t§qaruw"ts

"t"tnr ut,
o,
uo?rsnpo"rru?
14. P. B. Jones
THE OPTICAL MODEL IN NUCLEAR AND
PARTICLE PHYSICS
IS. R. K. Adair and E. C. Fowler
STRANGE PARTICLES
16. R. Wilson
THE NUCLEON· NUCLEON INTERACTION:
EXPERIMENTAL AND PHENOMENOLOGICAL
ASPECTS
17. J. F. Denisse and J. L. Decroix
PLASMA WAVES
18. W. F. Brown, Jr.
MICROMAGNETICS
19. A. Rose
CONCEPTS IN PHOTOCONDUCTIVITY AND
ALLIED PROBLEMS
20. A Guinier and D. L. Dexter
X.RAY STUDIES OF MATERIALS
21. T. G. Northrup
THE ADIABATIC MOTION OF CHARGED
PARTICLES
22. G. Barton
INTRODUCTION TO ADVANCED FIELD THEORY
23. C. D. Jeffries
DYNAMIC NUCLEAR ORIENTATION
24. G. B. Benedek
MAGNETIC RESONANCE AT HIGH PRESSURE
25. D. L. Dexter and R. S. Knox
EXCITONS
26. D. H. Sampson
RADIATIVE CONTRIBUTIONS TO ENERGY AND
MOMENTUM TRANSPORT IN A GAS
27. P. A. Carruthers
INTRODUCTION TO UNITARY SYMMETRY

Additional volumes in preparation

INTRODUCTION TO
UNITARY SYMMETRY

PETER A. CARRUTHERS
Cornell University, Ithaca, New York

1966
INTERSCIENCE PUBLISHERS

a division of John Wiley & Sons,

New York· London· Sydney
Copyright © 1966 by John Wiley & Sons, Inc.
All Rights Reserved. Reproduction or use in whole or in part of
this material is permitted for purposes of the United States
Government.
Library of Congress Catalog Card Number 66-20390
Printed in the United States of America
 Preface

In recent years the "eightfold way" symmetry scheme has

greatly clarified the systematics of the strong interactions of
elementary particles. In this theory the previously known con-
servation laws for isospin and strangeness have been unified in an
especially elegant manner. It has been known for a long time that
the strongly-interacting particles can be grouped into multiplets
whose members have nearly the same mass. The associated law
of conservation of isospin is enforced mathematically by requiring
the theory to be invariant under certain transformations of the
group S U(2). The latter group is generated by unitary transfor-
mations (with determinant unity) in a two-dimensional complex
vector space. In the new theory it becomes useful to group
together various isospin multiplets (composed of particles with
identical spin, parity and baryon number). The isospin and
strangeness quantum numbers can then be associated directly
with the eigenvalues of diagonal operators in a certain class of
representations of the group S U(3), the latter being the group of
linear unitary transformations (having unit determinant) in a
three-dimensional complex vector space.
The aim of this book is to provide physicists with a practical,
elementary introduction to the use of unitary symmetry in strong
interactions. It is assumed that the reader has a thorough under-
standing of the rotation group based on the algebra of the
"angular momentum" operators. Whenever possible an S U(3)
problem is solved in analogy to a familiar problem in S U(2). A
knowledge of group theory, especially representation theory, is
recommended but is not essential.
No attempt has been made to give proper credit for priority.
References have been cited in proportion to their familiarity to
the author and probable usefulness to the reader. We only wish
v
vi INTRODUCTION TO UNITARY SYMMETRY

to recall the important role played by Sakata, Ikeda, Ogawa, and

Ohnuki in introducing SU(3) (and in assigning the pseudoscalar
mesons to the regular representation) and the development of
the" final" version of the theory (the" eightfold way") by Gell-
Mann, Ne'eman and Okubo.
Many interesting and important subjects have been omitted
from this volume. Nothing has been said about the weak inter-
actions, mainly because the author does not feel competent to
write on this subject. Although the author is especially interested
in the connection between symmetry and dynamics, only one
section (7.3) is given to this aspect of SU(3) on account of the
perishable nature of "knowledge" in the field of strong inter-
action dynamics. Similarly, the brave and possibly profound
speculations on the existence of symmetries higher than S U(3)
have been completely ignored.
The following general references should be called to the reader's
attention.
l. R. Behrends, J. Dreitlein, C. Fronsdal, and B. W. Lee, Rev.
Mod. Phys., 34, 1 (1962).
2. R. E. Cutkosky, Ann. Rev. Nucl. Sci., 14, 175 (1964).
3. M. Gell-Mann and Y. Ne'eman, The Eightfold Way, W . A.
Benjamin, Inc., New York, 1965.
4. H. J. Lipkin, Lie Groups for Pedestrians, Interscience, New
York,1965 .
5. S. Okubo, Lectures on Unitary Symmetry (University of
Rochester preprint), unpublished.
6. J. J . de Swart, Rev. Mod. Phys., 35, 916 (1963).
About equal parts of the book were written at Cornell Uni-
v ersity, the California Institute of Technology and the Aspen
I nst itute for Theoretical Physics. The support of the Alfred P.
Sloan Foundation throughout this period is gratefully acknow-
led ged. The author is indebted to Professor J. J. de Swart for
permission to reproduce his tables of isoscalar factors.
 Contents

1. SU(2): Angular Momentum and Isospin 1

1.1 Rotations and Angular Momentum 1
1.2 Isospin 5
1.3 Field Operators for Arbitrary Isospin Multiplets 13
1.4 Violation of Isospin Conservation by Electromagnetism 18

2. SU(3) : Symmetry in Strong Interactions 22

2.1 Introduction 22
2.2 Infinitesimal Generators of S U(3) . 24
2.3 SU(2) Subgroups of SU(3) 33
2.4 Reflection Symmetries of Weight Diagrams 39
2.5 The Complex Conjugate Representations 3* 42
2.6 The Regular Representation; The F and D Matrices 49

3. Construction of Representations of SU(3) 56

3.1 Introduction 56
3.2 The Raising and Lowering Operator Technique 58
3.3 Casimir Operators for SU(3) . 72
3.4 Physical Interpretation of the Representations 76
3.5 The Reduction 8 x 8 . 84

4. General Properties of Representations of SU(3) 97

4.1 Introduction 97
4.2 SU(3) Clebsch-Gordan Coefficients. 99
4.3 Orthogonality Properties of the Representation Matrices. 102
4.4 Irreducible Tensors; Wigner-Eckart Theorem for SU(3) . 104
4.5 Symmetry Properties of the SU(3) Clebsch-Gordan Coeffi-
cients 105
4.6 Construction of Representations by Tensor Analysis 108

5. SU(3)-Invariant Vertices and Amplitudes 116

5.1 Construction of Invariant Couplings 116
5.2 Branching Ratios in Strong Decay Processes 128
5.3 SU(3) Invariance and Scattering Amplitudes . 136
vii
viii INTRODUCTION TO UNITARY SYMMETRY

6. Broken SU(3) . 142

6.1 Introduction 142
6.2 Mass Formulas (Neglecting Electromagnetism) 145
6.3 Mass Formulas (Including Electromagnetism). 152
6.4 Magnetic Moments 159
7. Crossing Symmetry of Strong Interactions 165
7.1 Crossing Matrices for Arbitrary Isospin . 165
7.2 SU(3) Crossing Matrices 176
7.3 Dynamical Relation of the Baryon Octet and Decuplet 180
Appendix I. Isoscalar Factors . 191
Appendix II. Vector and Axial Vector Currents; Algebra of Currents 206
Author Index 219
Subject Index 223
 Chapter 1

SU(2) Angular Momentum and Isospin

1.1 Rotations and Angnlar Momentnm

Consider a fixed vector r = (x,y,z). If we rotate the coordinate
system by an angle 8 about an axis given by the unit vector n
the fixed vector r appears to be r' = (x' ,y' ,z') in the rotated frame,
where
r' = r cos 8 + r x n sin 8 + n(n·r)(1 - cos 8) (1.1)
This expression clearly exhibits the fact that three independent
parameters are required for the unique specification of a rigid
rotation, and also that r'2 = r2. The usual way of writing (1.1)
is in terms of an orthogonal matrix with components Ojj:
(1.2)
The redundancy of this description (0 has only three independent
parameters among its nine components) is not very convenient for
practical calculations. Other well-known descriptions were inven-
ted long ago (Euler angles, quaternions, Cayley-Klein parameters).
A simple description of rotations is provided by the close con-
nection of transformations in spin space to rotations in ordinary
three-dimensional space. The usual spinor has two complex com-
ponents VI and v 2 • Under a rotation of the coordinate system by an
angle 8 about n, the spinor V = (VI ,v 2 ) transforms to v':

v' = Uv; U = exp (i;n.a) 1 cos~ + in.asin;; (1.3)

(1.4)

I
2 INTRODUCTION TO UNITARY SYMMETRY

a is the Pauli spin matrix vector. It will be observed that U is

unitary, utu = 1, and has det U = 1. The transformations U
clearly form a group and are specified by the parameters n, 8
associated with a rotation. The correspondence is not unique
since the physically equivalent angles 8 and 8 + 217 correspond to
± U; one says that the spinors form the basis of a double-valued
(irreducible) representation of the rotation group.
We can now explicitly exhibit the correspondence of the spin
transformation and the rotation, Eq. (1.1). Define the traceless
Hermitian matrix R by

R == r _ a= (z
x + iy
x--ziY ) (1.5)

Then we have the following theorem

R' = r' - a = URU-l == U(r- a)U-l (1.6)

where U is given by Eq. (1.4) and r' by Eq. (1.1). (This elegant
result is due to W. R. Hamilton; U and R are essentially quater-
nions.) The proof is as follows:
URU-l = R + [U,R]U-l
= r - a + (r x n)-asin8 + 2sin2t8[(r x n) x n]-a
r' - a = r - a + (r x n)- asin 8 + (1 - cos 8)[(n- a)(n-r) - r- a]
r' - a = [r cos 8 + r x n sin 8 + n(n-r)(1 - cos 8)]-a (1.7)
in agreement with (1.1). Liberal use has been made of the identity

