large logarithm resummation in detail only for the top quark contribution.19). However. µ0 denotes the renormalization scale used in the effective theory when the comparison of the full and the effective theory amplitudes is performed. xt ) + O(αs ) (12. Tree level d s → s d amplitude generated by Leff . The Wilson coefficient C(µ) depends on the scale µ at which the effective theory fields and parameters are renormalized (see Section 7.17(a) with only top quarks in the electroweak loop.e. 12. this means requiring equality of the box diagrams in ¯ Fig.

Consequently.20. 12.e. are UV-divergent. they contain logarithms ln(µ2 /Q 2 ). even if we restrict ourselves to the leading terms in the evaluation of C(µ0 ). as in Fig. logarithms from all orders of the QCD perturbation series for the full Standard Model amplitudes. i. 12.19.18 with those in Fig. the box 0 low 2 diagram in Fig. After renormalization. Each of the diagrams shown has its partner in which fermion lines in the four-fermion vertex are differently contracted. not much larger than unity). Let us have a closer look at how the size of µ0 is being specified.20. up to 2 non-leading corrections in Q 2 /(MW or m 2 ).448 12 Standard electroweak theory Fig.18 contains logarithms ln(MW /Q 2 ) and ln(m 2 /Q 2 ). shown in Fig.20. In order to make the latter two t 0 0 logarithms small (i. It turns out that such large logarithms are absent so long as the renormalization scale µ0 is chosen to be of the same order as the heavy particle masses MW and m t in the full Standard Model diagrams. the t low low very logarithms we would like to resum. 12.e. On the other hand. 12. . we must make sure that no large logarithms reside in the neglected higher order QCD corrections to this coefficient coming from comparing amplitudes like the one in Fig. we choose µ0 of the same order as MW and m t . The dependence on ln Q 2 cancels in the low difference of the full and effective theory amplitudes to all orders in QCD. Thus. 12. One-loop QCD corrections to the d s → s d amplitude generated by the ¯ ¯ effective lagrangian. (This is a general fact: Wilson t low coefficients are insensitive to low energy behaviour of the considered theory). and which enters with a relative minus sign due to the Fermi statistics of quark fields. the QCD corrections to the leading order result for the coefficient 2 C(µ0 ) depend on αs ln(MW /µ2 ) and αs ln(m 2 /µ2 ). The QCD corrections to the tree level diagram. 12.

in the so-called vacuum saturation approximation. .‡ Consequently. Performing renormalization group evolution of the Wilson coefficient would be necessary even if the matrix element was perturbatively calculable.4)). renormalization group evolution of the Wilson coefficient from µ0 ∼ MW to µlow = (a few)× QCD must be performed before calculation of the final low energy amplitude.7.† Nevertheless. We recall the RGE for C(µ). the argument for using low µ in calculation of the matrix element would be identical to the one we have used to fix the size of µ0 . we need to evaluate the K0 → K0 transition amplitude generated by Leff .151) ¯ s γ µ PL d 2 → Z 2 ¯ s γµ PL d ¯ s γ µ PL d (12. This matrix element is usually parametrized as follows (for states normalized to unity): 2 ¯ s d3 x K0 |(¯L γµ dL )(¯L γ µ dL )|K0 = 1 BK (µ)m K f K s 3 (12.152) The constant Z 2 in the above equation is the usual QCD renormalization constant of the quark wave-function (see (8. The where m K is the kaon mass. (7. and f K parameter BK (µ) cannot be calculated perturbatively because it depends on QCD interactions at energy scales close to QCD . i. The RGE for the Wilson coefficient C(µ) is discussed in Section 7.150) 161 MeV is its decay constant. it is still a function of the renormalization scale µ used for UV renormalization of amplitudes generated by the considered four-quark interaction.138).7 Flavour changing neutral-current processes 449 ¯ Once the Wilson coefficient C(µ) is found. 0 low Thus.12. (7. as well as renormalize couplings and fields in the effective part: C(µ) → Z c C(µ) and ¯ s γµ PL d (12. In order to render the amplitudes generated by Leff finite. This requires finding the matrix element of the four-quark operator present in Leff between the kaon and antikaon states at rest.150).153) † The factor BK (µ) is defined so that BK (µ) = 1 when only the vacuum state is introduced as an intermediate state between the two currents in the matrix elements in (12.e. We would then argue that the presence of large renormalization scale µ0 worsens the perturbative expansion of the matrix element due to large logarithms ln(µ2 /Q 2 ). we need to perform the standard renormalization of the LQCD part of it.139): µ d − γc C(µ) = 0 dµ (12. maximal energy scales available in lattice calculations are much smaller than QCD × (size of the lattice)/(lattice spacing). ‡ For instance. Non-perturbative methods used for evaluation of BK (µ) are not applicable when µ is arbitrarily large.

154) We are going to use this equation for evolving C(µ) from µ0 down to µlow at the leading order in QCD (i. the renormalization constant Z c in the MS scheme is found from pole parts of one-loop diagrams shown in Fig. colour indices are contracted within brackets. 2 Up to O(Q 2 /MW ). By power counting.157) The pole parts of Fig. (d) reads L1 = 2 × (−X )(−igs )2 (−i)i2 µε q q dn q [ ¯ s γµ PL (/ + m d )γν T a d ][ ¯ s γ ν T a (/ + m s )γ µ PL × 2 (2π)n (q 2 − m d )(q 2 − m 2 )q 2 s d] (12. one finds the pole part of L1 L1 pole =− d ¯ s γ ν γ ρ γ µ PL T a d (12. (d) and the ‘negative partners’ of Fig. The SU (3) generators and gauge coupling are denoted by T a and gs .450 12 Standard electroweak theory where γc = −µ dZ c 1 . the contribution to the effective lagrangian of Fig. we can verify that only the double / term in the numerator q can contribute to UV divergence. 12. Thus. dµ Z c (12.156) Substituting this result into the considered amplitude. low (b) and the ‘negative partners’ of Fig.148) with the external fields after its insertion into (2. Thus.e.20(a). 12. The factor of 2 is the combinatorial factor for the diagrams summarized by L1 .20. 12. 12. respectively.20(c). the only integral we need is dn q qρqσ ig ρσ + finite terms = (2π)n (q 2 − m 2 )(q 2 − m 2 )q 2 32π 2 ε s d 2 Xgs ¯ s γµ PL γρ γν T a 2ε 16π (12. It can be traced back to the possibility of two different (but indistinguishable) contractions of the fields in the four-fermion operator in (12. we need to find the explicit O(αs ) contribution to the renormalization constant Z c .155) 2 where X = G 2 MW λ2 C(µ)/4π 2 abbreviates the factor standing in front of the t F four-quark interaction in Leff . 12.20(c). 12. (b) are found in an analogous manner and yield the contribution to the effective lagrangian: L2 pole =− 2 Xgs ¯ s γν T a γρ γµ PL γ ρ γ ν T a 16π 2 ε d ¯ s γ µ PL d . As usually.20(a).202) to build the Green’s function 0|T s ¯ d s ¯ d |0 (observe that for L1 all contributions are distinguishable and see Fig.19). At the leading order in QCD. in the so-called leading logarithmic approximation).

After all that we find 3 4 Lk k=1 pole = 2 Xgs 12 ¯ s γ µ T a PL 16π 2 ε d ¯ s γ µ T a PL d d − 16 ¯ s γ µ PL 3 d ¯ s γ µ PL (12.98) for anticommuting fields. which gives pairs of indistinguishable contractions). After all these simplifications. we can get rid of the SU (3) generators T a by applying the identity (T a )αβ (T a )γ δ = 1 δαδ δγβ − 1 δαβ δγ δ 2 6 (12. Now.159) Finally.5 on muon decay). ( f ) (and their ‘negative’ companions) are summarized by L3 pole =+ L4 pole 2 Xgs ¯ s γν T a γρ γµ PL 32π 2 ε 2 Xgs ¯ s γµ PL γρ γν T a =+ 32π 2 ε d ¯ s γ ν T a γ ρ γ µ PL ¯ s γ µ PL γ ρ γ ν T a d d d Compared with L1 and L2 .96).158) is simplified by using γν γρ γµ γ ρ γ ν = 4γµ + O(ε) and (T a T a )αβ = 4 δαβ .7 Flavour changing neutral-current processes 451 Pole parts of Fig. we apply the identity (12.161) with colour indices contracted within brackets.202) (we recall once again that for L1 and L2 all contractions are distinguishable and the factor of 2 is necessary to match the combinatorics of the tree level diagrams described by the operator in (12.12. It holds in four dimensions only. we require that the divergent . 12.148).158) In order to simplify the Dirac algebra in the first expression. L3 and L4 have no combinatorial factors of 2 since two indistinguishable contractions appear when L3 (L4 ) is used in (2. The second term in (12.160) together with the Fierz rearrangement formula (12. we obtain the final result 4 Lk k=1 pole =− 2 Xgs ¯ s γ µ PL 12π 2 ε d ¯ s γ µ PL d (12.20(e). Summing up all contributions one finds for the pole part 4 Lk k=1 pole =− 2 Xgs 16π 2 ε ¯ s γµ γρ γν PL T a d ¯ s γ ν γ ρ γ µ PL T a d d d + ¯ s γν γρ γµ γ ρ γ ν PL T a T a d ¯ s γ µ PL − ¯ s γµ γρ γν PL T a d ¯ s γ µ γ ρ γ ν PL T a (12. but we are allowed to use four-dimensional Dirac algebra so long as we are interested only in calculating one-loop renormalization constants in the MS scheme (see also the discussion in Section 12.

153) is easily found to be C(µ) = exp ln µ ln µ0 d(ln µ ) γc (µ ) C(µ0 ).168) where b1 is the coefficient in one-loop RGE for the QCD coupling constant (see Chapters 6. we substitute N f = 4 to b1 . π (12. Consequently.169) The value of b1 = − 1 (11 − 2 N f ) depends on the number N f of quark flavours 2 3 active in the considered effective theory.163) Since the quark wave-function renormalization constant is known to be (see Chapter 8) Z2 = 1 − 2 gs 4 + O(gs ) 6π 2 ε (12.166) 2 gs αs 4 2 + O(gs ) = 1 + + O(αs ). In the full Standard Model. the b quark should be decoupled.162) which implies that 2 Zc Z2 − 1 = − 2 gs 4 + O(gs ) 12π 2 ε (12. we obtain (up to higher orders in αs ) C(µ) = exp = exp 1 π 1 b1 ln µ ln µ0 d(ln µ )αs (µ ) C(µ0 ) dαs αs (µ0 ) C(µ0 ) = αs αs (µ) −1/b1 αs (µ) αs (µ0 ) C(µ0 ) (12. 4π 2 ε πε (12. 8 and (6. we substitute N f = 5 to b1 when running the Wilson coefficient from µ = µ0 to µ = µb ∼ m b .45)): µ dαs αs 2 = αs b1 + O(αs ) dµ π (12. the charm quark can be . the top quark is already absent.165) The solution to the RGE (12.164) the renormalization constant Z c must be equal to Zc = 1 + Consequently. (12.167) Thus. Below µb .452 12 Standard electroweak theory part of the one-loop diagrams we have calculated is cancelled by the counterterm 2 Leff = −X (Z c Z 2 − 1) ¯ s γ µ PL d ¯ s γ µ PL d (12. At this latter scale. we have N f = 6. In the effective theory introduced at the scale µ0 . Finally. γc (µ) = αs (µ) 2 + O(αs ).

next-to-leading contributions to the matching on the full theory side are given by two-loop diagrams like the one shown in Fig. one needs to include O(αs ) contributions to the matching between the full and effective theories at the scale µ0 . As mentioned earlier. Running the Wilson coefficient down to the low energy scale µlow at the next-to-leading order in QCD requires knowing the two-loop contribution to the anomalous dimension γc .170) is replaced by 2µ0 . Achieving this resummation has been the main reason for introducing the effective lagrangian (12. one finds C(µ0 ) including the O(αs ) part of it. we have resummed powers of αs ×(large logarithm) from all orders of the usual perturbation series.148). 12.18. In other words.7 Flavour changing neutral-current processes 453 Fig. i. In order to achieve this. These scales have to be of the same order of magnitude as MW . m b and m c . On the effective theory side.170) automatically reduces its dependence on the arbitrariness in the choice of scales µ0 . Details of the next-to-leading QCD calculation are presented in Buchalla. too. Resummation of large logarithms can be systematically performed at higher orders in QCD. a factor of 2 is still allowed. summing all terms αs (αs ln(µ0 /µb ))n etc. one has to evaluate finite parts of the diagrams presented in Fig. Buras & Harlander (1990).170) The expression we have obtained for C(µlow ) is perturbative in αs but exact in αs ln(µ0 /µb ). the leading order (LO) contribution to the αs (µb ) αs (µc ) 6/25 αs (µ0 ) αs (µb ) 6/23 ˜ S(xt .e. and to the running of the Wilson coefficient down to the scale µlow . respectively. when µ0 in (12. Including the O(αs ) part of C(µlow ) in (12. For instance.20. xt ) + O(αs ). 12. it requires calculating the O(αs ) part of C(µlow ) exactly in αs ln(µ0 /µb ). αs ln(µb /µc ) and αs ln(µc /µlow ). However. Equating the two amplitudes renormalized at the scale µ0 ∼ MW . decoupled at µ = µc ∼ m c . In the MS-scheme. At the next-to-leading order. αs ln(µb /µc ) and αs ln(µc /µlow ).21. it is given by pole parts of two-loop corrections to the effective four-quark vertex. µb and µc . Sample two-loop QCD corrections to the effective vertex.21. The ultimate expression for the Wilson coefficient at scales µlow < µc is thus C(µlow ) = αs (µc ) αs (µlow ) 6/27 (12.12. 12. for example. 12. An example of such a correction is shown in Fig. changing them by. .

