Sie sind auf Seite 1von 412
Aspects of symmetry Selected Erice lectures of SIDNEY COLEMAN Donner Professor of Science, Harvard University CAMBRIDGE UNIVERSITY PRESS Cambridge New York New Rochelle “Melhourne Sydney Published by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 1RP 32 East 57th Street, New York, NY 10022, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia This collection © Cambridge University Press 1985 First published 1985 First paperback edition 1988 Printed in Great Britain at the University Press, Cambridge Library of Congress catalogue card number: 84-21364 British Library cataloguing in publication data Coleman, Sidney Aspects of symmetry: selected Erice Lectures of Sidney Coleman. 1. Particles (Nuclear physics) 2. Symmetry (Physics) I. Title 539.7721 QC793.3.HS ISBN 0 521 26706 4 hard covers ISBN 0 521 31827 0 paperback DE To my mother Contents Preface xiii Acknowledgements xiv 1 An introduction to nnitary symmetry 1 The search for higher symmetries 1 I The eight-baryon puzzle 1 .2. The elimination of Gp 4 2 SU) and its representations 5 2.1 The representations of SU(m) 5 2.2. The representations of SU(2) 6 2.3. The representations of SUG) 7 2.4 Dimensions of the IRs 8 2.5 Isospin and hypercharge 9 2.6 Isospin-hypercharge decompositions 10 2.7 The Clebsch-Gordan series 12 2.8 Some theorems 15 2.9 Invariant couplings 17 2.10 The problem of Cartesian components 17 2.11 SU(2)again 18 2.12 SU(3) octets: trilinear couplings 19 2.13 SU(3) octets: quadrilinear couplings 20 2.14 A mixed notation 21 3 Applications 23 3.1 Electromagnetism 23 3.2. Magnetic moments: baryons 24 3.3 Electromagnetic mass splittings 25 3.4 Electromagnetic properties of the decuplet 26 3.5 The medium-strong interactions 26 4 Ideas of octet enhancement 28 Bibliography 35 2 Soft pions 1 The reduction formula 36 2) The weak interactions: first prmeiples 40 3. The Goldberger Treiman relation amd a fret glance 4° Ahard look at PCAC 42 5 The gradient-coupling model 48 viii Ubon > Contents Adler's rule for the emission of one soft pion 47 Current commutators 50 Vector-vector commutators 50 Vector-axial commutators 51 Axial-axial commutators 51 The Weinberg- Tomozawa formula and the Adler—Weisberger relation 52 Pion—pion scattering é a Weinberg 57 Kaon decays 60 Appendix |: Notational conventions 63 Appendix 2: No-renormalization theorem 63 Appendix 3: Threshold S-matrix and threshold scattering lengths 64 Bibliography 65 Dilatations Introduction 67 The formal theory of broken scale invariance 68 Symmetries, currents, and Ward identities 68 Scale transformations and scale dimensions 70 More about the scale current and a quick look at the conformal group 71 Hidden scale invariance 76 The death of scale invariance 79 Some definitions and technical details 79 A disaster in the deep Euclidean region 80 Anomalous dimensions and other anomalies 82 The last anomalies: the Callan—Symanzik equations 84 The resurrection of scale invariance 88 The renormalization group equations and their solution 88 The return of scaling in the deep Euclidean region 90 Scaling and the operator product expansion 93 Conclusions and questions 96 Notes and references 97 Renormalization and symmetry: a review for non-specialists Introduction 99 Bogoliubov’s method and Hepp’s theorem 99 Renormalizable and non-renormalizable interactions 104 Symmetry and symmetry-breaking: Symanzik’s rule 106 Symmetry and symmetry-breaking: currents 108 Notes and references 11! Secret symmetry : an introduction to spontaneous symmetry breakdown and gauge fields Introduction 113 Secret symmetries in classical field theory 115 The idea of spontaneous symmetry breakdown 115 Goldstone bosons in an Abelian model 118 Goldstone bosons in the general case 119 The Higgs phenomenon in the Abelian model 121 Yang-Mills fields and the Higgs phenomenon in the general case 124 Summary and remarks 126 Secret renormalizability 128 The order of the arguments 128 Renormalization reviewed 128 Functional methods and the effective potennal 132 Contents The loop expansion 135 A sample computation 136 The most important part of this lecture 13% The physical meaning of the effective potential 139 Accidental symmetry and related phenomena 142 An alternative method of computation 144 Functional integration (vulgarized) 145 Integration over infinite-dimensional spaces 145 Functional integrals and generating functionals 148 Feynman rules 152 Derivative interactions 154 Fermi fields 156 