(a-A)(a-B) = A-B + ia-(A x B) (1.8)

In words, (1.6) tells us the following: If we associate with the

vector r the matrix R in the two-dimensional spin space, the
unitary transformation U in this space is in direct correspondence
with the rotation in three dimensions. Further, (1.6) suggests that
a transforms as a vector with a' = U-1aU given by Eq. (1.1)
with r replaced by a. (Check this as an exercise.) Thus r- a =
r/ · o'.
SU(2): ANGULAR MOMENTUM AND ISOSPIN 3

We shall also define the above transformations in a way that is

not so directly tied to physics. Consider the set of linear unitary
transformations in a complex two-dimensional space

(1.9)

subject to the unimodularity condition

det U = 1; (1.10)

The unitarity-unimodular condition implies

a* y*) = ( S - fl) => S = a* (1.11)

( fl* S* -yay = -fl*
so that the transformation matrix U is

U- a fl) . 1 (1.12)
- ( -f3* a* '

The matrices U clearly form a group, known as S U(2), the group

of unitary 2 x 2 matrices with unit determinant. The corre-
spondence between (1.4) and (1.,) is
() . . ()
a = cos 2" + ~nz SIll 2"
(1.13)
fl = l·( nx - .). ()
SIn 2"
~ny

We have discovered two irreducible representations of the

rotation group, irreducible because it is impossible to form subsets
oflinear combinations of the basis functions [(x,y,z) and (V 1 ,V 2 ) for
the two cases discussed] that transform only among themselves
under arbitrary rotations. (For rotations about the z axis, x and
y span an invariant subspace, as does z alone. In this case the
representation is reducible and the matrix 0 is in block diagonal
form.) There is no question about which representation is more
4 INTRODUCTION TO UNITARY SYMMETRY

fundamental, since we can build the three-dimensional representa-

tion out of direct products of the spinors, but can never obtain
the double-valued representations by reducing tensors formed
from x j _
We mention here the basic weapon one has for generating
representations of some group once a representation has been
found. This is most easily expressed in terms of basis functions.
Suppose we have a set of functions o/j, j = 1, 2, . .. , N, which are
transformed into each other under the operations (designated R)
of the group. Then the transformed functions are

Ro/j == all = L Dij(R)o/i

j
(1.14)

and the matrices D form an N-dimensional representation of the

group. Then automatically the products 0/10/1'" o/z generate a
direct product representation, e.g.,

(o/jo/f)' = L
k,l
Dkl(R)Dlj(R)o/ko/l (1.15)

but these representations will in general be reducible. [For

example, from the tensor XtYf where XI and Y1 obey (1.1) and
i = 1,2,3 one can form the symmetric traceless tensor Tij =
XiYf + XiYi - iStix ,y, the skew tensor Sti = x iY1 - xjYt and the
scalar X· y. T lj , St}, X· Y form bases for the irreducible representa-
tions contained in the direct product 0 X 0.]
In the usual treatments of angular momentum the problem is
generally stated and solved in a slightly different way. One seeks
all possible representations of the (Hermitian) angular momentum
operators J i obeying the following algebra:

(1.16)

Eijk being the completely antisymmetric Cartesian tensor (e123 = 1).

In the usual way one seeks eigenfunctions of J2 = JiJ t and J 3 ,
and finds that the eigenvalue of J2 is j(j + 1), where 2j is any
nonnegative integer and j3 runs from - j to +j. Addition of
angular momentum proceeds by taking direct products as in
SU(2): ANGULAR MOMENTUM AND ISOSPIN 5

(1.15); since the standard procedure is generalizable, we sketch it

again. Consider the (2jl + 1)(2j2 + 1) states "'h ml"'iz
m 2· To reduce
the resulting representation into irreducible representatioI).s, one
begins with the" stretched" configuration "'hh (a)"'f2f2 (b), which
is the only state with J a = jl + j2 and has j = jl + j2. Remem-
bering that J = Ja + Jb and introducing the standard raising and
lowering operators J ± = J 1 ± iJ2 , the remaining states withj =
jl + j2 are found by applying J _. The multiplet with j = jl +
j2 - 1 is then begun by constructing the state orthogonal to that
with j = jl + j2' ja = jl + j2 - 1 and again applying J _ re-
peatedly. Finally, one gets one multiplet each for j running from
jl + j2 to Ij2 - jll. The coefficients relating the direct product
basis functions to the eigenfunctions of J2 and J a,j 1 2,j 22 are the
Clebsch-Gordan coefficients.
It will be observed that every representation of the commuta-
tion relations, and hence of the rotation group, can be obtained
by adding up spin 1/2 enough times, since (1.16) is equivalent to
the following definition of the rotation operator (h = 1)
R(n,O) = exp (iOn. J) (1.17)
The lowest dimensional (nontrivial) representation of (1.16) is
J= tao
1.2 Isospin
As in the case of the electron spin, the discovery of isospin
arose from the detection of previously unsuspected energy
degeneracies. There is every reason to believe that the strong
interactions possess, to a high degree of accuracy, the symmetry
known as "isospin conservation." This symmetry is commonly
taken to have the following meaning. The observed members of a
multiplet (dimension 2 T + 1) of strongly interacting particles
transform into each other under transformations of S U(2)
(equivalently: "rotations in isospin space") according to an
irreducible representation of dimension 2 T + 1. (The multiplets
are believed to be "pure" to an accuracy of order 1/137 due to
electromagnetic effects.) The interactions between the strongly
6 INTRODUCTION TO UNITARY SYMMETRY

interacting particles are invariant under these same transforma-

tions. Everybody is familiar with the dramatic conceptual
simplification brought about by this symmetry. Extensive treat-
ments are given in the books by Roman (1) and by Marshak and
Sudarshan (2).
Although isospin is an extremely familiar concept, we shall
review certain aspects of the formalism that help elucidate
similar procedures in unitary symmetry. We shall label the
components of the most fundamental mathematical entity in the
theory (the two-component spinor) with the names p (proton) and
n (neutron) (T3 + ± 1/2) in honor of their historical priority.
The basic spinor,

N = (~) (1.18)

is subject to infinitesimal transformations of S U(2)

N -* N' = (1 + iO(1·t't")N (1.19)

where 't" is formally the same as G in Eq. (1.4) and 0(1 is a real,
infinitesimal vector. (Recall that only infinitesimal transforma-
tions are needed.) To conduct an interesting experiment we have
to bring two nucleons together. The product wave functions
break up additively into parts of distinct isospin (according to the
law of vector addition of angular momenta) which may behave
quite differently. Mathematically, this resolution of a direct
product into terms of definite isospin (by means of the Clebsch-
Gordan coefficients) corresponds to a decomposition into irredu-
cible representations. If we consider two nucleons (labeled 1 and I

2) with spinors Nt(l) and Ni2), where i and j are + or - cor-

responding to p or n, the product function Nt(l)Ni2) transforms
under infinitesimal transformations according to the operator

(1.20)

so that the infinitesimal generator is T = t('t"l + 't"2). The trans-

formed state is clearly a linear combination of the original states
and so is manifestly a representation. This representation is,
SU(2): ANGULAR MOMENTUM AND ISOSPIN 7

however, reducible. The resulting irreducible representations are

the isospin I and 0 states:

T=I T = T3 = 0
I p(I)p(2)
I I
o V2 [p(I)n(2) + n(I)p(2)] V2 [p(I)n(2) - n(I)p(2)] (1.21)

-I n(I)n(2)

The same transformation that arranges the states N j (I)NA2)

into the isospin states (1.21) transforms the product matrices in
Eq. (1.15) to block diagonal form. Labeling these matrices with
the relevant isospin, this result is usually written symbolically as

D(1 /2) X D(1/2) = D(O) + D(l) (1.22)

Next consider the somewhat less familiar idea of the complex

conjugate representation. Since N transforms as

N~N' = UN (1.23)

the complex conjugate of this shows that the complex conjugate

spinor N* = (::) , which transforms as

N* ~ N*' = U*N* (1.24)

also generates a representation of S U(2) in terms of the matrices

U*. [Note: In field theory p will designate the Dirac field that
destroys a proton and creates an antiproton; p* will be the adj oint
field, creating a proton and destroying an antiproton; etc.] In the
case of SU(2) [but not SU(3)] the complex conjugate representa-
tion is equivalent (but not identical) to the representation having
the spinor N as the basis. The criterion for equivalence is the
existence of a similarity transformation by a matrix S such that
U = SU*S-l (1.25)
8 INTRODUCTION TO UNITARY SYMMETRY

Using the infinitesimal transformation we see that

N_(l +~Sex . 't")N

(1.26)
N* _ (1 - ~ Sex.'t"*)N
so that equivalence obtains for 8 such that
't" = -8't"*8- 1 (1.27)

Since 71 = 71*,73 = 73* and 72 = -72* we need 8 to obey

718 + 871 = 0
728 - 872 = 0 (1.28)
738 + 873 = 0
Eqs. (1.28) are clearly solved by 8 OC 72- A convenient (unitary)
8 is 8 = i72- Multiplying the second member of (1.26) by 8 we see
that the spinor Nc

Nc == 8N* = i-r2N* = ( 0
-1 0 Ii (Ii)
1)(P) -p
(1.29)

transforms identically to N

Nc - No' = (1 + ~ Sex.'t")Nc (1.30)

In using the usual Clebsch-Gordan coefficients it is essential to

have all states transforming in the same way_ Equation (1.29)
shows how the extra minus sign arises in the antinucleon wave
functions_ For instance, the T = 0 and T = 1 wave functions for
a nucleon-antinucleon system are (an overall minus sign has been
inserted)
T=1 T=O
-pn
1 1
V2 (pp - nit) V2 (pp + nit) (1.31)

np
SU(2): ANGULAR MOMENTUM AND ISOSPIN 9

The isotopic triplet has (for S wave, J = 0 configuration) zero

baryon number and odd intrinsic parity, just as the pions (7T± ,7T O)
do. Such a correspondence does not, of course, imply that pions
are bound states of NN.
The spinor N* is said to transform contragradiently to N. The
inner product Nt N = Li N j * Ni is a scalar under rotations, as
follows from (1.26) or equivalently