what matters in practical calculations are only orders of magnitude of renormalization scales.454 12 Standard electroweak theory coefficient C(µlow ) changes by αs (2µ0 ) 6/23 − 1 C LO (µlow ) αs (µ0 ) αs (µ0 ) 2 =− (ln 2) · C(µlow ) + O(αs ) (12. Consequently.273) C LO (µlow ) = εK Im(M12 A2 ) 0 exp(iπ/4). This is ¯ because the dominant contribution to A0 originates from the tree level s → u ud ∗ weak transition.270). On the other hand. the above dependence on µ0 can serve as an estimate of the error we make by neglecting this very term. 2 the µ0 dependence of C(µlow ) becomes a O(αs ) effect (see.44). the imaginary part of M12 is equal to 2 2 ¯ ¯ Im M12 = Im K0 |HW |K0 + O(HW ) = − Im K0 |HW |K0 + O(HW ) (12. the amplitude A0 is approximately real.265) would be equal. We recall that (1.261). when the O(αs ) term is calculated explicitly. √ 2 m|A0 |2 (12. In order to simplify further discussion.172) where M12 is the hermitean part of the weak hamiltonian. M¨ nz & Pokorski (1995) for an extensive discussion of this point in the u case of the decay b → sγ ). Therefore. Buras.171) π Before the O(αs ) term in (12. for example. Note that ¯ if A0 was exactly real. not their precise values.174) .44) of the CKM matrix.170) is found. the effect of changing the scale by a factor of 2 can always be absorbed into unknown higher order corrections. Choosing the proper order of magnitude for the renormalization scale is important for (asymptotic) convergence of the perturbation series. Our approximate expression for εK now becomes proportional to Im M12 εK Im M12 exp(iπ/4) √ 2 m (12. we shall briefly describe how the evaluated Wilson coefficient is used to find the C P-violating observable εK (defined in Chapter 1). Thus. A0 is to a good approximation proportional to V12 V11 which is real in the standard parametrization (12. Misiak.173) As follows from (1. one finds that its µ0 dependence is precisely such that it cancels the µ0 dependence of the leading term. When this parametrization is used. The same phenomenon can be proven to happen at any order in perturbation theory. On the other hand.273) and A0 is defined in (1. m is introduced in (1. In the remainder of this subsection. The renormalization-scale-dependence is always formally a higher order effect. then the parameters εK in (1. we shall assume that we use the parametrization of the CKM matrix as in (12.270) and ε in (1.

7 Flavour changing neutral-current processes 455 The matrix element in the r. Using (12. 0) = + 2(1 − x)3 4(1 − x)2 ˜ ˜ ˜ ˜ S(xc .s. 0) − 2 S(x.148) between the kaon and antikaon states. When the other other contributions cannot be neglected.146).12.170). 0) − S(xc . the |λt |.t λi λ j M12 (i j) (12.174).180) . Although the double top-quark contribution is really the largest in Im M12 . j=u.175) The part proportional to Im λ2 in the above expression is given by the matrix t element of the effective four-quark interaction present in (12. we can write it as a sum of contributions proportional to different products of CKM matrix elements Im M12 = Im i. 0) = xc − ln xc + xt2 − 8xt + 4 3xt 2 ln xt − + O(xc ) 2 4(1 − xt ) 4(1 − xt ) and the µ-independent parameter BK is defined by BK = BK (µ)[αs (µ)]−2/9 [1 + O(αs )] (12. x) + S(0. 0) − S(xt .17 and by QCD corrections to it.c.174) is generated by the box diagram in Fig.h.176) with C(µlow ) given by (12. xt ) + S(0. we find Im M12 = − 2 G 2 MW F 2 Im(λ2 )C(µlow )BK (µlow )m K f K t 12π 2 + (contributions proportional to other products λi λ j ) (12. one obtains a result which is usually written in the following form Im M12 = − 2 G 2 MW F 2 BK m K f K Im(λ2 )S(xc )η1 c 12π 2 2 + Im(λt )S(xt )η2 + 2 Im(λt λc )S(xc . 12.178) where −3x 3 ln x x 3 − 11x 2 + 4x ˜ ˜ ˜ S(x) = S(x. xt ) = S(xc . xt )η3 (12. of (12.179) The quantity η2 which parametrizes the short-distance QCD effects in the double top quark contribution is directly related to the calculated expression for the coefficient C(µ). Similarly to (12.150) and (12.177)                  (12. Namely η2 = [αs (µc )] 9 2 αs (µb ) αs (µc ) 6/25 αs (µ0 ) αs (µb ) 6/23 + O(αs ) (12. because |λc | contributions are included and the unitarity of the CKM matrix is used.

it is technically much more involved.3 Check that the relation ρ = 1 follows from the SUV (2) symmetry of the Higgs potential after spontaneous symmetry breaking (like in the σ model). with help of an effective theory and renormalization group evolution of Wilson coefficients. N . due to a singularity in the phase space integration.181) ¯ 12.9 Calculate the box diagram for K0 –K0 mixing in the unitary gauge. n = 1. one gets the tree level relation: ρ≡ 2 MW 2 cos2 θW M Z = n 2 2 tn (tn + 1) − t3n vn 2 2 2 n t3n vn (12. .) + O(MW /E). . Problems 12. 12.8 Calculate the forward–backward and left–right asymmetries for the process e+ e− → Z 0 → f¯ f for s 1/2 ≈ M Z .2 Show that a convenient measure of the C P violation in the K0 –K0 mixing is the parameter J = Im(Vub Vcd Vud Vcb ). . i. . 12. . 12.e. . Z L . with vacuum expectation values vn .182) − (1 − |Vud | − |Vcd | − |Vcb | − |Vub | + |Vud | |Vcb | + |Vub | |Vcd |2 )2 12.4 Calculate the β-functions for the renormalization group evolution of the gauge couplings in the Standard Model. However. Show that the equivalence theorem also holds for decays h 0 → W − W + → 4 f and h 0 → G + G − → 4 f (for MW < Mh < 2MW ). G 0 . 12. .e. 3 2 12.) = A(G ± . Show that in the limit Mh MW the same result is obtained by calculating the decay width for h 0 → G + G − where G ± are the Goldstone bosons in the ’t Hooft–Feynman gauge. .6 Calculate the partial decay width for h 0 → W + W − . . . .1 Show that with N Higgs multiplets with weak isospin Tn2 = tn (tn +1). This is an exam± 0 ple of the ‘equivalence theorem’: A(WL . Show that it can be expressed in terms of four moduli of the elements of the CKM matrix: 4J 2 = 4|Vub |2 |Vcb |2 |Vud |2 |Vcd |2 2 2 2 2 2 2 2 (12. ¯ 12.5 Derive RGE for Yukawa couplings in the Standard Model. for vanishing G ± f f coupling).456 12 Standard electroweak theory The calculation of η1 and η3 is performed with the use of the same technique as η2 . even in the limit of zero fermion masses (i.7 Show that (h 0 → γ γ ) ∼ (g 2 /MW )α 2 Mh . Construct the effective lagrangian for this process.

although external with respect to the strong interaction gauge group. acquire important dynamical sense. representation of the gauge group and for N > 2 to a complex representation of the chiral group (see Appendix E). i. As discussed in Chapter 9. Moreover. the conservation of the U (1) current corresponds to baryon number conservation whereas the conservation of the UA (1) current is a problem (the so-called UA (1) problem): it can be seen that the spontaneous breakdown of the SU (N ) × SU (N ) implies the same for the UA (1) but there is no good candidate for the corresponding Goldstone boson in the particle spectrum. In the framework of the renormalization programme in QCD the finiteness of the matrix elements of † We take quarks as massless. It is clear from the preceding discussion that Green’s functions involving chiral or axial currents are of direct physical interest. Bell & Jackiw 1969).13 Chiral anomalies 13. i. real. they are gauge currents of the weak interaction gauge group. For massive quarks the chiral group is softly broken but.e. this is irrelevant for the discussion of anomalies.† Fermions belong to a vector-like. To be specific let us consider QCD. In this chapter we systematically discuss the fermion anomaly in (3 + 1)-dimensional quantum field theory (Adler 1969. We know from Chapters 9 and 10 that the Noether currents corresponding to the global flavour symmetry.e. 457 . apart from being invariant under gauge transformations in the colour space its lagrangian is also invariant under the global SU (N ) × SU (N ) × U (1) × UA (1) chiral group of transformations acting in the flavour space. Its existence can be traced back to the short-distance singularity structure of products of local operators.1 Triangle diagram and different renormalization conditions Introduction Anomalies have already been mentioned in this book on several occasions. The axial non-abelian currents couple to Goldstone bosons (pseudoscalar mesons) and the left-handed chiral currents couple to the intermediate vector bosons. as we shall see.

The basic structures to study are the triangle diagrams depicted in Fig.1. In particular the problem of chiral anomalies in more than four space-time dimensions and of gravitational anomalies as well as new techniques for their calculation . The triangle diagram is anomalous because it is impossible to add a polynomial so that all the classical symmetries are simultaneously preserved. in general. 13. Historically. As we know. as in the standard SU (2) × U (1) model of weak interactions. also provide a beautiful manifestation of the UV–IR interrelation in the theory: short-distance singularities manifest themselves in low energy theorems. They may require an additional renormalization. must be separately verified. Relations obtained in this way between various renormalized Green’s functions involving currents can then be summarized in operator language: certain currents have anomalous divergences. In the next section we will see explicitly that the anomaly constitutes a breakdown of gauge invariance. This freedom is used in the renormalization procedure when we get rid of all infinite pieces. Ward identities. This is an important constraint on any theory. Anomalies reflect the absence of a renormalization procedure such that Green’s functions respect all the symmetries of the lagrangian. being a generalization of Gauss’ theorem. However. In the following we calculate explicitly the triangle amplitude and study its Ward identities. the main tool for studying anomalies is the Ward identities in perturbation theory. To complete the introductory remarks it is worth mentioning that anomalies are also of basic importance for most theories going beyond the Standard Model. In other words an anomaly means a deviation of the renormalized theory from the canonical behaviour. We consider the U (1) and UA (1) currents of one-fermion free-field theory. In this way anomalies can be seen experimentally. the anomaly spoils the conservation of the currents coupled to gauge bosons and gauge theories with gauge fields coupled to non-conserved currents are inconsistent. the requirement of unitarity in each channel makes any one-loop amplitude well-defined up to a polynomial in the external momenta. All the necessary generalizations can be easily introduced at the end. All internal quantum numbers of fermions are suppressed in this calculation. Otherwise. these combinations must be anomaly-free. if gauge fields are coupled to linear combinations of vector and axial-vector flavour currents. Anomalies originate in the short-distance singularities of the free fermion propagator which generate renormalization ambiguities in the Green’s functions containing chiral currents.458 13 Chiral anomalies the T -ordered products of such local composite operators as chiral currents is an additional requirement which. in perturbation theory. It turns out that in four space-time dimensions the Green’s function which cannot be renormalized preserving all the classical symmetries of the lagrangian is the one containing two vector and one axial currents or three chiral currents.

t).1 Triangle diagrams 459 Fig.1 and the fact that the canonical equal5 5 5 time commutators [ j0 (x. at the classical level. jµ (y. [ j0 (x. In momentum space. Zumino. t). jµ . they read −( p + q)µ 5 µνδ ( p. For the time being we treat them as abelian currents of the one-fermion free-field theory which. Frampton & Kephart (1983). t)]. 13. q) (13. q) =0 (13. say.1) for m = 0 or.13. (y. Wu & Zee (1984)). and two vector currents jµ (x). t). q) = qν 5 µνδ ( p. q) = pδ 5 µνδ ( p. We shall come back to some e of these subjects later on.1.2) = qν 5 µνδ ( p. or of one 5 axial vector current jµ (x). t)] and [ j0 (x. q) = 2m 5 νδ ( p. have been vigorously studied (for example. Calculation of the triangle amplitude L We consider a vacuum matrix element of three chiral currents jµ (x). q) 5 µνδ ( p. AlvarezGaum´ & Witten (1983). for m = 0 −( p + q)µ and pδ 5 µνδ ( p.3) . for example. If we allow for a Dirac mass term in the lagrangian then the axial-vector current is partially conserved 5 L ∂ µ jµ (x) = −2∂ µ jµ (x) = 2mi ¯ γ5 Using the techniques of Section 10. t) vanish one can derive the standard Ward identities for the matrix elements L L 5 0|T jµ (z) jν (y) jδL (x)|0 and 0|T jµ (z) jν (y) jδ (x)|0 . jµ (y. q) =0 (13. are conserved.