Ghost fields 158 The Feynman rules for gauge field theories 159 Troubles with gauge invariance 159 The Faddeev-Popov Ansatz 160 The application of the Ansatz 163 Justification of the Ansatz 165 Concluding remarks 167 Asymptotic freedom 169 Operator products and deep inelastic electroproduction 169 Massless field theories and the renormalization group 171 Exact and approximate solutions of the renormalization group equations 174 Asymptotic freedom 176 No conclusions 179 Appendix: One-loop effective potential in the general case 180 Notes and references 182 ‘Classical lumps and their quantum descendants Introduction 185 Simple examples and their properties 187 Some time-independent lumps in one space dimension 187 Small oscillations and stability 191 Lumps are like particles (almost) 192 More dimensions and a discouraging theorem 194 Topological conservation laws 195 The basic idea and the main results 195 Gauge field theories revisited 198 Topological conservation laws, or, homotopy classes 202 Three examples in two spatial dimensions 205 Three examples in three dimensions 208 Patching together distant solutions, or, homotopy groups 209 Abelian and non-Abelian magnetic monopoles, or, (G/H) as a subgroup of 7,(H) 215 Quantum lumps 223 The nature of the classical limit 223 Time-independent lumps: power-series expansion 225 Time-independent lumps: coherent-state variational method 232 Periodic lumps: the old quantum theory and the DHN formula 22 A very special system 246 A curious equivalence 246 The sceret of the soliton 250 5.3 5.4 Contents Qualitative and quantitative knowledge 252 Some opinions 253 Appendix |: A three-dimensional scalar theory with non-dissipative solutions 254 Appendix 2: A theorem on gauge fields 256 Appendix 3: A trivial extension 257 Appendix 4: Looking for solutions 257 Appendix 5: Singular and non-singular gauge fields 259 Notes and references 262 The uses of instantons Introduction 265 Instantons and bounces in particle mechanics 268 Euclidean functional integrals 268 The double well and instantons 270 Periodic potentials 277 Unstable states and bounces 278 The vacuum structure of gauge field theories 282 Old stuff 282 The winding number 284 Many vacua 291 Instantons: generalities 295 Instantons: particulars 297 The evaluation of the determinant and an infrared embarrassment 300 The Abelian Higgs model in 1+1! dimensions 302 *t Hooft’s solution of the U(1) problem 307 The mystery of the missing meson 307 Preliminaries; Euclidean Fermi fields 311 Preliminaries: chiral Ward identities 314 QCD (baby version) 316 QCD (the real thing) 323 Miscellany 324 The fate of the false vacuum 327 Unstable vacua 327 The bounce 329 ‘The thin-wall approximation 332 The fate of the false vacuum 334 Determinants and renormalization 336 Unanswered questions 339 Appendix 1: How to compute determinants 340 Appendix 2: The double well done doubly well 341 Appendix 3: Finite action is zero measure 344 Appendix 4: Only winding number survives 345 Appendix 5: No wrong-chirality solutions 347 Notes and references 348 1/N Introduction 351 Vector representations, or, soluble models 352 $* theory (half-way) 352 The Gross-Neveu model 358 The CP*-' model 362 Adjoint representations, or, chromodynamics 368 The double-line representation and the dominance of planar graphs 368 3.2 3.3 3.4 3.5 3.6 Contents Topology and phenomenology 373 The ’t Hooft model 378 Witten’s theory of baryons 386 The master field 391 Restrospect and prospect 396 Appendix 1: The Euler characteristic 397 Appendix 2: The ’t Hooft equations 398 Appendix 3: U(N) as an approximation to SU: Notes and references 401 Preface I first came to Erice in 1966, to lecture at the fourth of the annual schools on subnuclear physics organized by Nino Zichichi. I was charmed by the beauty of Erice, fascinated by the thick layers of Sicilian culture and history, and terrified by the iron rule with which Nino kept students and faculty in line. In a word, I was won over, and I returned to Erice every year or two thereafter, to talk of what was past, or passing, or to come, at least insofar as it touched on subnuclear theory. Eight of these lectures, or more properly lecture series, are collected here. No attempt has been made to bring the lectures up to date. Typo- graphical errors, when spotted, have been corrected, and references to works to be published have b anged to references to published works. (I thank Hugh Osborne for wn on this dull task.) Otherwise, these