NtN --+ NtUtUN = NtN (1'.32)

Note that N*tN is not a scalar but NetN or NetNe or NtNe are
scalars.
We illustrate these considerations by constructing the isospin-
conserving Yukawa couplings between baryons and pseudo-
scalar mesons. First consider the nucleon-eta coupling (7] has
T = 0). Omitting space-time quantities to account for spin and
parity the obvious invariant is
(1.33)

wrItmg the more suggestive N in place of Nt. [We are now

using the symbol N to stand for the Dirac field operator. The
unitary operators required to generate transformations like (1.19)
are discussed in Appendix II.] The other invariants involving
Ne either do not conserve baryon number (NeN or NNe) or are
equivalent to (1.33) (NeNe). A more interesting example is the
vertex 8AK. The A has to absorb a K to become a 8. Since
the conjugate of K = col (K+ ,KO) does not transform like 8 =
col (8°,8-) the quantity 3KT A is not an isoscalar. However
Ke = col (K o, - K-) does transform properly so that
gKM;3AKe + H.c. (1.34)
is the appropriate form.
N ext we consider the regular representation of S U(2). Notice
that the three pions form an isotopic triplet and give rise to a
three-dimensional matrix representation of S U(2). Also notice
that the number of independent generators of infinitesimal trans-
formations is just three. Because of the fundamental structure of
10 INTRODUCTION TO UNITARY SYMMETRY

the commutation rules we can construct a special set of represent a-

tion matrices tk from the structure constants iSijk defined by the
commutation rules

(1.35)
as follows:
(1.36)

This can be checked by using the identity SijkSZmk = Di/D jm -

DimD jZ ' (Later we shall see that these results can be greatly
generalized.) This set of matrices is explicitly

and clearly all representations with T = 1 are equivalent to

(1.37). As an exercise we suggest that the reader show (n.t)3 =
n · t, and hence, that the rotation operator is

R(n,O) = exp (iOn· t)

= 1 + in·t sin 0 + (n.t)2(cos 0 - 1) (1.38)

Apply this to ordinary spatial rotations and show that (1.38)

applied to r = (x,y,z) gives r' as given in Eq. (1.1). Thus (x,y,z)
"transforms according to the regular representation" of the
rotation group. This exercise shows that the pion isovector cp =
(</>1</>2</>3) transforms according to the regular representation. The
connection to the usual complex field </> that creates 1T - and
destroys 1T+ is </> = (</>1 + i</>2)/V2.
Suppose we discover somehow that we have the triplet (</>1</>2</>3)
transforming according to (1.37) and (1.38)

(1.39)

(for instance, by finding a bound-state pole in the T = 1, NN

amplitude). Then how are we to construct the isospin-conserving
SU(2): ANGULAR MOMENTUM AND ISOSPIN 11

pion-nucleon Yukawa interaction? Clearly, we must form a linear

combination of the Nfil amplitude

fillet/Nf (1.40)
such that (using matrix rotation)

(1.41 )
is invariant under S U(2) transformations. Alternatively, since the
direct product in (1.40) has the reduction of Eq. (1.22), only the
T = 1 combination makes (1.41) invariant [D(1) x D(l) = DW) +
D(l) + D(2) but D(l) X D(O) = D(l)]. Invariance requires

(N' = UN, ep' = Vep) (1.42)

where U = 1 + (ij2)8a ' T and V = 1 + i8a·t. Thus we obtain
the condition
(1.43)

or, using the infinitesimal generators:

(1.44)
using (1.36). The unique solution of (1.44) is (to a constant factor)

(1.45)
so one obtains the standard answer

(1.46)
Note that by sandwiching the three Pauli matrices T between the
basic spinors we obtain an isovector V = NtTN transforming
according to the regular representation:

(1.47)
as is easily shown. We have emphasized these somewhat esoteric
points in order that the analogous results in S U(3) will not seem
astonishingly new.
As a final example, consider the 7TJ:J: coupling. Now all three
particles have isospin 1; in analogy to the pion field <p we write
12 INTRODUCTION TO UNITARY SYMMETRY

~ = (L\,.E2 ,.ES )' Following the same argument as in the TTN case
we seek a 3 x 3 matrix C such that

(1.48)
is invariant. The matrix C satisfies

[thO] = ieij/,Ck (1.49)

so that Ck == t k. Hence the 7T.E.E Yukawa coupling is

gnEEEt.E . re = - ignEEE. x ~·re (1.50)

the second form being more common.

We have done most of the work towards deriving the general
isospin-invariant Yukawa coupling of the eight baryons
(N,/l,.E,E) with the eight pseudoscalar mesons (7T,K,K,7]). For
later reference we write this down, omitting as usual the required
Dirac matrix Ys:
gnNNN'tN're + g"AE(.J~·re + H.c.)
+ g"EE( - iE. x ~ ·re)+ g"EE8'tE're
+ gKNA(N/lK + H.c.) + gKNE(N'tK-~ ~ H.c.)
+ gKEA(S/lKc + H .c.) + gKEE(8'tKc ' ~ + H.c.)

+gnNN NN7] + gnAAA/l7] + gnEEE.·~7] + gnEE8E7] (1.51)

where the particle symbols have the usual significance

(1.52)
SU(2): ANGULAR MOMENTUM AND ISOSPIN 13

1.3 Field Operators for Arbitrary Isospin Multiplets

In order to make use of the extensive results (3,4) availahle for
the group S U(2) it is often essential to pay close attention to
phase conventions among states and operators. For example,
when constructing the SU(2) invariant 7TNN coupling it was con-
venient to use the vector form of the pion field 1t = (7Tl,7T2,7T3).
However, physical states require the use of the spherical hasis, in
which T3 is diagonal. The commonly used complex fields cP, CPo, cp.
(cp =(7Tl - i7T2)/V2 creates 7T- and destroys 7T+, CPo creates and
destroys 7T O, cp. = (7Tl + i7T2)/V2 creates 7T+ and destroys 7T-) do
not create states related hy the standard (Condon-Shortley) phase
convention unless a judicious minus sign is supplied to the
7T+ state. This can he seen hy explicitly constructing the isospin
operator for the pion field. Rather than work out this special case,
we turn to the general prohlem (5). As the space-time properties
are of no special concern here, we consider hoson fields. We shall
consider two types of fields: (1) self-conjugate fields for which a
particle and its antiparticle helong to the same isospin multiplet
and (2) pair-conjugate fields, for which the antiparticles of a given
isospin multiplet constitute a distinct isospin multiplet. The two
fields shall he called cp(Jl) and if/ Jl ), respectively, where p. denotes
the decrease in T3 which results from application of the field to a
state. Clearly, the pion triplet (7T+ ,7T O,7T-) is a familiar example of
a self-conjugate isomultiplet. The conjugate douhlets of K
mesons (K+ ,KO); (j(o,K-) are typical pair-conjugates.
For the self-conjugate field we introduce the operator aJl(k)
which destroys particles of momentum k, isospin T (lahel sup-
pressed), and T3 component p.. The Hermitian conjugate operator
aJl *(k) is required to create an isomultiplet of cp-particles of
momentum k
(1.53)
in which the different states are related hy the standard phase
convention
T± I Tp.) = r±(p.) I T,p. ± 1)
(1.54)
r±(p.) = [(T + p.)(T ± p. + 1)]1/2
14 INTRODUCTION TO UNITARY SYMMETRY

Introducing a phase factor 'YJIl T of unit magnitude, we may write

</J(Il)(X) = L [all(k)fk(X) + 'YJ1l Ta_/(k)fk*(x))

k
(1.55)

(1.56)

where k·x = wt - k.x, w = (k2 + m2)1 /2, and Vis the quantiza-
tion volume. The functions fk(X) are normalized by

(1.57)
+-+
The symbol AooB signifies A(oBJot) - (oAJot)B.
The condition (1.53) suggests that </J(Il)* be required to trans-
form as a spherical tensor (3,4), i.e.,

(1.58)

(1.59)

Here ex is the set of three real parameters describing S U(2) trans-

formations and T is the isospin operator. [The reader may enjoy
demonstrating that (') an(') a -1 = exp ( - iex· t)n where the regular
representation matrices t were defined in Eq. (1.36).] The DVIl are
the standard representation matrices (in the convention of
Edmonds (4» usually labeled by three Euler angles a, fl, y. If we
choose (') a to be a rotation by 7T about the 2 axis, we find, using
D VIl (O,7T,O) = (-I)T+ lloll. -v:
e t "T 2 </J(Il)*e- t "T 2 = (-I)T+Il</J(-Il)* (1.60)

The phase 'YJ1l T is therefore required to obey

'YJ_Il T = (-1)21l'YJIlT (1.61)
We may construct the isospin operator in terms of the fields </J(Il)
and the (2 T + I)-dimensional matrices t l , t 2, t3 by means of the
standard isospin matrices t ± ==tl ± it 2, t3:
(t 3)IlV == (/L I t3 I v) = /LOllv
(1.62)
(t±)IlV == (/L I t± I v) = [(t ± /L)(t + /L + lW I20 Il,V±1
8U(2): ANGULAR MOMENTUM AND ISOSPIN 15

Taking (2.5) into account, one may write the isospin operator in
the form

(1.63)

By substituting the explicit expressions (1.55) and (1.62), one

finds

T3 = 2: fLajJ. *(k)ajJ.(k)
jJ.,k

T± = 2: U(t + fL + l)(t ± fL)F12 (1.64)

jJ.,k

If the operators T ± are to conform to Eq. (1.53), the coefficients

T}jJ. T must satisfy
(1.65)