We also keep in mind that. a contribution from the triangle diagrams shown in Fig. 13. p) (13. q) = Fµδν (q. anticommuting γµ γ5 = −γ5 γµ and reversing the order of factors Tr[γα γν γβ γδ γγ γµ γ5 ] = Tr[γµ γγ γδ γβ γν γα γ5 ] (13. p) (13. As we shall see. at the one-loop level. q) is the Fourier transform of 0|T P(z) jν (y) jδ (x)|0 with P = ¯ γ5 .8) We also notice that the sign of the mass factor is irrelevant because the only nonvanishing mass-dependent terms in the trace are proportional to m 2 . where the anomaly is actually placed.1)–(13. q) where p and q are momenta conjugate to x and y.1.1) or (13. as we shall see. q) (13.s.7) This follows from (13. q) + Fµδν (q. p ↔ q and δ ↔ ν. We shall first consider the diagram on the r. q) = 2Fµνδ ( p. of Fig. an extra factor (−ig) appears in each such vertex. The first observation worth making is the symmetry property Fµνδ ( p.3). The amplitude corresponding to this diagram reads Fµνδ ( p. Thus the result we will obtain is valid for the m = 0 case. q) = (−1) and d4 k i i i Tr γν γδ γ µ γ5 (2π)4 /+/ −m /−m /− p−m k q k k / 5 µνδ ( p. respectively. It depends on our choice of the renormalization constraints. too.460 13 Chiral anomalies 5 µνδ ( p. dictated by the physics of the problem. q) consistent with the . For clarity of the following discussion it is useful to first write down the most 5 general form of the amplitude µνδ ( p.2) and (13. the current matrix elements get. q) is superficially linearly divergent.4) = 5 d4 x d4 y d4 z exp[−i( px + qy + r z)] 0|T jµ (z) jν (y) jδ (x)|0 5 and νδ ( p. the anomaly is massindependent. By power counting the amplitude Fµνδ ( p.3). 5 Fourier transform of 0|T jµ (z) jν (y) jδ (x)|0 5 4 µνδ ( p.1 and from crossed diagrams with the external lines interchanged.5) by using the change of variables k = −k. q)(2π) δ(r is the + p + q) (13. if currents are coupled to gauge fields with some coupling constant g. It is impossible to renormalize this amplitude preserving simultaneously all three identities (13. In a quantized theory. 13.5) = Fµνδ ( p.6) We take here a non-zero fermion mass but.h. these contributions destroy the naive Ward identities (13.

10).10) Then we must have (13. Thus we insist on vector 5 current conservation for the renormalized vertex µνδ 5 The Pauli–Villars regularization of the amplitude µνδ ( p. Using again power counting arguments it is clear that only the functions A1 and A2 are divergent. that we insist on the vector current conservation qν for the renormalized vertex 5 µνδ ( p. q) = −B3 (q. q) pν εµδαβ p α q β + B2 ( p. q) pδ εµναβ p α q β + B4 ( p. for instance. whereas the functions Bi are finite.9) where As and Bs are scalar functions. = pδ 5 µνδ ( p. There are many ways to calculate anomalies. q)εµνδα p α + A2 ( p.11) determine the divergent amplitudes Ai in terms of finite Bi s. p).11) is independent. Ovrut (1983) and references therein). q)εµνδα q α + B1 ( p. q) = −A2 (q. p) implies that A1 ( p. in principle. q) = −A2 (q. However. p) follows so that only one of the equations (13. q)qδ εµναβ p α q β (13. p). If the regularization procedure satisfies the constraints which we want to impose as our renormalization constraints then in the limit of no regulator we obtain the anomalous Ward identity we are looking for. From Bose symmetry A1 ( p. in fact only logarithmically divergent.5). q) 5 µνδ . It reads 5 µνδ ( p. B2 ( p. q) =0 (13. Dimensional regularization in this case has some problems with ambiguities in the extension of γ5 to n dimensions (for a review of this matter see. Imagine. q) = A1 ( p.1 Triangle diagrams 461 requirements of parity and Lorentz invariance. Given this identity.13.11) A1 + p · q B1 + q 2 B2 = 0 A2 + p 2 B3 + p · q B4 = 0 and indeed the conditions (13. these functions can be expressed in terms of Bs when definite renormalization conditions are imposed. The terms pν ενδαβ p α q β and qµ ενδαβ p α q β are not linearly independent.5) by some regularization prescription and to calculate Ward identities for the 5 regularized µνδ . q) consists of considering it to be a function of the mass of the fermion circulating in the loop: . satisfying different renormalization constraints. q)qν εµδαβ p α q β + B3 ( p. by additional finite subtractions one can always transform it into another anomalous Ward identity. p). The requirement of Bose symmetry under interchange 5 5 of the two vector currents µνδ ( p. necessary for the functions A1 and A2 . q) = µδν (q. B1 ( p. Subtractions are. Let us follow this approach choosing to work with the Pauli–Villars regularization (Itzykson & Zuber 1980) of the amplitude (13. for instance. Let us choose as our renormalization conditions the relations (13. q) = −B4 (q. One is to regularize the amplitude (13.

M) (13.16) The last term vanishes for the same reason as above.13). M) is finite. m) − 5 µνδ ( p. q. m) is cancelled by µνδ ( p. q. m).12) 5 The regularized amplitude µνδ ( p. M) = 0 Indeed. because it satisfies our renormalization conditions fixing Ai in terms of Bi First we show that the regularized amplitude satisfies the ‘normal’ Ward identities −( p + q)µ 5 µνδ ( p. The first two terms are second rank pseudotensors depending on only one momentum variable and hence vanish. m. q. M) = 5 µνδ ( p. M). M). At the end one lets the mass M go to infinity and the final answer is finite with no additional subtractions. The last term gives 5 just the Ward identity (13. q. q. m) − 2M 5 νδ ( p. q) 13 Chiral anomalies 5 = µνδ ( p. This can easily be checked by expanding the integrand in the integrals over d4 k in powers of 1/k 2 : the linear 5 5 divergence of µνδ ( p. m. 4 M) dk 1 1 γν γδ γ5 − (m → M) Tr 4 (2π) /+/ −m /−m k q k d4 k 1 1 + 2i Tr γν γδ γ5 − (m → M) (2π)4 /−m /− p−m k k / d4 k 1 1 1 γν γδ 2mγ5 − (m → M) + 2i Tr 4 (2π) /+/ −m /−m /− p−m k q k k / (13. q. q.14) pδ 5 µνδ ( p. m. A regulated amplitude is defined as the difference between the given amplitude and the same amplitude taken at some other value M of the mass (M is the regulator) 5 µνδ ( p. m. The first term also vanishes because we can shift the momentum integration variable to k = k − p and . M) (13. using the relation p + / = (/ + / − m) − (/ + p − m) we can write down / q q k k / the following equation: −( p + q)µ = 2i 5 µνδ ( p. m. M) = 2m 5 νδ ( p. m. q. m.13) (13. q. M) = q ν 5 µνδ ( p.462 5 µνδ ( p. Using p = / (/ − m) − (/ − p − m) this becomes k k / pδ 5 µνδ ( p. q. q.15) All the integrals are finite. q. M) = 2i d4 k 1 1 Tr γν γµ γ5 − (m → M) 4 (2π) /+/ −m /− p−m k q k / d4 k 1 1 − 2i γν γµ γ5 − (m → M) Tr 4 (2π ) /+/ −m /−m k q k (13. Consider now p δ µνδ ( p. q. q. q. m.

q.19) (13. q. m) vertex can be calculated explicitly since the Feynman integral is convergent: 5νδ ( p.21) is to calculate explicitly the finite functions Bi starting with the unregularized amplitude (13. q. This shift of the integration variable is legitimate because the integral is finite.18) and the factor 2 in front of the integral in (13. m) 5 µνδ = 2m 5 νδ ( p.1 Triangle diagrams 463 again apply the same argument.11). q. In a quite analogous way we can show that qν 5 µνδ ( p. Another way to obtain the anomalous Ward identity (13.13). our regularization procedure is consistent with our renormalization requirements (13. q. . † Note the difference between a shift of the integration variable by a constant and a change of variables k → −k used in proving (13.13.17) takes account of the crossed diagram. M) = 0 Thus.21) The anomalous term is m-independent.10). The νδ ( p. q. m) − (i/2π 2 )ενδβα p β q α = qν =0 (13. Our last step is to take the limit M → ∞ in the Ward identity (13. q. M) = +(i/2π 2 )ενδβα pβ qα and the final form of the axial Ward identity consistent with our renormalization constraints reads −( p + q)µ pδ 5 µνδ 5 µνδ ( p.20) It is now straightforward to calculate the limit M→∞ lim 2M 5νδ ( p. m) (13.1).17) d4 k 1 1 1 4 (k + q)2 − m 2 k 2 − m 2 ( p − k)2 − m 2 (2π) where F( p.16) and integrating over d4 k one gets F( p. This we leave as an exercise for the reader. q. m) = 2i 1 1 d4 k 1 Tr γν γδ γ5 4 (2π) /+/ −m /−m /− p−m k q k k / = 2 × 4miενδβα pβ qα F( p. m) = − 1 16π 2 1 1−x dx 0 0 dy [q 2 y(1 − y) + p 2 (1 − x)x + 2 pq x y − m 2 ]−1 (13. q. The anomaly results from the non-vanishing of the regulator term in 5 (13.13) in that limit.7). If it were linearly divergent the translation of the integration variable might involve a non-vanishing surface term which would spoil the argument† (see Problem 13. m.5) and then to use renormalization constraints (13. Using the Feynman parametrization (B. m) = i (13.

that the latter is satisfied for a1 − a2 = i/2π 2 Since a1 ( p.22). q) = −a2 (q. At the level of the divergent L Feynman integral the bare amplitude µνδ ( p. of Fig. Adding a finite polynomial .21). p) from the Bose symmetry under interchange of the two vector currents discussed earlier. q. q) can also be calculated from the triangle on the r. q) formally satisfies the following relation L µνδ ( p. q) to distinguish it from the AVV amplitude considered earlier) under interchange of any pair of vertices. The divergent amplitudes A1 and A2 are again uniquely specified by the condition (13.23) (13. q. q. Finally let us consider the triangle on the l. with (13.22) as our renormalization constraint. of Fig. with an additional finite subtraction for the A1 and A2 a1 εµνδα p α + a2 εµνδα q α we see. m) (13.h.s.3). m) 5 µνδ ( p.22) as the renormalization condition one gets pδ qν 5 µνδ ( p.1. q) (which we denote 5 by ˜ µνδ ( p.25) One cannot satisfy simultaneously all three Ward identities (13.1 with the left-handed currents at each vertex and in the limit m = 0. our previous renormalization conditions are inappropriate since now for the Green’s function of three identical currents we 5 must require Bose symmetry of the renormalized µνδ ( p. we conclude that a1 = −a2 = i/4π 2 Thus. q) = 1 2 µνδ ( p. instead. m) (13. q.10) lead to the result (13.22). q) (13. q) is the three-vector current amplitude which has no anomaly. 13. q) − 1 2 5 µνδ ( p.464 13 Chiral anomalies Different renormalization constraints for the triangle amplitude The renormalization conditions (13. However.21). by comparing (13. It may happen that we must. insist on the relation −( p + q)µ 5 µνδ ( p. L Thus formally the anomaly of µνδ ( p.2) and (13.s. Indeed. m) = 2m 5 νδ ( p.24) = −(i/4π 2 )εµναβ p α q β = (i/4π 2 )εµδαβ p α q β (13.h. 13.26) where µνδ ( p.21) and (13. Whether we should impose them or not depends on the physics of the problem. We can get the right answer by applying a finite renormalization to the previous result (13. This will be illustrated in the next section.

q) = −(i/6π 2 )ενδαβ p α q β    5 p δ ˜ µνδ ( p. q) = (−i/2π 2 + a1 − a2 )ενδαβ p α q β    δ ˜5 α β p µνδ ( p. (13.32) respectively. to illustrate the meaning of arbitrariness in renormalization conditions for the triangle diagram. the same triangle amplitude leads to three different sets of anomalous Ward identities depending on the imposed renormalization conditions.1 Triangle diagrams 465 term.30) Thus.28) (13. Important comments In the next section we will discuss in more detail the anomaly problem in the context of specific physical theories. δ ↔ ν one gets the following equations −i/2π 2 + a1 − a2 = a2 a1 = −a2 Therefore a1 = −a2 = i/6π 2 and  5 −( p + q)µ ˜ µνδ ( p. where Dµ = ∂µ + ig Aµ . The latter must be chosen according to the physical requirements of the problem under consideration. q) = +(i/6π )εµδαβ p q (13.30) with (13.26). Fµν = ∂µ Aν − ∂ν Aµ .27) Insisting on symmetry under the interchange −( p + q) ↔ p. we consider two models given by the lagrangians / L = Z 2 ¯ iD and R L = Z 2 ¯ R i/ ∂ R L / + Z 2 ¯ L iD L − 1 Z 3 Fµν F µν 4 − 1 Z 3 Fµν F µν 4 (13.31) (13. q) = a1 εµδαβ p q (13. q) = −(i/6π 2 )εµναβ p α q β    ν ˜5 2 α β q µνδ ( p. Here.29) (13. q) = a2 εµναβ p q    ν ˜5 α β q µνδ ( p. µ ↔ δ and p ↔ q.23).13. to the amplitudes A1 and A2 we get (for m = 0)  5 −( p + q)µ ˜ µνδ ( p. L We recall that our present interest is in the amplitude µνδ ( p. q) so we combine (13.