are unaltered reprints of the original publications. Numerous debts are acknowledged in the individual lectures, but there is one overriding debt that must be acknowledged here. None of this would have existed were it not for Nino Zichichi, Of course, he is the creator and director of the subnuclear school, and of the International Center for Scientific Culture ‘Ettore Majorana’ which encompasses it, but, more than that, he is personally responsible for each and every one of these lectures. The lecture notes would never have been written were it not for his blandishments and threats, transmitted in a fusillade of urgent cablegrams and transatlantic phone calls at odd hours of the morning. This book may be the least of his many accomplishments, but one of his accomplishments it is, and it should be counted as such. Finally: These lectures span fourteen years, from 1966 to 1979. This was a great time to be a high-energy theorist, the period of the famous triumph of quantum field theory. And what a triumph it was, in the old , sense of the word: a glorious victory parade, full of wonderful things 2 An introduction ta unitary symmetry envisioned here is something like that which prevails for isospin symmetry; if we turned off the weak interactions and electro- magnetism, the world would be isospin symmetric, and all the particles in an isotopic multiplet would be degenerate; in fact, the world is not isospin symmetric, and the masses are slightly different.) This is sometimes phrased by saying that the strong- interaction Hamiltonian is a sum of two terms: one due to ‘very strong interactions’, which is symmetric under the higher symmetry group, and one due to ‘medium-strong interactions’, which is not. The second viewpoint is one which has met with considerable success, and we intend to explore it in more detail here. Now, what do we know about the structure of such a ‘higher symmetry group’? Well, in the first place, there are several assumptions we can make, not because they are forced upon us by the problem, but because they are straightforward generalizations of what we know about the isospin group, and because they simplify the problem considerably. We assume, in the limit of exact symmetry, that: (i) The group commutes with the Poincaré group. (ii) The group acts on the Hilbert space of the states of the world as a group of unitary operators. (This is connected with conser- vation of probability.) (iii) The one-particle states form an invariant subspace under the action of the group. The many-particle asymptotic states trans- form like tensor products of one-particle states. (iv) The group commutes with the S-matrix. A group which satisfies these conditions is usually called ‘an internal symmetry group’. It is worth noting that ina field theory any group whose generators are obtained by integrating, over all space, the time components of a set of conserved Hermitian currents is an internal symmetry group. In addition to these assumptions, there are several conditions forced on the group by the nature of the problem. (i) The group turns baryon states at rest, with spin up, into baryon states at rest, with spin up. Therefore, it must have an eight- dimensional representation. (ii) Since the whole point of the hypothesis is to explain the approxi- mate equality of the baryon masses, this representation must be irreducible. (iii) (This is a rather fine technical point, but it is necessary in order The search for higher symmetries 3 to avoid some pathological cases.) The eight-dimensional Tepresentation should be closed, in the sense that any matrix which is the limit of a sequence of representation matrices should bea representation matrix itself. Since the S-matrix is continuous, if we are given a representation that is not closed, we can always close it, and thus obtain a representation of a larger group which is still an internal symmetry group. (iv) The representation must be faithful. All observed particles can, in principle, be constructed out of the eight baryons and their antiparticles. Thus, any group element which acts trivially on the baryons will act trivially on everything known, and might as well be forgotten. These conditions are sufficient to ensure that the higher symmetry group must be a compact Lie group. This is a class of mathematical objects which has been thoroughly investigated; all the compact Lie groups have been classified, and so have all of their representations. So