One member of the multiplet can be chosen to have T} = 1,

whereupon Eq. (1.65) indicates that T}jJ. T alternates in. sign as we
go from one member of the multiplet to its neighbor. If the
multiplet has a neutral member with T3 = 0 (as occurs for integral
T, the only observed types of self-conjugate isospin multiplets)
the corresponding field component is real if we make the choice
T}o T = 1. Then we find T}jJ. T = (-I)jJ.. Although this is the con-
ventional choice for this special case, it is more convenient to
adopt a universal convention. To discover the most useful choice,
consider the case in which 2 T is an odd integer. Then a real T}jJ. T
is ( _ I)T ± jJ. + 1. The choice of sign is arbitrary but should be
adhered to. We choose the plus sign in accordance with the defini-
tion (1.59) of l') a' The extra factor of unity has been introduced
so that this T}jJ. T reduces to ( -1)jJ. for the only existing nontrivial
self-conjugate boson fields, which have integral isospin. We
therefore have
(1.66)
The presence of the factor (_I)T+jJ. can be understood easily in
terms of the analysis of the complex conjugate representation of
8U(2) (6). The matrix CjJ.v with elements (-I)T+jJ.ojJ., -v applied to
16 INTRODUCTION TO UNITARY SYMMETRY

the complex conjugate basis gives a basis transforming in the

standard way. The requirement that both components of 4>(tL)
transform in the same way under S U(2) transformations has
already been imposed in Eq. (1.58).
A similar analysis for the" pair field," for which we introduce
the operators atL and btL referring to the distinct multiplets,

tfo(tLl(x) = L [atL(k)f,,(x) + ~tLTLtL *(k)f" *(x)]

tL.k
(1.67)

shows that ~tL T = '7tL T. [The atL * and btL * are required to create the
two isospin multiplets, within which the convention (1.54) is
maintained.] The isospin operators are

T = LtLV i Jd3xtfo(tL)*ttLJJotfo(V) == (tfo, ttfo)

~

T± = L [(t
Il.k
=+= fL + l)(t ± fLW/2{a tL *(k)atL ±1(k) + LtL±1*(k)b_ tL (k)}

(1.6I!,V
It is frequently useful to discuss the behavior of states and
operators under the transformations of charge conjugation and
G conjugation. We define charge conjugation by the following
transformation, which connects a state IfL ) with its antiparticle
state 1-
fL)':
(1.69)
omitting an arbitrary phase factor.
For a 4>-type field, l-fL)' is a_/IO), whilc for the tfo-field it
is LtL *10) . Corresponding to (1.69), one has the operator
transformations
CatL *C-1 = '7tL Ta _/
(1. 70)
CatL *C-1 = '7tL Tb _/
for the 4> and tfo fields, respectively. Comparison with the defini-
tions (1.55) and (1.67) shows that
C4>(tL)C -1 = 4>(tL)*
(1.71)
Ctfo(tL)C -1 = tfo(tL)*
8U(2): ANGULAR MOMENTUM AND ISOSPIN 17

Applying (1.71) to the isospin operators (1.63) or (1.68) shows

that (explicitly for (1.68»
(1. 72)

The transpose tk of the matrix tk is given by

'I = tl
12 = -t2 (1.73)

i3 = t3
so that (1.72) reduces to

CT1 C-l = - Tl
CT2C-l = + T2 (1.74)
CT3C-l = - T3

(The same result clearly holds for the <p field.)

Another operator useful for the description of strong inter-
actions is G conjugation, defined as (7,8)

G = CR (1.75)

where R is rotation by 7T about the 2-axis in isospin space:

(1.76)

We have previously noted the relation

(1. 77)

This relation clearly holds for if/ u ).

From the explicit forms of the isospin matrices, Eqs. (1.62),
one notes the relations

(tl)UV = (t 1 )-u.-v
(t2)uv = -(t2)-u.-v (1. 78)

(t3)uV = -(t3)-u.-v
IS INTRODUCTION TO UNITARY SYMMETRY

Therefore under transformation by R, as indicated in Eq. (1.77),

the isospin operators (1.63) or (1.6S) transform as follows:
RT1R-l - Tl
(1.79)
RT3R-l - T3
Combining Eqs. (1.74) and (1.79) now reveals the most im-
portant property of the G operator:
GTG-l = T (l.SO)
Finally, the fields c/> and if; transform under G conjugation as
follows:
Gc/>(/J,)G-l = (-I)T+/J,c/>(-/J,)*
(LSI)
Gif;(/J,)G-l = (-I)T+/J,if;(-/J,)*

Note the familiar result that the pion field 1t = (1Tl,1T2,1T3) is odd
under G conjugation. This follows from (LSI) on using 1T(±l) =
+(1Tl ± i1T2)/V2,1T(O) = 1T3·
The use of isospin conservation to simplify the description of
strong interactions is too well known to require review here. A
thorough discussion may be found in references 1, 2, and 9.

1.4. Violation of Isospin Conservation by Electromagnetism

Experience shows that isospin is conserved to within a few per
cent so long as the relative kinetic energy of the colliding particles
is substantially greater than the mass differences within isospin
multiplets. As remarked earlier, the fractional deviations ~M/M
from the average mass M of a multiplet are of the same order of
magnitude (a few per cent) as the fine-structure constant e2/41T ~
1/137 which enters into simple theoretical estimates of ~M/M.
It is commonly believed that departures from perfect S V(2)
isospin symmetry can be completely ascribed to electromagnetism
(apart from tiny effects due to weak interactions). This belief
persists despite the absence of any widely accepted, numerically
successful calculations of these effects. We shall adopt this posi-
SU(2): ANGULAR MOMENTUM AND ISOSPIN 19

tion here, not only because the hypothesis is simple and plausible,
but also because no good competitive ideas seem to exist.
Since charge varies from state to state within an isospin multi-
plet in accordance with the Gell-Mann-Nishijima relation
Q = T3 + ty (1.82)
where the hypercharge Y is composed of the strangeness Sand
baryon number B, the electromagnetic current cannot commute
with the isospin operator T. The presence of electromagnetic
couplings therefore" violates" the conservation of T that follows
from the supposedly S U(2) invariant strong interactions. (It is
popular to pretend that the electromagnetic couplings can be
turned off, and that in such a limit perfect S U(2) isospin sym-
metry obtains.) This" violation" of the symmetry is not random,
however, but is exactly prescribed by (1.82), which states that the
electromagnetic current is composed of an isoscalar part (Y) and
the third component of an isovector.
This transformation property, though not so powerful in its
implications as complete isospin symmetry, leads to definite and
meaningful predictions by means of the Wigner-Eckart theorem,
which essentially states that the matrix elements of operators
having the same S U(2) transformation properties are propor-
tional to one another. More precisely, consider the spherical tensor
T/ of rank J having 2J + 1 components (f1- = -J, .. . , +J). By
definition, a spherical tensor operator of rank J transforms
according to the irreducible representation DJ of SU(2) :

(')"T/(,)" -1 = L Tu/Du'/
u'
(1.83)

(A particular instance of this definition has been used in Eq.

(1.58).)
Consider the matrix elements of T/ between states /ajm) and
/f3'j'm') where j(j + 1) and j'(j' + 1) are the eigenvalues of J2
and m,. m' the azimuthal projection J 3 ; a, f3 denote other quantum
numbers necessary to specify the state. Then the Wigner-Eckart
theorem states that
(f3'j'm' / T/ /ajm) = 8m '. m + uC(jJj';mf1-)(f3'j' I/ TJ I/aj) (1.84)
20 INTRODUCTION TO UNITARY SYMMETRY

where C(jJj' ;mp.) is the Clebsch-Gordan coefficient in the notation

of ref. 3 and the "reduced matrix element" <,B'j' II TJ IIai> is
independent of the azimuthal quantum numbers.
As an application of (1.84) let us consider two examples involv-
ing electromagnetism. First, consider the magnetic moment
operator p., which is linear in the current and therefore is of the
form 8 + V 3 , according to (1.82), where 8 is an isoscalar
operator and V3 transforms as the third component of an iso-
vector. According to (1.84), within an isospin multiplet the matrix
elements of V3 are proportional to those of T3 (with a common
factor). Hence for the E triplet (10)

p.(E+) = 8 +v
p.(EO) = 8 (1.85)
p.(E-) = 8 - v
from which we obtain the relation

(1.86)

No restrictions are obtained for the other baryon multiplets.

Next consider the photoproduction of pions from nucleons,
y + N -+ 7T + N. Treating the electromagnetic interaction to
lowest order gives a matrix element of the form

<7TNI (8 + V3 ) IN ) (1.87)

where the interaction Hamiltonian 8 + V3 is - f d3xj . A, j

being the current and A the vector potential. The scalar part
cannot change the isospin of the initial state as shown by

(1.88)

while the isovector part V3 can change the isospin:

(1.89)
Thus three reduced matrix elements

(1/211 8 /11/2), <3/211 V3111/2), and <1/211 V3111/2 )

SU(2): ANGULAR MOMENTUM AND ISOSPIN 21

are required to describe the four charge state reactions y + p --+

p + 1T O, y + p--+n + 1T+, Y + n--+p + 1T-, and y + n--+n +
O
1T • It happens that the low energy region is dominated by the

3-3 resonance (T = J = 3/2 +), so that in this special case one

amplitude is adequate. Then the Clebsch-Gordan coefficients give
simple ratios for the charge state ratios.