35) can be verified using the Feynman rules for the operator insertion εµνρδ Fµν (x)Fρδ (x) = 4ε µνρδ ∂µ Aν (x)∂ρ Aδ (x) (13.33) and jµ (x) = Z 2 ¯ (x)γµ (x) (13. but Z 5 is not finite (we will come back to this point shortly). as we know from the previous subsection. This conclusion is no longer valid in the presence of the anomaly. at this level Z 5 = 1. The current is still 5 5 multiplicatively renormalized R jµ = Z 5 jµ . without additional counterterms. (13. Gauge invariance of the theory requires the vector current conservation. 13.1 we have proved that a conserved or partially conserved current satisfying normal Ward identities is not renormalized. which in terms of the renormalized fields are as follows: 5 jµ (x) = Z 2 ¯ (x)γµ γ5 (x) (13. for instance vector and axial-vector currents in the theory (13. is not coupled to any gauge field. Considering. However.36) (see Section 7. Thus for the triangle diagram involving 5 jµ and two vector currents we must impose the renormalization conditions (13. In Section 10.34) A comment is in order here on the renormalization of currents in presence of the axial anomaly. The first theory is just QED of a single fermion field. In operator language this corresponds to an anomaly in the divergence of the axial-vector current 5 ∂ µ jµ (x) = (g 2 /16π 2 )εµνρδ Fµν (x)Fρδ (x) (13.466 13 Chiral anomalies Fig.6).2. for example.21). the Green’s function 0|T ∂µ Aν (z) . conserved in the Noether sense. The vector current is coupled to the electromagnetic field and the axial-vector current.31).10) which lead to the anomalous Ward identity (13.35) Eq. at the one-loop level the current matrix elements are finite after imposing certain renormalization conditions. Thus. We define the bare currents. and the lagrangians are written in terms of renormalized quantities.

4 is logarithmically divergent. the lagrangian (13. Using (13.2 and 13.1 we conclude that Z 5 is not finite as the operator Fµν F µν does not have finite matrix elements: e.3.13. 13.32).3 and in the lowest non-vanishing order in g 2 the anomaly is given by the diagram in Fig. in order g 2 the Fourier transform (13. Closely following the ˜ procedure of Section 10. 13. 13. ∂ρ Aδ (z)Aα (y)Aβ (x)|0 in momentum space one gets the Feynman diagram shown in Fig.3) can be used to study the renormalization properties of the axial-vector current.g. The triangle diagram with gauge source currents at each vertex should be evaluated imposing symmetry under interchange of any pair of indices. removed from the external lines). For instance. as usual.2 (propagators are. Therefore.4.4. the diagram in Fig.1 Triangle diagrams 467 Fig.30) we now get † Now with (−ig) at each vector vertex included. The anomalous Ward identity for 5 (derive ¯ it using the path integral technique of Sections 10.2 and (13.35) we can calculate perturbatively other anomalous Ward identities involving axial-vector 5 current jµ (x). 13. the derivative of the Green’s function 5µ = ¯ 0|T (x) j 5µ (0) ¯ (y)|0 is anomalous because the latter gets a contribution from the diagram in Fig. Using relations (13. Fig.21)† (with m = 0) of the divergence 5 ∂ µ 0|T jµ jν jδ |0 is given by the Feynman diagram in Fig. . 13. Our second example. 13.11) of Ward identities. is a theory in which only the left-handed chiral current couples to the gauge fields.26) and (13.35) follows if we recall the general form (10. 13.

35) for the bare current we have used the result obtained in the lowest order in g 2 in perturbation theory. and only if. particularly in supersymmetric theories. λb .37) Because of the anomaly the gauge source current is no longer conserved and the theory is not gauge invariant. 13. i. λb . λc may transform non-trivially under different internal 1 2 3 symmetry groups and be block-diagonal in the remaining internal symmetry spaces. Let the currents in the vertices have coupling matrices λa . the scale invariance anomaly discussed in Chapter 7 does obtain corrections because renormalization violates scale invariance. Equivalently one can say that the gauge field has been treated as a classical background field. like (13. . λc . For the vector abelian theory considered here this has been proved to all orders in perturbation theory (Adler & Bardeen 1969). According to the Adler–Bardeen theorem generalized to the non-abelian case a theory is completely free of anomalies if. 1 2 3 2 where the trace corresponds to summing over different fermions in the loop and the anticommutator is due to the symmetrization over the two diagrams in Fig.) The Adler–Bardeen theorem also holds in non-abelian gauge theory but it may be regularization-scheme-dependent (Bardeen 1972). does not interfere with chiral symmetry. (13. The proof is technical but its main point can be summarized as follows: the necessity of renormalization of the higher order corrections in g 2 . of photon insertions in the triangle diagram. In gauge theories with chiral gauge symmetry the anomalies must be absent for a consistent theory. will be discussed later. λc }]. Writing the operator equation (13. Our last remark in this section is concerned with generalizing the triangle diagram calculation to the case in which several species of fermion fields which have some internal degrees of freedom are present. (On the contrary.5 The necessary modifications of operator equations.35). so the chiral anomaly should arise only from the lowest order fermion loops. One believes that the anomalies are not modified by higher order corrections. all the triangle diagram anomalies are absent.31). The currents may be coupled to gauge fields or 1 2 3 not and the matrices λa . i. For the triangle diagram we get then an extra factor 1 Tr[λa {λb .468 13 Chiral anomalies the anomaly in the chiral current L ˜ ∂ µ jµ (x) = −(g 2 /48π 2 )Fµν F µν (13.e.35) is exact to any order in g 2 for the bare current in the regularized theory or for the renormalized operators (see also the next section) with g and Fµν being the renormalized quantities defined by the lagrangian (13.e.

13.31). it turns out that chiral invariance remains a symmetry of the quantum theory based on the lagrangian (13. in quantum theory if we want to preserve vector gauge invariance an anomaly appears in the divergence of the axial-vector current. However. the triangle diagram amplitude turns out to be finite even in the limit of no cut-off but the higher order corrections induce infinities in the axial-current matrix elements. Defining 5 ˜ ˜ (g 2 /8π 2 )[Fµν (x) F µν (x)]R = (g 2 /8π 2 )Fµν (x) F µν (x) + (Z 5 − 1)∂ µ jµ (x) (13. 13. Therefore due to the anomaly the axial-vector current is renormalized: R 5 jµ 5 = Z 5 jµ and we must re-express (13. 13.35) in terms of the renormalized operators 5 ˜ ∂ µ R jµ (x) = (g 2 /8π 2 )[Fµν (x) F µν (x)]R (13.39) However.38) we may write (13. If we introduce a cut-off which preserves photon gauge invariance then to any order in g 2 the divergence of the unrenormalized axial-vector current is given by (13.2 Physical consequences of the chiral anomalies 469 Fig.2 Some physical consequences of the chiral anomalies Chiral invariance in spinor electrodynamics The lagrangian (13.5.31) and it is possible to construct the corresponding charge Q 5 .1).31) of spinor electrodynamics is invariant both under U (1) and axial UA (1) transformations. it does not commute with the photon field and does not have the correct equal-time commutation relations with the renormalized spinor field. Not all of these can be absorbed in the renormalization constants present in the lagrangian (13. Nevertheless.35) in terms of the renormalized operators. this renormalized axial-vector current does not generate chiral transformations. The axial charge is not conserved. Following Adler (1970) we first define a new bare . As discussed in the previous section. due to the presence of a dimension four operator ˜ (Fµν F µν )R in the divergence of the current (see Section 10.35).

470 13 Chiral anomalies current: symmetry current S 5 ˜ j5µ (x) = jµ (x) − (g 2 /4π 2 )Aν (x) Fµν (x) (13. Under the gauge transformation δ (x) = −ig (x) (x) δ Aµ (x) = ∂µ (x) the symmetry current transforms as follows: S ˜ ˜ δ j5µ = (g 2 /4π 2 )∂ ν (x) Fµν (x) = (g 2 /4π 2 )∂ ν ( (x) Fµν (x)) (13. calculated by use of (13. its commutator with the canonical commutation relations.21) and (13.22). from (13. is [Q 5 (t). is conserved S ∂ µ j5µ = 0 (13.3) where not all finite gauge transformations can be reached by iterating infinitesimal transformations.42) (x).40) which. t) (13. But the associate charge Q 5 (t) = S d3 x j50 (x. This current is a S finite operator (Bardeen 1974) and does not need renormalization. The situation is different in non-abelian gauge theories (see Section 8.46) Gauge invariance of Q 5 in spinor electrodynamics follows from the vanishing ˜ of (x) F0k (x) at spatial infinity. t)] = −γ5 (x) it commutes with the photon field and it is gauge-invariant. Let us check the last property.44) (13.41) time-independent. We conclude that in spite of the presence of the axial anomaly chiral symmetry remains a good symmetry in massless spinor electrodynamics. (x. compare (13. .45) Therefore δ Q 5 (t) = S d3 x δ j50 (x) ∼ ˜ d3 x ∂ ν ( (x) F0ν (x)) = ˜ d3 x ∂ k ( (x) F0k (x)) = 0 (13.43) is from (13.41) Its existence is directly related to our discussion of different renormalization conditions of the triangle diagram.35). The current j5µ is conserved but it is explicitly gauge-dependent due to the presence of the gauge field in its definition and therefore is not an observable operator.

Thus.47) we have used (2.2 Physical consequences of the chiral anomalies 471 γ π 0 → 2γ The relevance of the chiral anomaly to the decay π 0 → 2γ relies on our identification of pions as the Goldstone bosons of the spontaneously broken chiral symmetry of QCD (see Chapter 9). the axial-vector flavour currents of the SU (2)× SU (2) symmetric QCD lagrangian (9.49) plus terms of at least third order in ki from the second diagram: the factor k1 k2 comes from the gauge-invariant coupling (through the field strength tensor Fµν ) of external photons and an extra factor (k1 +k2 ) comes from the anomalous divergence term. we use the PCAC (partial conservation of the axial-vector current) approximation relating physical quantities to the results obtained in the exact symmetry limit p 2 = m 2 = 0. of (13. the Pauli matrix σ 3 is the neutral axial-vector coupling matrix and Q is the diagonal quark electric charge 5 matrix. 13.47). of (13. (2. Next. k2 ) α = (i/2π 2 ) Tr[ 1 σ 3 Q 2 ]ενδβα k1 k2 2 β (13. for µνδ defined by (13.48) where O(k1 . To calculate the π 0 → 2γ decay rate we 5a consider the amplitude ( jµ (x) is the renormalized axial isospin current) 53 d4 x exp(−iq x) γ1 (k1 )γ2 (k2 )|T S jµ (x)|0 = d4 x d4 y d4 z exp(+ik1 y) (13.6. The effect of the last diagram is only the coupling-constant renormalization.47) is given by the results of the previous section with the triangle diagram specified as shown in Fig.h. k2 ) has no pole in p 2 . With u and d quarks of three colours we get Tr[ 1 σ 3 Q 2 ] = 3 × 1 ( 4 − 1 )e2 = 1 e2 2 2 9 9 2 (13. see (9.188).1) acquire dynamical sense through their coupling to pions.196)) q 2 → p2 lim 53 d4 x exp(−iq x) γ1 (k1 )γ2 (k2 )|T S jµ (x)|0 = f π ( pµ / p 2 ) γ1 γ2 |S|π 0 + O(k1 .h.50) . We now take the divergence of (13.s.65) 5a 0| jµ (x)|π b = i f π pµ δ ab exp(−i px) where pµ is the pion four-momentum. The fermion lines are quark lines.194).68) and the equations of motion for the gauge fields Aµ (x). (2. k2 ) (13.4) we get the following: pµ 5 µνδ (k1 .195) and (2.s. (2. In this limit there is a pion pole contribution π to the l.47) 53 ν δ × exp(+ik2 z) exp(−iq x) 0|T jµ (x) jν (y) jδ (z)|0 ε1 (k1 )ε2 (k2 ) In (13. for example.13. In this case. The divergence of the r.47) (see.