finding all the compact Lie groups with eight-dimensional representations is merely a question of knowing what mathematics books to look in. However, there is another condition, which arises because we want our higher symmetry group to contain the isospin—hypercharge group, the group cf the old familiar symmetries of strong-interaction physics. This is: (v) The group should contain a subgroup isomorphic to the isospin— hypercharge group. Furthermore, when we restrict ourselves to this subgroup, the eight-dimensional representation should decompose into an isodoublet of hypercharge one (N), an iso- singlet of hypercharge zero (A), an isotriplet of hypercharge zero (2), and an isodoublet of hypercharge minus one (=). To determine the groups which satisfy this last condition, as well as all the others, requires some independent effort on the part of the investigator; however, the work is straightforward and can be done in a few hours. At the end of all this one has a long, and unenlightening, list. However, it has a surprising property; every group on it contains either SU(3), the group of all unimodular unitary transformations on a three-dimensional complex vector space, or Gg, a group called ‘the connected part of the minimal global symmetry group’, whose structure I will explain below. These two groups are minimal: if their predictions are wrong, the predictions of any other higher symmetry group must be wrong, and 4 An introduction to unitary symmetry the whole idea of higher symmetry must be abandoned. We will begin with Go, since that turns out to be wrong; SU(3), of course, turns out to be right — otherwise I would not be giving these lectures. 1.2 The elimination of Gy Go is the direct product of three factors of SU(2), Go ZSU(2)@SU(2)@SU(2). Thus, every element of Gy can be written as a triplet of elements of SU(2), G15 92593), the generators of Gp are three commuting ‘angular-momentum vectors’, 1, T?, 1, and the irreducible representations of Go are labeled by three ‘spins’, (S15 82) 83). Isospin and hypercharge are imbedded in Gy in the following way: Ta1 412, and Y=2/9), The eight baryons transform according to the representation G, 0, )@G, 4, 0). (Thus, we have the desired isospin-hypercharge decomposition J=4, Y= +1, and J=0, Y=0.) This is a reducible representation of Go; there- fore Gp does not meet our conditions — though it is contained in many groups which do, The easiest way to display a group which does meet the conditions and which contains Gg is to add to Gy a discrete element which has the effect of interchanging I'?) and IS). The representation above then becomes an irreducible representation of the enlarged group. One particular element of Gy is Rao, R is a hypercharge reflection operator, RYR''=—-Y RIR“'=1. Simply from R invariance one can deduce an almost endless list of contradictions with experiment. I will give three here. (i) For every particle or resonance there must be another particle or resonance, of the same spin and nucleon number, with opposite hypercharge, and with approximately the same mass. Thus, there must be a low-lying #* resonance with hypercharge —1. Such a resonance does not exist. SU(3) and its representations 5 (ii) It is easy to show from R invariance, isospin invariance, and the assumption that the electric current transforms like the electric charge (ie. like /,+ Y/2), that the clectromagnetic self-energy of the &* should be the same as that of the Z~. In fact, they are separated by 13 MeV. (iii) From the same assumptions, the magnetic moment of the A should be zero. In fact, it is of the order of magnitude of the nucleon moments. Thus, Go must be rejected. All that is left, our last hope, is SU(3). 2 SU(3) and its representations In this section we will develop some of the properties of SU(3) and its representations, and also develop some methods for doing simple SU(3) calculations. We will begin by making some remarks about SU(n). 