References
1. P. Roman, Theory of Elementary Particles, North-Holland, Amsterdam, 1960.
2. R. E. Marshak and E. C. G. Sudarshan, Introduction to Elementary Particle
Physics, Interscience, New York, 1960.
3. M. E. Rose, Elementary Theory of Angular Momentum, Wiley, New York,1957.
4. A. R. Edmonds, Angular Momentum in Quantum Mechanics, Princeton
University Press, Princeton, N.J., 1960.
5. P. Carruthers and J. P. Krisch, Ann. Phys. (N. Y.), 33, 1 (1965).
6. E. P . Wigner, Group Theory, Academic Press, New York, 1959, p. 288.
7. L. Michel, Nuovo Cimento, 10, 319 (1953).
8. T. D. Lee and C. N. Yang, Nuovo Cimento, 13, 749 (1956).
9. G. Kallen, Elementary Particle Physics, Addison Wesley, Reading, Mass., 1964.
10. R. E. Marshak, S. Okubo, and E. C. G. Sudarshan, Phys. Rev., 106, 599 (1957).
 Chapter 2

SU(3): Symmetry in Strong Interactions

2.1 Introduction
In Chapter 1 we discussed the fact that the strongly interacting
particles all seem to belong to isospin multiplets. [For experi-
m ental evidence one may consult the review of Dalitz (1) or that
of Rosenfeld et al. (2).] If we now compare particles which have
different isospin, but are otherwise identical (having the same
spin, parity, strangeness, and baryon number), further interesting
patterns emerge (Fig. 2.1). These isospin multiplets, when
grouped in this way, suggest a degree of organization more per-
vasive than that made possible by isospin conservation. It is
further apparent that any "higher symmetry" introduced to
account for these similarities is much less accurate than the
approximation of isospin conservation. Indeed, the masses of the
isospin multiplets comprising the" super-multiplets" are so dis-
parate, especially for the pseudoscalar mesons, that one may
doubt whether considerations based on symmetry arguments have
any relevance to the real world. In some cases, however, the
masses are close enough to suggest a vestigial degeneracy, masked
by some moderately strong symmetry-breaking interaction, in the
absence of which the members of the super-multiplets really
would be degenerate. This idea has a counterpart in the way in
which electromagnetism "breaks" isospin conservation. The
electromagnetic interaction contains a part proportional to the
third component of an isovector and so is not invariant under the
appropriate S U(2) transformations. In Chapter 1 we discarded
the (weak) electromagnetic interaction and studied the hypothet-
22
SU(3): SYMMETRY IN STRONG INTERACTIONS 23
y

n p

:r- A O
:ro :r+
T3
-I _12 2 1
a-
0 -I
a
0
O

(0)

Y
Y
KO
0
K+
0
K:o oKR+

Tr" TJ rr
o rr+ p- po p+
T3 T3
-I -~
I -I -~ WI ipo~ 1
2

-I -I
0
K-
0-
KO K*- R*o
(b) (c)

Y
NR- NAG NR+ NR++
0 0 0 0

yR_ y,Ro y*+

1 1 1
-1 -I -~ 0 ~ 1 i T3
a:- -I au 0

,n- -2
(d)

Fig. 2.1. Supermultiplets of particles of the same spin, parity, and baryon
number fall into simple geometrical patterns when placed on a hypercharge-3rd
component of isospin plot.

ical limit of perfect S U(2) symmetry. We follow a similar (less

plausible) procedure here. At present, it has to be admitted that
the unusual success of predictions based on perfect SU(3) sym-
metry (in apparent defiance of the large symmetry-breaking
terms) is not understood in any deep way.
24 INTRODUCTION TO UNITARY SYMMETRY

To begin, let us ignore the mass splittings so that there is

(theoretically) a set of degenerate states with identical J and P.
Later, we can worry about whether the observed splittings are
due to a violation of the symmetry by something" extraneous"
(e.g., as a magnetic field applied to an atom lifts the degeneracy
in J z ), or whether they arise spontaneously in a dynamical way.
One of the most exciting questions presently under investigation
is the relation of these" internal" quantum numbers (isospin and
strangeness) to dynamics. Such relations do not seem startling in
older branches of physics. Consider the fact that (at least at low
temperatures) atoms like to form crystals. Here the mathe-
matical principle is that one must minimize the free energy. With
sufficient skill one can dissect the partition function in such a way
to show that (for example) sodium ordinarily likes to be in the
fcc configuration. Although it is often incredibly difficult to really
calculate the most favorable symmetry for a given type of atom,
nobody is astonished at the existence of crystals or the decisive
effect of their symmetry on all physical properties.
We seek, then, a generalization of the isospin idea that includes
the hypercharge quantum number yet preserves isospin [S U(2)] in-
variance. (Here we have introduced the hypercharge Y = S + B
where S is strangeness and B the baryon number.) One possible
scheme is the following: since the isospin transformations cor-
respond (homomorphically) to rotations in a three-dimensional
space, one might consider the representations of the four-dimen-
sional orthogonal group. There is some historical precedent for
such a move, e.g., the Lorentz group, a special case of the complex
orthogonal group 0 4 , contains the rotation group as a subgroup.
This approach is discussed in reference 3,. It should be clear,
however, that the isospin group is really S U(2) and not 0 3 •
Hence no enormous surprise should be felt at the greater success
obtained by generalizing SU(2) to SU(3).

2.2 Infinitesimal Generators of SU(3)

To discover SU(3) let us recall our discussion of isospin. There
the fundamental entity was the nucleon spinor. Not only could
 ' U( !I) : S YMMETRY IN STRONG INTERACTIONS 25

W l1 huild up all representations of 5U(2) with this spinor, we could

ev e n pre tend that all nonstrange strongly interacting particles

could hc built up from hound states of nucleons and antinucleons.
Unf'nrLunately the idea of strangeness seems quite remote from
thaL Lheory. One can, however, introduce strangeness hy adding
one more "fundamental" haryon to the nucleon douhlet. The
hCSL candidate for this is the A, since it is an isotopic singlet. Thus
we have the hasic triplet

. (2.1)

where we pretend as usual that p, n, and A are degenerate. Now

all particles are to he huilt up out of if; and anti-if;. This is the
Sakata model (4) which was developed in the context of the
group U(3) hy Ikeda, Ogawa, and Ohnuki (5). It turns out that
the specific assignments in (2.1) are incorrect, and the Sakata
model has to he abandoned in its details. We shall use the his-
torical approach because of its intuitive appeal and because it
leads naturally to the more successful "eightfold way" of Gell-
Mann (6) and Ne'eman (7). The quadratic structure of Lagrangian
theories dictates that the symmetry transformations be unitary:

if;' = Uif; utu = 1 (2.2)

Since the determinant of a unitary matrix is etn , T} real, we can
separate this phase transformation (it corresponds to baryon
conservation) and deal entirely with unimodular transformations:
det U = 1 (2.3)
The group associated with continuous unitary unimodular trans-
formations U in n dimensions is called 5U(n).
We write the basic 5U(3) transformation as
U = exp (ia.. ),/2) (2.4)
where a. is an eight-component real vector and the Aj are a set of
eight traceless 3 x 3 Hermitian matrices, in exact analogy to the
26 INTRODUCTION TO UNITARY SYMMETRY

Pauli spin matrices a of SU(2). For most purposes we only need

to work with the infinitesimal transformation

U = 1 + tiOtt'A (2.5)

The Aj are required to be traceless by the condition

t
det U = 1 + 2" ott·Tr A + O(Ott)2 = 1 (2.6)

neglecting as always terms of second order in Ott. There are

3 2 - 1 = 8 independent 3 x 3 traceless 'Hermitian matrices (cf.
22 - 1 = 3 independent 2 x 2 traceless Hermitian matrices aj)
which we shall write down shortly.
The "self-representation" (2.4) generated by transformations
on the basic triplet (2.1) is referred to by the symbol 3, or
by D(I, 0) [or simp~y (1,0)], in a notation to be explained in
Chapter 3.
As in the case of angular momentum it is sometimes convenient
to use the Hermitian generators and sometimes the raising and
lowering operators. The latter are especially useful in constructing
representations. For S U(3) we can define six operators that per-
mute the various particles p, n, and A. The normalization is
chosen to give maximum symmetry to the diagram displaying the
eigenvalues of the diagonal generators defined below.

~ J. G° D
1

E,

°
E, ~ J. (~ °° ~)
°
E, ~ J. G°° ~)
°
SU(3): SYMMETRY IN STRONG INTERACTIONS 27

I t will be noticed that most standard works (e.g., ref. 8) use E _ 2

fo r our E 2 • The present choice increases the symmetry between
I~l' E 2 , and E3 and permits one to write the commutation rules
in a m ore succinct form than is otherwise possible. E_2 changes a
I I into a p. In addition to these six matrices we need two traceless
ma trices, independent of (2.7). These can be chosen to be diagonal.
We choose one (HI) to correspond to T3 [(1/2,-1/2,0) for p, n, and
II, r es pectively] and the other (H2 ) to be the same for p and nand
di ffe rent for A, so that H2 will be simply related to strangeness or
hyper charge.

o o
-1 1 ~ ) (2.8)
o o -2

Clearly we can express the Aj and also the matrix U of Eq. (2.5)
ill terms of (2.7) and (2.8). The physical interpretation of Yand
'l'3 require us to choose HI and H2 diagonal. Next we write down
the commutation relations between these generators. As we know
from Chapter 1, the algebra defined by these commutators poses
1\ m uch more general problem than the specific three-dimensional

onc. In particular we can seek all possible represcntations of the

a lgebra (i.e., matrices satisfying the commutation rules); in so
do ing we also find all representations of the group (9).
Before writing down the commutation relations it is helpful to
lIisc uss the eigenvalues of HI and H2 in the defining three-
dimensional representation. For purposes of discussion we trcat
lhe basic triplct !f in (2.1) as a wave function so that the three
cntries are the probability coefficients for p, n, and A. W c now
drop the p, n, and A notation in favor of numbering these
respective positions 1, 2, and 3. Now a "1" state has wave
f unction !f1 = c(,j (</>,0,0), where </> is the spatial wave function,
ctc. Dropping the common factor </>, the wave functions !fk
with lth component SkI (k,l = 1,2,3) are eigenfunctions of the
vector H:
(2.9)
28 INTRODUCTION TO UNITARY SYMMETRY

(2.10)

1
(2.11)
3

These two-component vectors mle are the weight vectors of the

defining representation 3. The corresponding weight diagram is
displayed in Fig. 2.2, along with the root vectors r(ex); ex = 1,2,3,

1
Ir(ex)1 = VS (2.12)

As shown below, the weights [eigenvalues of H, Eq. (2.9)] of any

representation lie on a two-dimensional lattice whose translation
vectors are any two of the root vectors. We also make the defini-
tion
r( - ex) == - r(ex) (2.13)

which is demanded by the structure of the commutation rules set

down below. The effect of the matrices Ea on the .pie is given in
the following

while E1P vanishes, etc.