Q}]Fµν F µν 2 2 (13. k2 . 13. respectively. 13. k2 ) in (13.48) γ1 γ2 |S|π 0 = 1 e2 β iε k k α εν εδ (2π )4 ∂( p − k1 − k2 ) + O(k1 . Terms O(k1 .6.7.51) Our calculation of the divergence of (13.472 13 Chiral anomalies Fig. from (13.47) and (13. of course. p) in (13.47) is.48) and O(k1 k2 . equivalent to calculating 5a perturbatively the matrix element γ1 γ2 |T S∂ µ jµ (x)|0 with the divergence given by 5a ˜ ∂ µ jµ (x) = (e2 /8π 2 ) 1 Tr[ 1 σ a {Q. Fig. From Lorentz and parity invariance α ν δ γ1 γ2 |S|π 0 = f ( p 2 )iενδβα k1 k2 ε1 ε2 (2π )4 δ( p − k1 − k2 ) β (13.51) are at least second and third order in momenta.7.53) . Therefore. 13.52) The contributing diagrams are those shown in Fig. p) 2 νδβα 1 2 1 2 f π 4π (13.

13. We must have D µ jµ = 0 to a maintain gauge invariance. The experimental value is Exp = (7. Secondly. In the absence of the anomaly one would get f (0) = 0 and PCAC could not be reconciled with the π 0 → 2γ decay rate. the agreement with the experimental value relies on the existence of three colour states of quarks.54) ε1 .5 eV). UA (1) problem Global symmetries of the QCD lagrangian with massless quarks are (see Section 9. Comparing (13.7 ± 0.ε2 d 3 k 1 d3 k 2 β α ν δ | f (0)ενδβα k1 k2 ε1 ε2 |2 (2π )4 δ( p − k1 − k2 ) 2π 6 4k10 k20 (13.8. We study now the anomaly of the UA (1) current. We make two important observations. q) of the Fourier transform of the matrix element . 13.51) and the last equation we conclude that f (0) = and the rate for π 0 → 2γ is = = 1 2m π 2 1 fπ e2 4π 2 = α fπ π (13. Chiral anomaly for the axial U (1) current in QCD.55) is entirely given by the axial anomaly.10). For the triangle diagram 5 a we effectively get the condition ∂ ν 0|T jµ jν jδb |0 = 0 and therefore we must impose the renormalization constraints (13. by analogy with (13. The UA (1) current is a singlet under the colour group and the a relevant triangle diagram is shown in Fig.2 Physical consequences of the chiral anomalies 473 Fig.63 eV 2 64π 3 f π (Bose statistics implies integration over a 2π solid angle only). A derivation of the equation D µ jµ = 0 in a quantized gauge-invariant theory is given in the next section. 13.1): chiral SU (n) × SU (n) invariance and invariance under abelian vector U (1) and axial-vector UA (1) transformation. 5 the divergence −( p + q) µνδ ( p. Thus.8.55) α m3 π = 7. The result (13.21).

These properties offer a mechanism of breaking of the UA (1) which does not imply the existence of the massless Goldstone boson coupled to the gauge-invariant UA (1) current.57) and of field configurations with a non-zero topological charge (see Section 8. The operator equation (13. this contribution is not independent of the triangle anomaly: given the latter. there is no good candidate for the UA (1) Goldstone boson in the particle spectrum. However. square diagrams also contribute to the anomaly for the divergence of the axial current. for instance. the anomalous divergence of the UA (1) current involves not only terms quadratic in gauge fields but also terms of the third order in gauge fields 5 ∂ µ jµ (x) = −(n/16π 2 ) Tr[4∂µ Aν ∂ρ Aσ + 4∂µ Aν [Aρ .35) can also be easily generalized to the present case if we remember that UA (1) current is a colour singlet. due to the self-coupling of the non-abelian gauge fields.111). µν 2 A new feature of (13. Aσ ]]ε µνρσ = −(n/4π 2 )εµνρσ ∂µ Tr[Aν ∂ρ Aσ + 2 Aν Aρ Aσ ] 3 (13. For an explicit discussion of the square diagrams in the spirit of the previous section see. T a = 1 λa . unlike the case of the spontaneously broken SU (n) × SU (n). As in the first subsection of this section we observe that the very existence of the anomaly does not destroy the UA (1) symmetry since the conserved current S j5µ = j5µ − 2n K µ .58) Thus.474 5 a 0|T jµ jν jδb |0 reads 13 Chiral anomalies −( p + q)µ 5 µνδ ( p.57) of the axial-vector current is crucial for resolving the so-called UA (1) problem in QCD.57) where Tr is in the internal quantum number space and G µν = igG a T a . All the anomalous Ward identities follow from the operator equation (13. it is determined by the gauge invariance of the axial-vector current. The anomaly (13. As with the SU (n) × SU (n) it does not seem to be the symmetry of the hadron spectra so UA (1) must be broken (U (1) manifests itself in baryon number conservation).56) is 5 ˜ ∂ µ jµ (x) = −(n/8π 2 ) Tr[G µν G µν ] (13.56) where n is the number of quark flavours. see (1.3). The only colour-invariant operator which for the triangle diagram gives the result (13. As we know from Chapter 9 the massless QCD lagrangian has the chiral (U (1) ∓ UA (1)) symmetry. q) 1 = (ig 2 /2π 2 )n Tr[ 1 λa 2 λb ]ενδβα p α q β 2 (13. The resolution of the problem is in the existence of the anomaly (13.57) is that. However. ˜ ∂ µ K µ = −(1/16π 2 ) Tr[G µν G µν ] and the conserved charge Q5 = S d3 x j50 .57). Aviv & Zee (1972).

however. see also Jackiw 1984 and our discussion in the next section). It will also be convenient to remember that by construction the fermion electric charge operator is given by Q = T 3 + Y . We can now systematically list possible anomalous triangle diagram contributions to the divergences of the gauge source a currents.13. However. i. Q 5 is not gauge-invariant: it is not gauge-invariant under large gauge transformations G N with the winding number N . Take first the SU (2) current jLµ : we can have the triangle diagrams a shown in Fig.2) −1 G N Q 5 G N = Q 5 − 2n N From Section 8. UA (1) symmetry is spontaneously broken because into the | + 2n the vacuum is not UA (1)-invariant. This does not limit the generality of our discussion: a theory is anomaly-free if. Anomaly cancellation in the SU (2) × U (1) electroweak theory We discuss this problem at the level of the triangle diagrams relegating to the next section the construction of the full anomalous divergence of a non-abelian current coupled to gauge fields.65) and (8.67) and Problem 8. For consistency of a quantum gauge theory the gauge source currents must be covariantly conserved. it can be seen from the chiral Ward identities that the spontaneous breakdown occurring because of the axial-vector anomaly and the existence of field configurations with non-zero topological charge (hence the set of the -vacua) is consistent with the assumption 5 of no Goldstone boson coupled to the gauge-invariant current jµ . the triangle diagram anomalies are absent. We have (see Section 8.and right-handed currents but with different U (1) charges Y .3 we also remember that GN | Thus we get G N exp(i or exp(i Q 5 )| =| + 2n The conclusion is that the UA (1) rotation by an angle changes the | vacuum vacuum. and only if. 13. In the electroweak theory the SU (2) gauge bosons couple only to the left-handed currents and the U (1) gauge boson couples to both the left.2 Physical consequences of the chiral anomalies 475 can be defined.58) . Thus.9 with the jLµ in one of the vertices and the corresponding Q 5 )| = exp[−iN ( + 2n )] exp(i Q 5 )| = exp(−iN )| (8. YR = Q and YL = (Q − T 3 ). anomaly-free (Gross & Jackiw 1972. Now.e.3 and in particular (8. Different -vacua are degenerate states since for massless fermions H commutes with Q 5 .

9(b) does not contribute because [Y. T b }] for Fig. T c }] Tr[T a {Y.9(b) for Fig. Y The anomalous contribution to the divergence of the U (1) current jµ = i i i i ¯ ¯ R γµ Y R ). T a }] = δ ab Tr[Q(T a )2 ] = 1 δ ab 2 i Q i (13. Y }] The first two traces have already been calculated. In this case all possible triangle diagrams are proportional to one of the following factors: 1 2 Tr[Y {Y.9(a) for Fig.26)). with left.60) .9. 13. T a }]. 13. 13. 13. T c }] = 0 for the group SU (2). 13. where the sum is over all fermions in the theory. Y symmetrized diagrams ( jµL is the left-handed part of the U (1) current). We immediately see that Fig. which contribute with opposite signs because of the (1 ± γ5 ) factors (see (13. We recall that the U (1) coupling is not vector-like because of different Y quantum number assignments to the left.59) where the sum is taken over the charges of all fermions in the loop. If we express Y in terms of Q the remaining contributions are 1 2 Tr[T a {Q.476 13 Chiral anomalies Fig.and right-handed fermions. T b }] = 1 2 Tr[Q{T b . and Tr[T a {T b .and right-handed fields.9(c) (T a = σ a are SU (2) generators). The last one appears in two types of loop. T b }] and 1 2 Tr[Y {Y. i ( L γµ Y L + can be studied in a similar way. T a ] = 0 and Tr T a = 0. 1 2 Tr[Y {T a . Y }] Tr[T a {Y. We calculate 1 2 Tr[Y {Y. It is clear from our discussion in the previous section that the eventual anomalous contribution of these diagrams after summing over all fermions in the loops is proportional to 1 2 1 2 1 2 Tr[T a {T b . Y }] = Tr Y 3 = Tr(Q − T3 )3 = Tr[Q 3 − 3Q 2 T3 + 3Q(T3 )2 − (T3 )3 ] (13.

h.35). for instance. in the operator notation we have.62) does not vanish due to some non-perturbative effects like tunnelling. Thus (13. Notice that the condition of vanishing divergence of two currents can always be imposed on triangle diagrams. Insisting on chiral gauge current conservation we are formally back to the renormalization conditions (13.2 Physical consequences of the chiral anomalies 477 Since Q is the same for fermions with either of the chiralities.61) is the only condition for the anomaly cancellation. U (1) vector current. of (13. acquires an anomaly. Since U (1) current is an SU (2) singlet. ∂ µ jµ = −(1/16π 2 ) Tr[ε µνρσ G µν G ρσ ] (13. with instantons as the dominant field configurations. An interesting µν possibility is that the vacuum expectation value of the r. The remaining terms give again Qi = 0 i (13.13. the Q 3 term is cancelled separately for each fermion when the sum over the two types of loop is taken. We conclude that the SU (2) × U (1) theory is anomaly-free if the sum of charges of all fermions in the theory vanishes. One would then expect ‘topological’ baryon number non-conservation. Restrictions on the fermion representations appear when we consider triangle diagrams with three gauge currents.61) since Tr[Q 2 T 3 ] = Tr[(T3 + Y )2 T3 ] = Tr[(T3 )3 + 2Y (T3 )2 + Y 2 T3 ] = Tr Y = Tr Q. analogously to (13. This is clear if we consider the triangle diagram with the baryon current and two gauge source currents. fractionally charged but coming in three colours. . or monopoles. The quarks and leptons discovered so far strikingly satisfy this condition: to each SU (2) doublet of leptons corresponds a doublet of quarks.10) and the anomaly is placed in the vector current.62) where G µν = igG a T a is the SU (2) field strength tensor.s. An interesting point to mention in the context of the SU (2) ×U (1) theory is that once we insist on conservation of the chiral source currents for the electroweak gauge fields (one needs suitable sets of fermion multiplets to ensure this property) the baryon number current.