2.1 The representations of SU(n) SU(n) is defined as the group of all unimodular unitary nxn complex matrices. This definition immediately tells us one representation of the group. If we let x' be a complex -vector, on which the group acts in the following manner U; xi Ujx’, then it is clear that the space of all x! forms a basis for a representation of U(n). A basis for another representation is formed by a set of vectors Yi, which transform according to U: YUly,, where Ul= 03, (We use an overbar throughout to indicate complex conjugation.) The notation we have used, with its upper and lower indices, mimics that of ordinary tensor analysis. This mimicry is not deceptive, for, as a consequence of the unitarity of U, U: xyiox,y', and thus, just as in ordinary tensor analysis, the summation of upper and lower indices is an invariant operation. By the usual method taking direct products — we can form, from these primitive objects, the spaces of all tensors of rank n+m, with n upper and m lower indices, Each of these spaces clearly forms the basis for a representation of U(n). 6 An introduction to unitary symmetry These representations are, however, not necessarily irreducible. If a;, is a tensor of rank two, we may divide it into the sum of a symmetric and an antisymmetric part 4 j= Aint Hip This separation is invariant under the action of the group; thus we have divided the original representation space (of dimension n?) into two invariant subspaces (of dimension n(n+ 1)/2 and n(n—1)/2, respectively). Likewise, if aj is a mixed tensor, we may divide it into two parts, laa aj=- dak +aj, where a=0, and again we have two invariant subspaces, in this case of dimension ! and n?—1, (Note that in this case we cannot make a further reduction by symmetrizing and antisymmetrizing, because the indices transform differently under the action of the group.) One of the problems we set in the introduction was finding the irreducible representations of SU(3). Our general method of attack will be to (i) construct all tensors with a given number of upper and lower indices ; (ii) divide them invariantly into as many parts as we can; (iii) discard the parts which we can show lead to representations equivalent to those obtained from tensors of lower rank; and (iv) identify the remaining parts with (hopefully) new irreducible representations. This method is, in principle, capable of generating all representations of SU(n), for any n. However, the combinatorics required to keep track of the irreducible tensors becomes formidable for n greater than 3, and other methods seem to be more efficient. We will first tackle SU(2), to warm up. 2,2 The representations of SU(2) It is a great convenience to introduce the antisymmetric two- index tensor e'/, Under the action of the group U: e4(det Uje'; but since U is unimodular, e/ is invariant. So is ¢,, the corresponding entity with two lower indices. Given any tensor, we can use the ¢ tensors to raise and lower indices, just as the metric tensor is used in ordinary SU(3) and its representations 7 tensor analysis. In particular, we may write any tensor invariantly in terms of a tensor with all lower indices, and thus there is no representation of SU(2) induced on a general tensor that is not equivalent to one induced on a tensor with only lower indices. Let Gin. .in be a tensor of this kind. We may divide a into two parts, which are respectively symmetric and antisymmetric under interchange of the first two indices. With the aid of ¢,; we may write the antisymmetric part in terms of a tensor of lower rank: iy, in)... =Fiyadiy...- Thus, we need only consider completely symmetric tensors. Thus we have (hopefully) an inequivalent irreducible representation of SU(2) associated with every space of completely symmetric tensors of a given rank, with all indices lower. (We adhere to an ancient convention, and call this rank 2s.) We will call the representation D®’, or (s) for short. The dimension of (s) is the number of linearly independent tensors of the proper sort. Since the tensors are completely symmetric, and since the indices are allowed to assume only two values, this is the same as the number of ways 2s objects can be divided into two sets. That is to say, dim (s)=2s+1, a result which should be familiar. 2.3 The representations of SU(3) Now let us apply the same techniques to SU(3). The invariant ¢ tensors now have three indices, and thus cannot be used to raise and lower indices. Thus we have to work with tensors that have both lower and upper indices: However, we may still use the ¢ tensors to write the antisymmetric part of any tensor in terms of a tensor of lower rank allay taler elt k Bie, and therefore we need only consider tensors completely symmetric in both their upper and lower indices. By similar reasoning, we need only consider traceless tensors. We will define the representation induced on the space of all traceless, completely symmetric tensors with # upper indices and m lower