SU(3): SYMMETRY IN STRONG INTERACTIONS 29

Fig. 2.2. Weight diagram for the defining representation 3 of 5U(3). The vectors
mt are the weight vectors, eigenvalues of the operator H. The r(a) are the root
vectors.

The result of computing the commutators of the eight matrices

E ±a (a = 1,2,3), HI' and H2 can now he summarized:

(2.15)

(no sum) (2.16)

(no sum) (2.17)

Here a, {3, and y run over 1, 2, and 3. In Eq. (2.15) the index y is
s ummed over. Equation (2.13) follows by requiring (2.16) to be
consistent for (3 = a. Further relations involving the E _ a follow
from (2.15) to (2.17) by Hermitian conjugation, recalling Eat =
E_ a •
We may now construct a set of Hermitian generators from the
set E±a' Hi. These are defined by the relations:

A4 = V6(E2 + E_ 2 )
(2.18)
A6 = V6(E3 + E_ 3 )
30 INTRODUCTION TO UNITARY SYMMETRY

Explicitly the matrices \ are given by:

-,

~)
1
Al =
(: 0
0
A2 =
(: 0
0 ~)
G ~)
0 0
A3 = -1
0
A4 =
(: :) 0
0 (2.19)

A5 =
(;
0
0
0 n A6 =
(: :)
V3
1
0
0
1

0 0

(: ~i)
0
1
A7 = 0 As = 0 0
V3
i 2
0 0
- V3
The first seven Ai have been chosen to resemble the ordinary Pauli
matrices. Note that AI' A2, and A3 generate the algebra of 5U(2).
In analogy to the Pauli spin matrix relations,

(2.20)
(2.21)

the A's have the following relations:

[Ai,Aj] = 2i];jkAk (2.22)

{Ai,Aj} = -t8 ij I + 2d tik Ak (2.23)
In (2.22) fiik is clearly antisymmetric in i and j, and also real by
virtue of the Hermiticity of the Ai and the antisymmetry. Using
the trace property
(2.24)
5U(3): SYMMETRY IN STRONG INTERACTIONS 31

we can see that fiji, is totally antisymmetric. Multiply (2.22) by

Al and apply (2.24):
4ifiil = Tr (AIAiAj - AI,\A j)
= Tr (AjAIAi - AjAj'\) (2.25)
= Tr Aj[AI,A;] = 4iflif
= -4if!lj
using the cyclic invariance of the trace and the basic definition
(2.23). Antisymmetry in the first and third indices is a consequence
of the relations already noted. Similarly, the real coefficients d jjk
are totally symmetric in i, j, and k. Note that the 4/3 in (2.23) is
necessary for the compatibility of (2.23) and (2.24).
The actual values Offijk and d ijk are found by patiently working
out the commutation properties of the matrices listed in Eq.
(2.19). The nonzero coefficients are given in Table 2.1.

TABLE 2.1
The Nonvanishing Structure Constants [,Jk of the Hermitian Generators Are
Given. The d'Jk' Which Are Useful as Clebsch-Gordan Coefficients, Are Defined by
the Anticommutation Rules of Eq. (2.23) for the 3 x 3 Generators (1/2)"')...
ijk [,Jk ijk d'Jk

123 1 118 1/'/3

147 1/2 146 1/2
156 -1/2 157 ,1/2
246 1/2 228 I/V3
257 1/2 247 - 1/2
345 1/2 256 1/2
367 - 1/2 338 I/V3
458 V3/2 344 1/2
678 V3/2 355 1/2
366 - 1/2
377 -1/2
448 - 1/(2V3)
558 -1/(2V3)
668 -1/(2V3)
778 -1/(2V3)
888 -1/V3
/ ' 32 INTRODUCTION TO UNITARY SYMMETRY

It will be noted that the Hermitian generators G = tA lead to

a more succinct expression for the commutation rules

(2.26)

(a,b,c = 1,2, ... ,8) than do the non-Hermitian generators E+a,Hi'

The structure constants ifabc are extremely important in the study
of the group and shall be the subject of much discussion in sub-
sequent sections. An n X n Hermitian representation of the
algebra (2.26) of SU(3) by eight matrices Ga leads to unitary
representations of the form exp (iex· G), where ex is a real eight-
component vector with continuously varying components
describing S U(3) transformations. The manifold of the param-
eters Clk always includes the origin, which corresponds to the
identity transformation of the group. Although the n x n matrices
2G a ' do not satisfy a relation like (2.23), the coefficients dabc are
also useful and will reappear later as Clebsch-Gordan coefficients
for the" regular representation" basis functions.
Finally, we include for completeness, one other useful sct of
generators used extensively by Okubo (10). One defines a set of
nine real traceless matrices A/ with matrix elements (A/)"v
(i,j,f1-,JJ = 1,2,3) by

(A/)jLV = Oivl>jjL - tOijOIlV (2.27)

Only eight of these matrices are independent since
All + A22 + A 33 = 0 (2.28)
The commutation rules are easily computed
[A/,A,k] = o/A/ - o/A/ (2.29)
and the connection of A/ to Ea, H j is as follows:

A21 = V6E_ I A 3I A 32

Al2 = V6EI AI 3 A 23 (2.30)

All - A22 = 2V3HI ; All + A22 = 2H2
We now show that the allowed weights must lie on the two-
dimensional lattice having any two of r(I), r(2), r(3) as primitive
SU(3) : SYMMETRY IN STRONG INTERACTIONS 33

Fig. 2.3. The allowed eigenvalues (weights) ofH lie on a two-dimensional hexagonal
lattice whose translation vectors are any two of the roots r(a).

t.ranslation vectors (Fig. 2.3). Suppose if!(m) is an eigenfunction of

II with eigenvalue m. Then Eq. (2.17) shows that the state
E"if!(m) has eigenvalue m + r(a), unless it is identically zero:
HEaif!(m) = [H,EaJif!(m) + EaHif!(m)
(2.31)
= (r(a) + m)Eaif!(m)
Similarly, E_aif!(m) either vanishes or has weight m - r(a).

2.3 SU(2) Subgroups of SU(3)

I n the Sakata model the first two components of the triplet if!
lice an isospin doublet (p,n). From Eq. (2.11) we see that this
ide ntification can be made more generally for the basic triplet,
t he correspondence being T3 = V3 HI. The third component of
l/'is accordingly an isosinglet. (Incidentally the hypercharge
operator in the Sakata model is Y = 2H2 + 2/3.) Moreover, the
operators V6 E ± 1 correspond to the raising and lowering opera-
tors T ± (see Eq. 2.7 or 2.14). The isosinglet is simply annihilated
by E±l.
From the commutation rules (2.16) and (2.17) it follows
im mediately that an isospin sub algebra composed ofthe operators

(2.32)

exists, independent of the sp·e cial features of the 3 representa-

tion. Setting a = f3 = 1 in Eq. (2.16) and (2.17) yields
(2.33)
34 INTRODUCTION TO UNITARY SYMMETRY

The connection with the cartesian components is as usual T ± =

Tl ± iT2 •
It is now easy to discover two other S U(2) suhalgehras. From
Eq. (2.31) we learn that the operators E±a increase the weight in
the direction of r(a). The equivalence of the three directions sug-
gests that V6E±2' V6E±3 are just as good raising and lowering
operators as V6E±l' Equating a and f3 in Eq. (2.16) leads to the
suspicion that the projection of H on the root vector r(a) gives
the third component of the" isospin " in the a direction. To verify
this we first define this projection

(2.34)
or explicitly

(2.35)

Taking the scalar product of r(a) with Eq. (2.17) (first replacing
a hyf3 in that equation) gives
(2.36)

1/3 while r(a)·r(f3) = -1(6 for a =1= f3, we find

[Jea,E±aJ (2.37)

[Jea,E ±oJ (2.38)

It is apparent from Eqs. (2.37) and (2.16) that 3Jea is the appro-
priately normalized third component of "isospin" in the a
direction.
In addition to the ordinary isospin operator T, we are led to
define U and V spin:
8 1/(:1 ): SYMMETRY IN STRONG INTERACTIONS 35

U± == V6E±3 U3 == 3Je3 (2.39)

V± == V6E±2 V3 == 3Je 2 (2.40)

Th e commutation rules are analogous to (2.33):

[U+,U_]=2U3 [U 3,U±] = ±U± (2.41)

[V+,V_] = 2V3 [V3,V±] = ± V± (2.42)
None of these subalgebras is an invariant s.ubalgebra (9) since the
operators in one sub algebra do not commute with those in another.
IThe algebra of S U(3) is said to be simple since there is no in-
variant sub algebra.]
The great practical utility of the S U(2) subgroups of S U(3)
having T, U, and V spin as generators has been emphasized by
Levinson, Lipkin, and Meshkov (11,12). We shall make extensive
li se of this formalism in order to exploit the familiar features of
S U( 2). The effect of the operators T ±' U ±' and V ± is indicated
in Fig. 2.4. The original commutation relations can be rewritten
in terms of the T, U, and V operators. Making the definitions

I ±a = V6 E ±a' I ± 1 = T ±' I ±2 = V ±' I ±3 = U ±

Iza = 3Jea (2.43)
we find that Eqs. (2.15)-(2.17) can be written as

[I±a,I±O] = +EaOyI'FY (2.44)

[I ± a,I '1'0] = 2o ao I za (2.45)
[Iz a,I ± 0] = ± 3r(a)· r(f3) I ± 0 (2.46)

Fig. 2.4. The operators T ±, U ±, and V ± change the eigenvalues of H by one

lattice constant (cf. Fig. 2.4) in the direction shown.
36 INTRODUCTION TO UNITARY SYMMETRY

Equation (2.46) can be simplified in the same way as Eq. (2.36).