. This then gives a limitation on the allowed fermion representations. The other case is when the group is not safe. We consider a theory of massless fermions coupled to non-abelian gauge fields. These are called ‘safe’ groups. As we know the triangle diagram anomaly is given by a multiple of Dabc = 1 2 Tr[T a {T b .A T α αµ = ¯ iγ µ αµ αµ − jV AV − jA AA αµ αµ (13. It may be that.478 13 Chiral anomalies Anomaly-free models We conclude with more general remarks. the theory is vector-like (Appendix E). They include SU (2). for an anomaly-free theory the quantity Dabc must vanish. like SU (N ) with N > 2 or SU (2)×U (1). the coupling constant g is included in the definition of Aµ . T c }] (13. but there exist some ‘safe’ representations for which Dabc vanishes.and right-handed couplings. i.63) where T a .A Aµ = iAV. Depending on the physical problem at hand we may need to use vector and axial-vector couplings or left. We take all three currents to be gauge source currents. T b and T c are generators of the group of transformations of currents attached to the three vertices. 13. For the purpose of the present discussion it is convenient to treat the gauge fields as classical background fields.e. This happens when fermions with either of the chiralities transform according to the same representation of the gauge group.3 Anomalies and the path integral Introduction For a systematic discussion of chiral anomalies it is convenient to use the generating functional and the path integral technique. One can distinguish three cases of anomaly-free models (Georgi & Glashow 1972). Dabc = 0. The transformation properties of the fermion and the gauge fields under infinitesimal † Since in this section fields Aµ are treated as background fields.64) and the T α s are the generators of the gauge group in fermion space. In the first case the gauge-invariant lagrangian density reads† L = ¯ iγ µ (∂µ + AV + γ5 AA ) µ µ where V. for all representations of the group. Two cases may be distinguished here. all orthogonal groups except S O(6) ≈ SU (4) and all symplectic groups. If this is not the case. The right-handed fermion loop and the left-handed fermion loop anomalies may cancel.

too.64) is also invariant under global U (1) and UA (1) transformations. only two types of Feynman diagram appear in our theory: diagrams in which the external fields couple to a free spinor line and diagrams in which the external fields couple .66) respectively.125)). At the L. In addition the same fermions may couple to other gauge fields.65) = (1 − γ5 δ δ A A A) A. that in a non-abelian theory a pure axial-vector interaction with Dirac fermions would not be gauge-invariant. It is often more convenient to use the left. However. A AV = −[ µ A A = ∂µ µ AA ] µ AV ] µ −[    (13.3).13. where as usual = i α T α .R where α α jL.67) can be regarded as dynamical gauge fields or as auxiliary background fields.3 Anomalies and the path integral 479 vector and axial-vector gauge transformations are = (1 − δ δ V V V) V A V = ∂µ µ AA = −[ µ −[ AA ] µ V. the physical interpretation of the non-abelian structure in the lagrangians (13. We see. A V ] = Dµ µ     A. The lagrangian (13.and right-handed fields and then  L = ¯ L iγ µ (∂µ + iAα T α ) L + ¯ R iγ µ (∂µ + iAα T α ) R   Lµ Rµ (13. in particular.R iγ µ ∂µ L.R Invariance under gauge transformations on the chiral fields requires standard transformation properties for the gauge fields Aµ (see Section 1. Depending on the physical problem considered.     V V. the gauge fields in the lagrangian (13.64) and (13. Since. no gauge bosons are coupled to the corresponding abelian currents. the gauge fields are classical background fields.R − jµ Aµ − jµ Aµ =  Lα Rα  L. by assumption.67) also depends on the problem.64) or (13.R = 0 (see (1.Rµ = ¯ L.R γµ TL. Of course. for instance. corresponding to a gauge group acting in a different space. the case with the flavour symmetry currents in QCD.R L.R V A classical level the U (1) currents are conserved: ∂ µ jµ = ∂ µ jµ = 0 and the αµ non-abelian currents are covariantly conserved: Dµ jL. This is.67) Lα Rα ¯ L.    (13. introduced to write down the generating functional for the anomalous Ward identities involving currents which in the framework of the considered theory are not coupled to any dynamical gauge fields.

70) by taking the derivative ∂ ∂ Aα (y) ν ∂ ∂ Aδ (x) β . Eq.1 to derive Ward identities we take the functional exp{iZ [Aµ . The answer has been given by Fujikawa (1980) who noticed that the anomaly is due to the non-invariance of the fermionic path integral measure under the transformation (13.64) and specify AA = 0. however. ¯ (x) → ¯ (x) exp[−iγ5 (x)] (13. µ Thus we consider the axial abelian anomaly in a QCD-like theory. We use the lagrangian (13. η]} = ¯ D D ¯ exp i µ d4 x (Lkin − jα Aα + η + ¯ η) ¯ µ (13. It reads 0 = δ Z [A. Abelian anomaly Let us first consider the generating functional for Ward identities involving the U (1) axial-vector current. η]/δ | ¯ 1 = exp{iZ [A.9). η.68) and use its independence of the choice of the integration variables changed according to the transformation (x) → exp[−iγ5 (x)] (x). are not altered). For instance ¯ ¯ µ β 5 α the anomalous Ward identity for the matrix element 0|T jµ (z) jν (y) jδ (x)|0 is obtained from (13.57) in particular. η]} ¯ × exp i =0 d4 z D D¯ + ¯ η) d4 x ( ¯ i/ ∂ µ − jα Aα + η ¯ µ 5 ˜ ¯ × {−(n/8π 2 ) Tr[G µν (z)G µν (z)] − ∂ µ jµ (z) + η(−iγ5 ) + (−i ¯ γ5 )η} (13.1. has no impact on the generality of the discussion of chiral anomalies due to the Adler–Bardeen non-renormalization theorem quoted in Section 13.70) generates the desired Ward identities when we differentiate it functionally with respect to the Aα . η and η and set η = η = 0. (13.9) which generates µ 5 Ward identities corresponding to ∂ µ jµ = 0 and is wrong in view of our perturbative calculations and of (13.70) One may ask where the anomalous term comes from in the path integral formalism. Applying the general formalism developed in Section 10.4. Naively we then get (10. This.69) (the gauge fields Aα . η. We shall come back to this point in Section 13. Using the latter we can anticipate the correct modification of (10.480 13 Chiral anomalies to a spinor loop. η.69).

R −[ L.66).71)–(13.72) δ δ = (iT α ) δ Aµ δ Aα µ D ¯ j µ (x) exp i / d4 x ¯ iD ˆ ≡ J µ (13.and right-handed fields separately form a group and we have δ AL.R . We again consider the functional exp{iZ [Aµ ]} = D D ¯ exp i / d4 x ¯ iD (13. Non-abelian anomaly and gauge invariance Our next task is to study the generating functional for the anomalous Ward identities involving non-abelian currents.73).73) and 1 δ Z [Aµ ] = δ Aµ (x) exp{iZ [Aµ ]} D ˆ and jµ (x) is the current coupled to the field Aµ (x) whereas J µ (x) is defined by (13. we study the gauge dependence of the functional Z [Aµ ].R (13.71) corresponding to the lagrangian (13.73) and that for δ Aµ depend.67). AL. First. The stands for Dirac or chiral fermion fields and Aµ for vector and axial-vector gauge fields or left.and right-handed gauge fields. if we use chiral fields. the gauge transformations on the left.56). other fields are suppressed. On the other hand.65) and (13.13. The explicit formulae (13. then Aµ (x) in (13.64) or (13. The gauge transformation of the fermion fields is again just a change of integration variables but under the accompanying transformation of the gauge fields Aµ (x) → Aµ (x) + δ Aµ (x) the functional Z [Aµ ] changes as follows Z [Aµ ] → Z [Aµ ] − 2 where d4 x Tr δ Z [Aµ ] δ Aµ (x) δ Aµ (x) (13.72) includes both AV and µ AA which transform under vector or axial-vector gauge transformation according µ to (13. Differentiating with respect to η and η and setting η = η = 0 one gets the anomalous Ward identity for ¯ ¯ 5 ¯ the matrix element 0|T jµ |0 .3 Anomalies and the path integral 481 and setting η = η = 0.R = ∂µ µ L. One gets ¯ 5 α α ˜ ∂ µ 0|T jµ (z) jν (y) jδ (x)|0 = −(n/8π 2 ) 0|T Tr[G µρ G µρ ] jν jδ |0 β β and in momentum space this is just equivalent to (13. of course. respectively.R ] = Dµ µ L. If we work with vector and axial-vector gauge fields.74) . on whether we use vector and axial-vector fields or chiral fields.

applied to Z [Aµ ]. X L. µ δ δ AA (x) µ (13.482 13 Chiral anomalies In each case we may rewrite (13.80) for chiral gauge transformations on the chiral gauge fields (all commutators are for matrices T α and not for fields which are c-numbers).72) and (13.75) we get δ Z =2 or δ Z [Aµ ]/δ where X V (x) = ∂µ δ δ AV (x) µ + AV (x). µ δ δ AA (x) µ + AA (x).R (x). µ δ δ AL.R (x) = ∂µ δ δ AL.79) for the axial-vector gauge transformations on the vector and axial-vector fields. µ δ δ AV (x) µ (13.R (x) µ = Dµ δ δ AL. µ δ δ AV (x) µ + AA (x). X A (y)] = c X A δ(x − y) (13.R δ(x − y) which.R (Y )] = cαβγ X L.77) (where X = iX α T α ) for the vector gauge transformations on the vector and axialvector fields.78) α (x)| (x)=0 d4 x Tr[X (x)Z [Aµ ] (x)] (13.R (x) µ + AL. † It is convenient to use a test functional.R (x). . X A (Y )] = cαβγ X V δ(x − y)     γ β α [X L. are equivalent to the relation δ 1 δ 2 Z [Aµ ] − δ 2 δ 1 Z [Aµ ] = δ[ 1.75) which is easy to verify. By explicit calculation† one verifies that the X s satisfy the gauge group commutation relations  γ β α [X V (x).73) using the following relation ˆ d4 x Tr[ J µ (x)Dµ (x)] = − ˆ d4 x Tr[D µ Jµ (x) (x)] (13.76) = −X α (x)Z [Aµ ] (13.82) expressing the group property of the gauge transformations. With the help of (13.R (x) µ (13.81) γ β α   [X A (x). and X L. 2] Z [Aµ ] (13. X V (y)] = cαβγ X V δ(x − y)      β α αβγ γ  [X V (x). X A (x) = ∂µ δ δ AA (x) µ + AV (x).

where J.88) which are (at α α the triangle diagram level D µ = ∂ µ ) ±(1/12π 2 )ε µνρσ Tr[T α ∂µ AL.R D µ jµ = 0 (13.89) . from the triangle diagram calculation in Section 13. for invariance of our quantum theory under one of the previously defined gauge transformations we must have δ Z [Aµ ] = 0 ⇒ X (x)Z [Aµ ] = 0 (13. by the absence of ¯ the functional integration over D Aµ and of the gauge field kinetic terms which are gauge-invariant.84) obtained in Section 1.3 on the classical level.76).90) (13.3 Anomalies and the path integral 483 Eqs.86) and the G A (x) in (13.1 we know partial contributions to the G L.R (x) in (13.86) Similarly.85) or equivalently α −X L. To study gauge invariance of the complete quantum theory it is sufficient to study the gauge invariance of the Z [Aµ ].R D µ jµ (x) = G L. η are the gauge and fermion field sources.88) Indeed. In a theory with only vector or only chiral gauge fields. The non-abelian source currents for gauge fields must be covariantly conserved at any order of perturbation theory for gauge invariance to be maintained in quantum theory.R ] ν σ and −(1/2π 2 )ε µνρσ Tr[T α (∂µ AV ∂ρ AV + 1 ∂µ AA ∂ρ AA )] ν σ ν σ 3 (13.82) have several implications.77) and (13.13. (13. Firstly.78)–(13. η] of the complete ¯ theory.83) where the X is one of the X s given by (13.80) depending on the considered gauge transformation. jµ ] = G A (x) V (13.87) or α −X A (x)Z [Aµ ] = G α (x) A (13. With dynamical gauge fields the generating functional Z [Aµ ] differs from the Z [J. using (13.81) or (13.R ∂ρ AL.R (x) (13. The anomaly means that the variation of Z [Aµ ] under gauge transformation does not vanish L. η.R (13. if we work with vector and axial-vector currents and insist on the vector V current conservation D µ jµ (x) = 0 then the divergence of the axial current is anomalous and A A A D µ jµ = ∂ µ jµ + [Aµ . (13.R (x)Z [Aµ ] = G α (x) L.73) we recover the equation V D µ jµ = 0 or L. η.

Observe that as long as for the anomaly the minimal form depending only upon the gauge fields and not other external fields is taken.87). Different numerical coefficients reflect different renormalization conditions for the two diagrams: conservation of the vector currents for the first one and symmetry between all three vertices for the second one.R L. All the anomalous Ward identities follow from the relations (13. (13.R L. respectively.91) The importance of this condition follows from the fact that the operator X is nonlinear in the gauge potential Aµ . One way to get the full anomalous terms G α (A) is to use the so-called Wess– Zumino consistency condition (Wess & Zumino 1971).R (x) the equation α α X L.R (x) = −cαβγ G L.86) and (13.R (x)G α (y) − X L.R (x)δ(x − y) L. To determine the effect of the anomalies. Therefore the Wess–Zumino condition completely determines G α [Aµ ] once the term (13.R L.R 2 L.R L.94) .85) and (13. After some calculation we can get G α (x) = ±(1/12π 2 ) Tr[T α ∂ µ εµνρσ (Aν ∂ ρ Aσ + 1 Aν Aρ Aσ )] (13.484 13 Chiral anomalies respectively.R The expression for the axial anomaly must satisfy the following consistency conditions (again.89) is given. since the G α (A)s factorize out of the path integral which cancels with the exp{−iZ [A]}. this current would be expected to transform covariantly under gauge transformations.81) or (13. In the second expression the first term comes from the AVV diagram and the second one from the AAA diagram.R β β γ (13.86) or (13.88) (we may need to introduce additional sources to the functional Z ). Consistent and covariant anomaly Finally we study the transformation properties of the anomalous current. Using (13. in addition to the gauge variation   δ A µ = Dµ      4 4 ˆµ (x)] δ Z [A] = −2 d x Tr[(δ Z /δ Aµ )δ Aµ (x)] = 2 d x Tr[Dµ J     4 α α   =− d x (x)G (A) (13. Naively.93) and it is much more complicated (Bardeen 1969).81)) V X α G A (y) = cαβγ G A δ(x − y) γ β A A X α G A (y) − X β G A (y) = 0 α β (13.92) L.88) follow from (13. see (13.82) and (13.R (y)G L.85) we get for the anomaly G L.