indices as D™™, or simply (n,m) for short. Eventually, we will show that this > family of representations forms a complete set of inequivalent irreducible 8 An introduction to unitary symmetry representations. However, it is more expedient to first extract some Properties of these representations that are of practical importance, and only afterwards to prove their completeness, inequivalence, and irreducibility. Therefore, for the time being, we will simply call them IRs. If you want to think of ‘IR’ as an acronym for ‘irreducible representation’, you are welcome (indeed encouraged) to do so, but, in fact, we shall not use this property until we have proved it. Note that a simple consequence of our definitions is that (@, m)=(m, n), where the overbar indices complex conjugation. 24 Dimensions of the IRs To calculate the dimension of an IR is a straightforward exercise in combinatorics. The space of all completely symmetric tensors with n upper indices and m lower indices can be decomposed into the space of all symmetric tensors and to a space of tensors equivalent to the traces. (By ‘the traces’ we mean those tensors which are obtained by summing one upper index with one lower index.) The space of the traces is equivalent to the space of all completely symmetric tensors with (n— 1) upper indices and (m— 1) lower indices. In representation language, (n, NO, m)=(, m) @[(n—1, 0)@O, m—1)]. Taking the dimensions of both sides dim (n, m)=dim (n, 0) x dim (0, m)—[dim (n—1, 0) xdim (0, m—1)]. By arguments similar to those we used in the discussion of SU(2), dim (n, 0)=4(n + 2(n+ 1)=dim (0, n), and therefore, dim (n, m)=4(n+ 1)@m+ I(n+ mt 2). There is an alternative method of designating representations, much used in the literature, in which a representation is labeled by its dimension. Since (n, m) and (m, n) have the same dimension, they are distinguished by labeling a representation by its dimension if n is greater than m, and by its dimension with an overbar if m is greater than n. Thus (1, 0) is often called 3, 0,1)” "3 ayy” » g 3,0)" * ” 10, (2,2)" " “27, etc SU(3) and its representations 9 (This is still not totally unambiguous, since dim (4, 0)= dim (2, 1)=15, but it suffices for practical purposes.) 2.5 Isospin and hypercharge Merely to state that the very strong interactions are invariant under a given higher symmetry group is not sufficient information to construct a physical theory. We also have to know how isospin rotations and hypercharge rotations, the old familiar symmetries of strong- interaction physics, are imbedded in the group. The most convenient way of specifying this, for SU(3), is by giving the isospin and hypercharge transformation properties of the fundamental representation (1, 0). (1, 0) is a three-dimensional representation. Therefore, when we restrict SU(3) to the isospin group SU(2), it can decompose in only three ways: into the sum of three isosinglets, into an isosinglet and an isodoublet, or into an isotriplet. The first case is mathematically impossible, for it implies that all the elements of SU(2) are inside the identity element of SU(3). The third case is mathematically possible, but physically uninterest- ing: if the fundamental triplet contains only integral isospin, then all the IRs (which are made from direct products of fundamental triplets) would contain only integral isospins, which is not very satisfactory for explaining nature. Thus, only the second possibility remains. We will choose our basis in unitary space so that the (1, 0) representation looks like g qt 12 > ~ 12 q i where we have labeled the basis vectors by the appropriate eigenvalue of Z,. All that remains is to assign the hypercharge. In order that the hypercharge differences between observed particles be integers, it is necessary that the hypercharge difference between the singlet and the doublet be of magnitude one. We choose the hyper- charge of the doublet to be the greater. (This is just a matter of convention, although it does not appear to be so at first glance. If we were to choose the opposite assignment, the representation (0, 1) would have our original assignment. Thus, the structure of all calculations would be the same, except that (m, m) would everywhere be replaced by (m, n). This only has to do with our conventions about how we write things on paper, not with what goes on in the world. Formally, this degree of freedom corres- ponds to the existence of an outer mitomorphism of SU(3).)