Equation (2.44) can also be summarized by
[T+,U+l = V_ [U+,V+l = L [V+,T+l = U_ (2.47)
and the three relations obtained by Hermitian conjugation.
We next describe how one may construct for each of the S U(2)
subgroups linear combinations of HI and H2 which are scalars.
We have seen that Je a = r(o:)·H is the third component of the
T, U, or V spin. Let us construct three vectors r'(o:) by rotating
the three root vectors r(o:) counterclockwise by 90° (Fig. 2.5).
(r/(0:),r2'(0:)) = (- r2(0:),r1(0:)) (2.48)

r'(I) (0, ~)
=

r'(2) = (!, -~) r'(o:). r(o:) = 0 (2.49)

2 2V3

r'(3) = ( -~, - 2~3)

The quantities Je a' are defined by the equality
Je a ' =r'(o:)·H (2.50)
Taking the scalar product of r'(o:) with both sides of Eq. (2.17)
gives
(2.51)
In addition, all the Jea and Jep' commute for all 0: and f3. Therefore
Je1 ' commutes with T, Je2 ' commutes with V, and Je 3 ' commutes
with U. Figure 2.5 should be consulted to clarify the simple
geometrical significance of this result.

Fig. 2.5. The relation between the root vectors r(a) and the set r'(a) is shown.
SU(3): SYMMETRY IN STRONG INTERACTIONS 37

The significance of these operators can be understood as follows.

In detail, Eq. (2.50) is

'10 ' 1 H
<.Iv 1 = Y3 2

Jez' 1H 1 H (2.52)
=2 1 - 2Y3 2

Je3 ' = -GH1 + 2~3Hz)

It will be recalled that H2 was introduced in order to describe
the hypercharge quantum number. The specific relation is, how-
e ver, model dependent. In the Sakata model, where the com-
ponents of the basic triplet are identified witp p, n, and A, the
connection is
(Sakata model) (2.53)
while the identification of the eight baryons with the components
of the regular representation (see below) leads to
(Eightfold way) (2.54)
I n general a linear relation between Y and H2 is called for. for
hrevity we discuss the Je/ for the case (2.54).
Comparing Eqs. (2.52) and (2.54) gives the following relation:
(2.55)
Moreover, the electric charge is given by

-2Y3 Je3 '

(2.56)
The conserved quantity related to the 2-direction

(2.57)
is of course not independent of Q and Y. Thus far the" Z charge"
has not been found to lead to any simplifications beyond the
practice of describing states by means of Y and Q.
 38 INTRODUCTION TO UNITARY SYMMETRY

constant Y

'~~Q
Fig. 2.6. Lines of constant charge Q. hypercharge Y. and Z = Y - Q in the
eigenvalue lattice are shown.

Equation (2.51) now shows that T, V, and V commute with

Y, Q, and Z, respectively.
[T,Y] o
[U,Q] o (2.58)
[V,Z] o
Thus all members of a T-spin multiplet have the same hyper-
charge, all members of a V-spin multiplet have the same charge,
and all members of a V-spin multiplet have the same value of
Z = Q - Y. This result is indicated in Fig. 2.6. To emphasize the
formal similarity of :Ie i and Je;, the lines of constant T 3 , V 3 , and
V3 are shown in Fig. 2.7.

constant U3

canstant T 3
Fig. 2.7. Lines of constant T 3 • U 3 • and V3 in the eigenvalue lattice are shown.
SU(3): SYMMETRY IN STRONG INTERACTIONS 39

The explicit matrices representing Y, Q, and Z in the defining

three-dimensional representation are

~ ~(: iJ
0
1
y
0

~ ~(: iJ
0
-1 (2.59)
Q
0

~ ~(: ~)
0
-2
Z
0
It should be stressed that these forms follow from having imposed
Y = 2H2 on all representations. In particular, we note that if
basic triplets exist, they must have fractional charges and hyper-
charges [Gell-Mann (13); Zweig (14)]. Whether or not such curious
particles really exist, it is very useful for mathematical reasons to
work with such triplets. [A variety of triplet theories have been
discussed by Gursey, Lee, and Nauenberg (15).]

2.4 Reflection Symmetries of Weight Diagrams

It is useful to supplement the particular transformations
(raising and lowering operations) of the S U(2) subgroups discussed
in the previous section. From the usual theory of angular
momentum we know that a rotation by 7r about the 2-axis changes
the sign of the eigenvalue m of T3:
(2.60)
so that exp (i7r T 2 ) is a "reflection" operator for the eigenvalue
m. "Reflection" operators for the T, U, and V multiplets are
defined by
RT = e htT 2
Ru = ei1tU 2 (2.61)
Rv = e i1tV
2
40 INTRODUCTION TO UNITARY SYMMETRY

It should be remembered that because of the double-valued

ch aract er of the half-integral representations of S U(2), the above
operators are not ordinary reflection operators [explicitly, RT2
has eigenvalue (_1)2T]. Nevertheless, we shall refer to the
operators (2 .61) as reflection operators for brevity. The present
discussion follows that of de Swart (16); a somewhat different
treatment has been given by MacFarlane, Sudarshan, and Dulle-
mond (17).
Suppose we apply R v (v = T, U, or V) to an eigenfunction of
H,o/(m). We show now that Rvo/(m) is again an eigenfunction of
H, with eigenvalue obtained by reflection of m in the plane per-
pendicular to the root vectors r(a) appropriate to the T, U, or V
directions, respectively. Thus the existence of one state always
implies the existence of others obtained by reflection (except in
the special case in which the weight lies in the reflection plane).
In order to evaluate

(2.62)

and similar expressions one uses the standard operator identity:

1
eJ.ABe-J.A = B + A[A,B] + 2f A2 [A,[A,B]] +.. . (2.63)

Thus (2.62) is equal to

. (i7T)2 (i7T)3
H + ~7T[T2,Hl + 2T [T2 ,[T2,H]] + 3T" [T2 ,[T2,[T2 ,H]]] + .. .
(2.64)
The required commutators are

(2.65)

so that further commutators give Tl or T 3 • Hence the series

(2.64) sums to

H + ((i;( + (~r + ... )r(l) T3 - r(I)( 7T - ;; + ... ) (2.66)

= H - 2r(I)T3
5U(3): SYMMETRY IN STRONG INTERACTIONS 41

T he result can be written in a neater form upon noting that

r(1 )T3 equals 3r(1)·Hr(1). We thus obtain
R T HR T -1 = H - 2r(1)r(1)·H (2.67)
by introducing new root vectors of unit length. The analogous
results for U and V spin are
RyHRy -1 = H - 2T(2)r(2). H
(2.68)
RuHRu-1 = H - 2r(3)r(3) .H
From the derivation (cf. Eq. 2.66) we see that the above results
still hold if Rv ~ Rv -1. One finds finally
HRvrjJ(m) = Rv[H - 2r(v)r(v}· H]rjJ(m)
(2.69)
= [m - 2r(v)(T(v). m)]RvrjJ(m)
The geometrical construction for the reflection RT is shown in
Fig. 2.8. We have thus obtained the important result that the
weight diagram of any representation must have reflection symmetry
in the planes orthogonal to the root vectors. The weights of states of
a n irreducible representation connected by reflections are said to
be equivalent weights.
It is also useful to notice the transformation properties of the
raising and lowering operators:
R T T±RT -1 = - T'f
R T U±RT -1 = V'f (2.70)

reflection
plane

Fig. 2.8. The geometrical relations involved in reflecting the weight m in the
plane perpendicular to the T3 axis are illustrated.
42 INTRODUCTION TO UNITARY SYMMETRY

and six other relations obtained by cyclic permutation in T, U,

and V. Except for signs (which are not obvious since there is no
unique rotation in S U(2) corresponding to a given reflection) the
results (2.70) can be guessed from Fig. 2.4. The results (2.70) can
also be used to express a reflection in one plane in terms of com-
pound reflections in the two remaining planes:
Rv = R T R u -1R T -1
(2.71)
Ru = R T R v R T -1
Equations (2.71) should be supplemented by the relations
obtained by cyclic permutation in T, U, and V.

2.5 The Complex Conjugate Representation 3*

In Section 2.2 a set of eight traceless 3 X 3 Hermitian genera-
tors were introduced and used to define the Lie algebra of SU(3).
An important fact is that there is an independent set of eight
traceless 3 x 3 matrices satisfying this algebra. Previously we
considered the infinitesimal transformation of the triplet .p

.p -+ (1 + is a. . G).p (2.72)

In addition to the representation matrices (2.5) we may obtain a

representation of SU(3) by complex conjugation:

(2.73)

The question then arises whether the two representations are

equivalent. In Chapter 1 we found that for S U(2) the complex
conjugate representation is equivalent to the defining representa-
tion. This is not the case for SU(3), however. That is, there is no
3 X 3 matrix S which satisfies SAS -1 = - A*. One way to show
this is to write the most general S in the form aI + b· A, and use
the properties of the Ai to deduce that a = bi = O. More directly,
for As the existence of S such that SAsS -1 = - As would imply the
contradiction that ± As have the same eigenvalues.
Having established that the set of components .pic == (.plc)*,
(k = 1,2,3), generate a representation (3*) distinctly different
5U(3): SYMMETRY IN STRONG INTERACTIONS 43

fr om the representation 3, we discuss the question of constructing

the corresponding raising and lowering operators. From the set of
generators p . c = -p..* (explicitly A~.5.7 = A 2 •5 •7 ; Ai.3.4.6.8 =
- A1.3.4.6.8) one defines a set E ± c, H/ by Eqs. (2.18), with the Ai
replaced by Ai c , For instance, we have