100) ˜α The current Jµ is also anomalous ˜ ˜ (Dµ J L. J µ ]α Bµ − [δ B G α (A)] α } (13.92) there exists the covariant non-abelian current L.97) ˆµ d4 x Jα {[Bµ (x).Rµ )α = G α (A) = ±(1/16π 2 )εµνρσ Tr[Tα G µν G ρσ ] (13. One can show (Bardeen & Zumino 1984) that in the case of the anomaly (13.Rµ + Pα (A) ˜L.Rµ = Jα Jα (13. The covariant current may have some physical significance when coupled to other external non-gauged fields.h.97) ˆµ α d4 x (δ Jα )Bµ (x) = − α ˆ ˜ d4 x {[ . By definition ˜ δ B Z [A] = −2 and ˜ δ B Aµ (x) = δ Aµ ≡ Bµ (x) The following operator equation holds ˜ ˜ ˜ δ B δ − δ δ B = δ[B.101) The anomalies (13. ] d4 x Tr[(δ Z /δ Aµ )δ Aµ ] = −2 ˆ d4 x Tr[ J µ Bµ (x)] (13. the second term is due to the anomaly. Bardeen & Zumino (1984) show that L. The physical sense of the original non-abelian current has been discussed earlier.92) and (13.s.96) Applying this operator to the functional Z [A] we obtain ˜ ˜ (δ B δ − δ δ B )Z [A] = = ˜ d x {[δ B G α (A)] 4 α  µ α ˆα )Bµ (x)}   (x) − (δ J     (13.101) are called consistent and covariant anomalies. Each can be used to study anomaly cancellation.Rµ ˆL.3 Anomalies and the path integral 485 we introduce the variation which defines the current.95) (13. gives the usual transformation property of the current. respectively. (x)]}α The gauge transformation properties of the non-abelian current follow from (13.13. .Rµ in gauge fields Its explicit form can be found by searching for a polynomial Pα ˆµ with an anomalous gauge transformation rule opposite to that of the current Jα .98) The first term on the r.Rµ Pα = ∓(1/24π 2 )εµνρσ Tr[Tα (Aν G ρσ + G ρσ Aν − Aν Aρ Aσ )] (13.99) L.

− pz ) ˆ pµ = i(∂/∂ x µ ) ˆ ˆ ˆ A0 = i A4 . ˆ Di = − Di . 3.103) ˆ where again µ = 1.106) ∂ ˆ ∂ ˆ ˆ µν ˆµ ˆν G a = − µ Aa + ν Aa + gcabc Ab Ac ν ∂x ˆ ∂x µ ˆ We define the Euclidean γ -matrices to be hermitean matrices γn obeying ˆ {γµ .108) µ.105) The field strength tensor is ˆ G iaj = G iaj where (i. The Dirac operator i D is hermitean. ν = 1.102) where µ = 1. −y. ˆ ˆµ Dµ = −∂/∂ x µ + ig Aa T a ˆ ˆ 4i G a = −iG a 0i (13.107) .3) x0 = −ix4 ˆ ˆ xi = xi p0 = −i p4 ˆ ˆ pi = pi xi = (−x. γν } = 2δµν ˆ ˆ and explicitly γ4 = γ0 . 2. 4. (13. . 3. 4 (13.110) γi = −iγi . − p y . ˆ =i¯ ¯ (13. 2. 3 (13. . First let us summarize the rules of continuation from Minkowski to Euclidean space. For the spatial coordinates we have (see Section 2.109) ˆ / and it is also hermitean. ˆ The matrix γ5 is taken as ˆ ˆ ˆ ˆ ˆ γ5 = −γ1 γ2 γ3 γ4 ˆ (13. The fermion fields and ¯ are independent variables in the path integral and in Euclidean space we define ˆ = .4 Anomalies from the path integral in Euclidean space Introduction In this section we discuss the techniques of anomaly calculation using the path integrals in Euclidean space (Fujikawa 1980). 2. j = 1. . 2. 3). We define the Euclidean vector potential Aµ as (13. ˆ i = 1. 4 and correspondingly (13.486 13 Chiral anomalies 13. −z) ˆ x µ = −xµ ˆ ˆ pi = (− px . ˆ Ai = − Ai (13. −∂/∂ xi )) µ ˆ D0 = i D4 .104) ˆ so that Aµ form a Euclidean four-vector. Thus for the covariant derivative Dµ = ∂µ + Aµ = ∂µ + ig Aa T a we get (∂µ = (∂/∂ x0 . .

Since i D is hermitean it has real eigenvalues λn ˆ / i D ˆ n = λn ˆ n (13.13.4 Anomalies from the path integral in Euclidean space 487 Under infinitesimal rotations in Euclidean space ˆ and ˆ † transform as follows: δ ˆ = 1 (γµ γν − γν γµ )ωµν ˆ ˆ ˆ ˆ ˆ ˆ 8 where ωi j = ωi j .111) ω0i = iω4i ˆ ˆ iS = − S where ˆ S= ˆ d4 x 1 ˆa ˆa G G 4 µν µν (13. ˆ and δ ˆ † = − 1 ˆ † (γµ γν − γν γµ )ωµν ˆ ˆ ˆ ˆ ˆ 8 (13. i D } = 0 ˆ / we get ˆ /ˆ ˆ i D γ5 ˆ n = −λn γ5 ˆ n (13.117) On the other hand eigenfunctions of the vanishing eigenvalue can always be chosen to be eigenfunctions of γ5 and thus to be chiral fields. This corresponds to stereographically projecting Euclidean four-space onto a four-sphere and changes the determinant but only by ˆ / a factor which is independent of the gauge field. The Euclidean action is (13. Since γ52 = 1 its eigenvalues ˆ are ±1 and γ5 (1 ± γ5 ) ˆ n = ±(1 ± γ5 ) ˆ n ˆ ˆ ˆ (13. For simplicity we assume that i D has a purely discrete spectrum. .113) ˆ ¯ / + ˆ (i D + iM) ˆ To complete this introduction let us discuss the eigenvalue problem for the ˆ ˆ / / hermitean Dirac operator i D in Euclidean space.114) and because ˆ {γ5 .116) (13.112) † ¯ so that ˆ 1 ˆ 2 is a scalar and we identify the transformation properties of ˆ as those † † of ˆ (in Minkowski space 1 γ0 2 is a scalar).118) In the rest of this section we work in Euclidean space and the caret is omitted. We observe ˆ / also that for λn = 0 chiral fields are not eigenvectors of i D ˆ / i D (1 ± γ5 ) ˆ n = λn (1 ∓ γ5 ) ˆ n ˆ ˆ (13.115) Thus non-vanishing eigenvalues always occur in pairs of opposite sign.

Under local infinitesimal axial transformations (13.122) is not invariant and following Fujikawa we can explicitly evaluate the Jacobian of the change of variables.121) †  / iD ϕn (x) = λn ϕn (x). d4 x ϕn (x)ϕm (x) = δmn  ¯ The coefficients an and bn are Grassmann variables. can be defined / precisely by expanding and ¯ in terms of the eigenfunctions ϕn of iD  † ¯ ¯ (x) = an ϕn (x). ϕn (x)bn (x) =   n n (13.123) If the measure were invariant under axial transformations then using the invariance of the path integral under a change of variables one could derive (10.124) or am = n † d4 x ϕm (x) exp[i (x)γ5 ]ϕn (x)an = n Cmn an (13.125) .120) (13.119) where and ¯ are two independent variables describing Dirac fermions coupled to gauge potentials transforming under a vector-like gauge group.119) the lagrangian is transformed into µ L (x) = L(x) + ∂µ j5 (x) and therefore / d4 x ¯ iD → / d4 x ¯ iD − d4 x µ (x)∂µ j5 (x) (13. We get (x) = n an ϕn (x) = n an exp[i (x)γ5 ]ϕn (x) (13. the measure (13.11) (with δ a L(x) = 0) implying the conservation of axial current at the quantum level. The measure becomes D D¯ = n dan m ¯ dbm (13.488 13 Chiral anomalies Abelian anomaly with Dirac fermions As mentioned in the previous section the anomaly of the ungauged axial U (1) 5 current jµ = ¯ γ µ γ5 can be understood as a non-invariance of the path integral measure under local transformations on fermion fields = exp[i (x)γ5 ] ≈ (1 + i (x)γ5 ) ¯ = ¯ exp[i (x)γ5 ] ≈ ¯ (1 + i (x)γ5 ) The generating functional in the Euclidean space exp{−Z [A]} = D D ¯ exp − / d4 x ¯ iD (13.122) up to an irrelevant arbitrary normalization factor. However.

Hence we may simply regularize the expression (13. i. we have C = exp[i (x)γ5 ] and det C = exp{i Tr[ (x)γ5 ]} = exp i n d4 x † (x)ϕn (x)γ5 ϕn (x) (13. In our problem we want to choose a regularization procedure which preserves the invariance under the vectorlike gauge group with gauge potentials Aµ .4 Anomalies from the path integral in Euclidean space 489 Thus (see (2.e. gauge transformed eigenfunctions are eigenfunctions of the gauge transformed operator.121) of the eigenfunctions of iD = i(/ + A) which ∂ / are gauge-invariant. with this fact in mind we have / been working in the basis (13. Actually.127) where the trace is taken over the whole Hilbert space.13.116)) dam = det−1 C m n dan (13. Going over to a .128) with a gaussian cut-off for large eigenvalues λn d4 x (x) n † ϕn (x)γ5 ϕn (x) ≡ lim M→∞ d4 x d x d4 x 4 (x) n † ϕn (x)γ5 ϕn (x) exp(−λ2 /M 2 ) n † / Tr[γ5 exp[−(iD )2 /M 2 ]ϕn (x)ϕn (x)] n ∗ [ϕn (x)]lb [(ϕn (y)]a k n = lim M→∞ (x) = lim M→∞ / (x) lim {γ5 exp[−(iD x )2 /M 2 ]}ab kl y→x (13.128) which is ill defined as it stands and must be regularized.129) where Tr means the trace over the group and the Dirac indices.126) and the Jacobian for D ¯ gives the identical factor. Furthermore. Hence the measure is multiplied by exp −2i d4 x (x) n † ϕn (x)γ5 ϕn (x) (13.

131) for (x) independent of x. Using (13. Using the result (13.490 13 Chiral anomalies plane wave basis and using the completeness relation for the ϕn (x) we get ∗ [ϕn (x)]lb [ϕn (y)]a k n = δ ab δ kl n x|n n|y = δ ab δ kl x|y d4 k x|k k|y (2π )4 d4 k exp(−ikx) exp(+iky) (2π )4 = δ ab δ kl = δ ab δ kl and therefore† d4 x = lim (x) n † ϕn (x)γ5 ϕn (x) (13. and writing † † † ϕn (x)γ5 ϕn (x) = 1 ϕn (1 + γ5 )2 ϕn − 1 ϕn (1 − γ5 )2 ϕn 4 4 † In Euclidean space we define ε 1234 = −1 and then Tr[γ5 γ µ γ ν γ ρ γ δ ] = 4ε µνρσ .128) and invariance of the path integral under the change of variables we finally get 5 ˜ ∂ µ jµ (x) = (i/8π 2 ) Tr[G µν G µν ] (13. γ ν ]G µν )2 exp(+k µ kµ /M 2 ) 4M 2 2! (2π )4 (x) ˜ (x) Tr[G µν G µν ] (13. As a by-product of this derivation of the abelian anomaly we can easily / understand the content of the so-called index theorem for the Dirac operator iD . The theorem relates the number of positive chirality minus the number of negative / chirality zero modes of the operator iD to the topological charge (Pontryagin index) of the background gauge field Aµ . (13.132) which corresponds to the previous result (13. .133) vanishes for eigenfunctions with eigenvalue λn = 0 since γ5 ϕn has eigenvalue −λn and is thus orthogonal to ϕn .123).130) M→∞ d4 x d4 x d4 x = lim = M→∞ d4 k / Tr[γ5 exp(+ikx) exp[−(iD )2 /M 2 ] exp(−ikx)] (2π )4 d4 k 1 21 (x) Tr[γ5 ([γ µ . observing that d4 x n † ϕn (x)γ5 ϕn (x) (13.57) obtained in Minkowski space by the triangle diagram calculation.131) 1 16π 2 where the last trace is over the group indices only.

135) The left-handed fermions L and the right-handed fermions ¯ L are independent variables. However. as we know from the / introduction to this section the operator iD (A) maps positive chirality spinors to negative chirality spinors / iD (A) L → R and consequently it does not have a well-defined eigenvalue problem and a welldefined determinant on the space of chiral fields. This suggests a possible approach to its calculation using the path integral formalism: calculate / Z [A] = − ln det iD (A) (13. i.e. as we know from the previous discussion.134) where n + (n − ) is the number of positive chirality (negative chirality) zero modes of / the operator iD and Q is the topological charge defined by (8. The topological charge density is connected to the anomaly density.3 we have understood the non-abelian anomaly as a breakdown of the gauge invariance of the functional Z [A]. . it can always be regularized in a gauge-invariant anomaly-free manner (in other words positive and negative chirality anomalies cancel each other). In this case det iD is real and. The only exception is when chiral fermions transform as a real representation of the gauge group.3 we assume here that the background gauge fields rapidly approach the pure gauge configuration of |xµ | → ∞ so that the Euclidean action is finite and the topological charge Q is well defined. Then we can formulate our theory in terms of the Dirac fermions = L + R and in Euclidean / space we are back to the hermitean eigenvalue problem iD n = λn n with real / λn .62). leftand right-handed chiralities transform under the same representation. In Section 13.4 Anomalies from the path integral in Euclidean space 491 for the eigenfunctions with eigenvalue λn = 0. As in Section 8.13. Non-abelian anomaly and chiral fermions We consider now the structure of the non-abelian anomaly in theories of chiral fermions coupled to an external dynamical or auxiliary gauge field Aµ and described by the effective action exp{−Z [A]} = D LD ¯ L exp − / d4 x ¯ L iD L (13.136) and its change under gauge transformation. we get n+ − n− = Q (13.

and different regularization methods we refer the reader to the original literature (Fujikawa 1984.6 replace photons by gluons and check that in QCD there is no anomaly for the axial isospin current. 13. For actual calculation. The main difficulty is to invent an appropriate regularization procedure.2 Show that Ward identities for the Green’s function of the three-vector currents are anomaly-free. the result e agrees with (13.137) But R + R ∗ is a real representation and consequently Z R+R ∗ [A] is gauge-invariant. Thus we see that the calculation of the non-abelian anomaly from the path integral is a well-defined problem. Of course.93) which in Euclidean space reads G α = Dµ L ∂Z i µνρσ ε Tr[T α ∂ µ (Aν ∂ρ Aσ − 1 iAν Aρ Aσ )] α = − 2 ∂ AµL 12π 2 (13.138) / / is meaningful since D −1 D maps a space of fermion fields with given chirality into itself.1 Derive the anomalous Ward identity (13. we could say / that only the phase of iD may be gauge-non-invariant. Alvarez-Gaum´ & Ginsparg 1984.139) Problems 13.492 13 Chiral anomalies Now consider chiral fermions in a complex representation R. The overall phase of the determinant is irrelevant for us and so is the dependence on A.3) is satisfied for this integral representation and studying the surface term induced by this change of the integration variable in the linearly divergent integral. We arrive at an interesting observation (Alvarez-Gaum´ & Witten 1983): in e Euclidean space only the imaginary part of the functional Z R [A] may suffer from the anomalies which give its anomalous variation under gauge transformation. / Equivalently. Nair & Trahern 1982. Marmo. . for example.21) starting with the unregularized integral representation (13. performing a shift of the integration variable such that (13. 13. if det iD was defined in the space of chiral fermions. A = 0. Balachandran. We can always imagine an extended theory with additional fermions in R ∗ so that Z R+R ∗ [A] = Z R [A] + Z R ∗ [A] = Z R [A] + Z ∗ [A] = 2 Re Z R [A] R (13.4. One can circumvent the / problem of defining det iD observing that it suffices to calculate the change of / ln det iD under a gauge transformation. Discuss this point also in Euclidean space using the material of Section 13.5). And / / / / ln det iD (A g ) − ln det iD (A) = ln det[−iD −1 (A)iD (A g )] (13. 13.3 In Fig. we can put. Leutwyler 1984).

4 Check that V V G α = −(1/4π 2 )εµνσ τ Tr[λα [ 1 Fµν Fσ τ + A 4 V V + 2 (Aµ Aν Fσ τ − Aµ Fνσ Aτ + 3 1 A A 12 Fµν Fσ τ V Fµν Aσ Aτ ) − 8 Aµ Aν Aσ 3 493 Aτ ]] where µν FV µν FA = ∂ µ V ν − ∂ ν V µ + [V µ . Note also that ˜ ζ A (t. τ ) and h(x. y. V ν ] + [Aµ . y. Aµ = +iλi Ai µ satisfies the consistency conditions (13.5 The determinant of an operator A with real positive discrete and increasing without bound eigenvalues λn : A f n = λn f n can be defined as det A = n λn = exp[−(dζ A /dt)(0)] where the so-called ζ -function. Aν ] = ∂ µ Aν − ∂ ν Aµ + [V µ . Aν ] i Vµ = +iλi Vµ . y. t)] n t→0 y→x . y. y) ≡ n ∗ f n (x) f n (y) = t λn h(x. τ ). V ν ] + [Aµ . x. 13.Problems 13. τ = 0) = δ(x − y) ∞ 0 1 (t) dτ τ t−1 h(x. τ ) obeys the so-called ‘heat equation’ A x h(x. Using the integral representation for the -function (t) = 0 ∞ x t−1 exp(−x) dx check that the ζ A -function can be calculated as ζ A (t) = where h(x. y. defined as ζ A (t) = n 1/λt n converges for Re t > 2 and can be analytically extended to a meromorphic function of t with poles only at t = 1 and t = 2 and is regular at t = 0. y. y. τ ) = n ∗ exp(−λn τ ) f n (x) f n (y) = x| exp(−Aτ )|y 1 (t) ∞ 0 dτ τ t−1 dx h(x.93). x. τ ) = −(∂/∂τ )h(x. τ ) The last equation can be used to regularize quantities such as + ˜ ϕn (x)γ5 ϕn (x) = lim Tr[γ5 ζ A (x.

134) starting with the anomalous divergence in Euclidean space 5 ∂ µ jµ = 2m ¯ γ5 − 2iQ(x) (Integrate over the Euclidean volume. x.7 Prove (13. Zumino. τ ) then has the asymptotic expansion / (Nielsen.) . 0) = (1/16π 2 )a2 (x. 13.6 Determine the abelian and non-abelian chiral anomalies in 2n-dimensional spacetime (Frampton & Kephart 1983. τ ) = for small τ . x) D (x − y)2 1 exp − 4τ 16π 2 τ 2 ∞ n=0 an (x. take the vacuum expectation value and then take the limit m → 0.494 13 Chiral anomalies with A = (D )2 . R¨ mer & van Nieuwenhuizen 1978) o h(x. Check that ˜ ζ ˆ 2 (x. y. Grisaru. y. y)τ n 13. Leutwyler 1984). The heat kernel h(x. Wu & Zee 1984.

Can we find an effective low energy description in terms of the Goldstone bosons only which would be SU (2) × SU (2)-invariant? Our goal is to implement the SU (2) × SU (2) symmetry in such a way that the two irreducible multiplets. an inspection of the lagrangian (9. The σ mass is proportional to the vacuum expectation value v and can be made arbitrarily heavy. However. the σ -exchange contribution to the low energy π π → π π scattering cannot be neglected when m σ → ∞. Imagine we are interested in the physics at low energy.38) explicitly breaks the SU (2) × SU (2) invariance of the lagrangian. with the σ meson decoupled. respectively. 2) under the chiral group splits then into massless Goldstone bosons and a massive σ = σ0 − v meson which transform under the unbroken subgroup SU (2) as a triplet and a singlet. It is natural to expect that the low energy region can be effectively described in terms of the Goldstone boson fields. Equivalently. The quadruplet of real fields 0 = (π 1 π 2 π 3 σ0 )T transforming as (2. for example. the triplet and the singlet of the unbroken SU (2) which are contained in the quadruplet 0 do not mix under general SU (2)× SU (2) 495 .14 Effective lagrangians 14. only.38) shows that this does not happen: the ππσ coupling is proportional to v and therefore. σ0 ) does no longer transform as a quadruplet under SU (2) × SU (2). Both the ππ ππ and the π πσ 2 couplings originate from the SU (2) × SU (2) invariant term (σ0 + π 2 )2 in the original lagrangian. We consider the σ -model with bosons only.1 Non-linear realization of the symmetry group Non-linear σ -model To become familiar with the effective lagrangian technique it is useful to follow the original approach and to begin with the discussion of the SU (2) × SU (2)-invariant σ -model in which the chiral symmetry has been spontaneously broken: SUL (2) × SUR (2) → SUV (2). decoupling of the σ means setting σ0 = v and π this breaks the invariance since (π . The reason is that a naive decoupling of the σ field by its elimination from the lagrangian (9. E m σ .

1) and (14. Let us take some 0 = (π (x). .2) and their matrix π representation in (14. First of all we observe that at a given point x the Goldstone boson fields can always be set to zero by a suitable redefinition of the vacuum.4) and 2 σ 2 (x) = π 2 (x) + σ0 (x) and therefore due to (14. Ya ).3) (14. Any chiral transformation on the vector 0 (x) can then be written as follows:     π1 (x) 0  π (x)   0     (x) = exp(−iαa G a ) 0 (x) =  2  = exp[−iξa (ξ.1) where the X a s are broken generators in (14.2) (14. α)X a ]  0  0   π3 (x)  σ (x) σ (x) 0 (14.3).3) a rotation of the vacuum by the angle ξ (x) sets the new Goldstone boson fields to zero locally at x. respectively.38) for the generators we can write    1    π (x) 0 −σ (ξ1 /ξ ) sin ξ  π 2 (x)   0   −σ (ξ /ξ ) sin ξ  2        = exp(−iξa X a )   = 0 (x) =  3  π (x)   0   −σ (ξ3 /ξ ) sin ξ  σ0 (x) where 2 2 2 ξ = (ξ1 + ξ2 + ξ3 )1/2 σ (x) σ cos ξ (14. σ0 (x))T in the originally defined vacuum. This can be achieved if we use the freedom of choice of the broken symmetry vacuum (Section 9. Tr[G j G i ] = δ ji and X a and Ya are broken and unbroken generators of SU (2)×SU (2).5) where G a ≡ (X a .496 14 Effective lagrangians transformations. In fact using the real representation (E.5) among the degenerate set |α |α = exp(+iαa X a )|0 The corresponding transformation of the field operators reads = exp(iαa X a ) exp(−iαa X a ) or in the matrix form = exp(−iαa X a ) (14.

semi-simple Lie group. ξa (x)s indeed transform linearly under the SUV (2). before we generalize our discussion. rotated. Thus.6) R is the matrix representation of the unbroken SU (2) under which the matrices X a transform exp(−iu c Yc )X a exp(iu c Yc ) = Rab X b i. for the purpose of further considerations let us check the transformation properties of the ξa (x)s under the unbroken SU (2) isospin group   0  0   exp(−iu a Ya ) 0 = exp(−iu a Ya ) exp(−iξa X a ) exp(iu a Ya ) exp(−iu a Ya )   0    0  0     = exp(−iξa Rab X b )   0  = exp(−iξb X b )  σ Hence ξb = Rab ξa   0 0   0  σ σ (14. Wess & Zumino 1969. Callan. Let G be a compact.14.7) . Coleman. belong to the same representation as X a s and transform according to R T . We consider a theory with some symmetry group G which is spontaneously broken to a subgroup H . a general chiral transformation on 0 (x) can be represented ξ by the new angles ξa (ξ . These results can be generalized as follows (Coleman. However.e. connected. every group element can be decomposed uniquely into a product of the form g = exp(iξa X a ) exp(iu a Ya ) (14.4) because σ 2 (x) = π 2 (x) + σ0 (x) is invariant under chiral rotations.1 Non-linear realization of the symmetry group 497 2 with the same σ (x) as in (14. vacuum. The vector (0 0 0 σ (x))T refers now to the new. Wess & Zumino 1969). in some neighbourhood of the identity of G. From the properties of the exponentials it follows that. Since σ (x) is an SUV (2) singlet we suspect that in general a chiral transformation can be factorized into an unbroken-SU (2)-subgroup transformation (which in our case acts on the SU (2) singlet σ (x)) and a rotation of the local vacuum orientation. α ) specifying the vacuum orientation which sets the transformed Goldstone boson fields locally to zero.

Indeed.11) we want to calculate matrix elements involving Goldstone bosons.7) = exp(−iαa G a ) 0 (x) = exp(−iαa G a ) exp[−iξa (x)X a ] (x) (14. .7). α ) and u(ξ . (14. α )X a ] exp[−iu a (ξ . 2 fπ = v2 (14. respectively. where h ∈ H . for example. Gasser & Leutwyler (1984)) and consists in gauging the full chiral group.‡ Thus.10) ξ ξ = exp[−iξa (ξ . g) and h = h(ξ. α )Ya ] (x) since exp(−iα ·G) exp(−iζ · X ) belongs to G and can be decomposed as in (14. We say that the fields (ξ.8) 0 (x)) (X a is the matrix representation of the broken generator X a appropriate for where under a general G transformation g(α) ξ ξ = ξ (ξ . on the multiplet . using (14. Given the transformation rule (14.7) means that once we have a parametrization l(ξ ) each group element g can be uniquely decomposed into a product g = l · h. g).498 14 Eff