E ±l C _ 1 (\ C •\ C) _ 1 (\ .\ ) -E~l (2.74)
- 2V6 III ± lll2 - 2V6 - I l l ± lll2

a nd of course, H/ = - H j ; the complete set is

(2.75)
This set of operators, while quite natural, is not unavoidable
n or even the most useful choice, as discussed below.
The weight diagram for the representation 3* is clearly obtained
b y inversion of that for 3, as shown in Fig. 2.9. For a general
irreducible unitary representation D of SU(3) with Hermitian
generators G the generators of the complex conjugate representa-
t ion can be taken to be - G*. D* will be equivalent to D only if
t he weight diagram is left unaltered by inversion.
We have noted in Chapter 1 that by a judicious choice of
p hases in the set of basis functions generating the complex
conjugate representation the usual Condon-ShortIey formulas for
the Clebsch-Gordan coefficients apply also to the complex con-
jugate basis. If we use the matrices (2.75) then this useful result
is not true for any of the S U(2) subgroups of S U(3). For the three

m2

-c
m3

m,

-c
m
-c
m, 2

Fig. 2.9. The weight diagram for the complex conjugate representation 3* is shown.
44 INTRODUCTION TO UNITARY SYMMETRY

states in 3(if;") the Condon-Shortley phase convention holds for

all the SU(2) subgroups:
T +if;2 = if;l T _if;l = if;2
U +if;3 = if;2 U _if;2 = if;3 (2.76)
V + if; 1 = if;3 V _if;3 = if;l
What are the analogous statements for the complex conjugate
representation if;", with components (if;,,)/ = a,,/~(x), where
~ = </>* is the space-time wave function [cf. if;" preceding Eq.
(2.9)]? The raising and lowering operators are

(2.77)
with analogous relations for Uc ± and Vc ±. We thus find

T + cif;l = -if;2 T _ cif;2 = -if;l

U + cif;2 = -if;3 U _ cif;3 -if;2 (2.78)
V + cif;3 = -if;I V _cif;l -if;3
so that, although the T ± c, U ± c, and V ± C act in the same direction
in the weight diagram as do T ±' U ±' and V ±' the convenient
S U(2) phase convention has been lost.
It is clear, however, that the Condon-Shortley rules cannot be
expected to hold since 3* contains the complex conjugate isotopic
doublet (if;1,if;2)' and the signs of this complex conjugate doublet
were not chosen properly relative to the doublet (if;1,if;2) in 3 (see
Eq. 1.29). Thus we are led to use a different set of basis functions
(or field operators, according to the occasion) Xl' X2' and X3'
related to the basis if;1, if;2, and if;3 of the defining representation
3 as follows:
Xl = _(if;l)*
X2 = (if;2)* (2.79)
X3 = (if;3)*
(When if; is a field operator the * operation includes, of course,
Hermitian conjugation.)
SU(3): S YMMETRY IN STRONG INTERACTIONS 45

Let Uk he the elements of the matrix representation 3

o/k' = Uklo/I (2.80)

~o that the o/k (o/k)* transform as
(2.81)

Then the components of X transform as

(2.82)

Specializing to infinitesimal transformations, the change U l2 * -+

- U I2 *, U13 * -+ - U 13 * requires Ale -+ -Ale, A2 c -+ -A 2 e, A4 e-+
- A4 c, and A5 c -+ - A5 c. Thus the modified complex conjugate
representation has generators
GI e G2 e = -G 2
G6 e = -G 6 GS e -G s
(2.83)
with corresponding raising and lowering operators

(2.84)
Hi" = -Hi

Identifying T ± c = V6 E ± I c, etc., we see that the Condon-

Shortley phases are secured for the matrices pf T and V spin hut
not for U spin. Nothing can he done ahout this, because of the
commutation rules. One could transfer the minus signs to V spin
hy changing the minus sign in Eq. (2.79) from Xl to X2' Such a
convention would seem meritorious since U spin has a more
direct physical meaning than V spin, at present. However, tahles
of SU(3) Clehsch-Gordan coefficients have heen prepared (16,
18- 19) using the conventions (2.84), so we shall adhere to this
choice.
Suppose we have an irreducible r epresentation D = exp (iCl' G)
46 INTRODUCTION TO UNITARY SYMMETRY

of dimension N, with basis functions cp(N; T, T 3 , Y). The complex

conjugate representation D* has the generators (2.75) rather than
the desired convention given by (2.84). Thus a further unitary
transformation S D* S -1 is required to obtain (2.84). To discover
an explicit form for S, first note that complex conjugation results
in the generators

(2.85)
He = -H

The signs shown in (2.84) demand that the matrix S have the
following properties:

SV±S-l = - V±
(2.86)
SHS-1 = H

The last equality requires that S depend on H. The next to last

equality further shows that S should be a function of Q, the only
combination of H j commuting with U. Now it is easy to see that
an appropriate unitary S is

S = (-l)Q == exp (i7T'Q) (2.87)

since T ± and V ± change the charge by one unit. Here we suppose

that Q is always given by T3 + t Y, with Y given by (2.54),
whether or not Q is integral, in order to obtain a single formula
for all representations. If this convention is adhered to, then the
elements of S are not, in general, real. For example, in the Sakata
model Y had to be defined by Eq. (2.53) 2H2 + i to obtain
charges (1,0,0) for (p,n,A). With our convention (2.87) the three
components of the triplet have "charges" (2/3, -1/3, -1/3) .
(Further properties of such hypothetical triplets are discussed in
Section 3.4.) To obtain a real S for 3 we then write S -;. S' = TJS
with TJ = exp (i7T'/3) so S = exp [i7T'(Q + t)]. The diagonal
elements of S are (-1,1,1). If we take the complex conjugate
of the triplet (2.72) and multiply by S we obtain the triplet 3* of
Eq. (2.79) whose generators have the desired form (2.84). Thus we
have the relations
SU(3): SYMMETRY IN STRONG INTERACTIONS • 47

while for the regular representation 8 and all representations N

contained in 8 x 8 x ... x 8 (cf. Chapters 3 and 4)
,p(N*;TT3Y) = einQ,p*(N;T,- T 3,- Y) (2.89)
We can obtain a deeper insight into the situation revealed by
Eqs. (2.88) and (2.89) by cla;;sifying the representations of S U(3)
according to their triality.
For orientation, notice that in S U(2) we can classify the repre-
sentations according to whether 2j is even or odd. Consider the
state of highest weight. Under a rotation by 211' the state is
multiplied by e2iJti so that the representations fall into two classes
according to whether e2iJti = ± 1. When we reduce direct products
we find that odd x odd = even, even x odd = odd, and even x
even = even. Alternatively, we can associate the number d = 1
with 2j = odd and d = 0 with 2j = even. [That is, d = 2j
(mod 2) .] Then when multiplying representations we have d 1 +
d 2 = d 3 (mod 2). In terms of basic doublets we say that odd
powers ,p,,p x ,p x ,p,. .. have .. duality" 1 and even powers
,p x ,p,,p x ,p x ,p x ,p, ... , have duality O. Thus d is given by
eind , the eigenvalue of e2inl z applied to the state of highest weight.
In S U(3) the representations fall into three classes of the types·
contained in the direct products
(3)3n, (3)3n+l, (3)3n+2; n = 0,1,2, ... (2.90)
The triality t is defined (mod 3) by writing the exponents occurring
in Eq. (2.90) in the form (3n + t). Thus the representation 3 has
t = 1, 3 x 3 = 6 + 3* has t = 2, and 3 x 3 x 3 = 10 + 8 +
8 + I has t = O. The three examples given suggest that 7J =
exp (i11'tf3). It is in fact clear that this choice renders the phase
factor real in general:
S = exp [i11'(Q + tf3)]
(2.91)

• The following discussion makes use of results obtained in Chapters 3 and 4.

48 INTRODUCTION TO UNITAH Y ~Y MMETRY

This expression generalizes de Swart's res ull for lh c eightfold

way (t = 0).
If we decompose the produce of representations having trialities
tl and t2 the resulting irreducible representations have triality
t3 = tl + t2 (mod 3) in analogy to the SU(2) case. For a proof of
this statement see references 21 and 22. For our purposes this
result is nearly obvious from the preceding examples.
Equations (2.88) and (2.89) illustrate how one can compute t.
Apply the operator exp (27Ti Y) to the state of highest weight in
the reducible representation. The eigenvalue exp (27Tit/3) defines
the triality (mod 3). As an illustration, note that for 3, exp (27Ti Y)
gives exp (27Ti/3). 3 x 3 then has eigenvalue exp (47Ti/3) for its
component of highest weight. Solving for t, we find t =
-1,2,5, ... , i.e., t = 2 (mod 3).
Apparently the only representations realized in nature have
triality zero. There seems to be no deep understanding of this fact.
A different approach to the triality concept has been given by
Okubo et al. (23), who point out that the occurrence of fractional
charge and hypercharge can be avoided if the group is not S U(3)
but U(3). Charge and hypercharge are made integral by the
redefinitions
Q' = Q + It y' = y + it (2.92)
(cf. Eq. 2.91; for the integral character of Y' recall the relation
Q = T3 + t Y). Recall that in terms of the traceless generators
A/ of Eq. (2.27) Q and Yare given by
Q= All
(2.93)
Y = All + A22

Defining the generators B/ by

B/ = A/' + la/I;
(2.94)
t = Bll + B22 + B33
where t may be shown (23) to have integral eigenvalues, we have
Q' = Bll
(2.95)
8U(3): SYMMETRY IN STRONG INTERACTIONS 49

while T3 is i(A/ - A22) or l(B/ - B 22). The generators Bvu are

simply those of U(3). It will be noted that the operator t is truly
additive and not defined merely mod 3.

2.6 The Regular Representation; The F and D Matrices

The traceless 3 x 3 Hermitian generators fA; were used in Eqs.
(2.22) and (2.23) to introduce the set of real coefficients Jabc and
d abc (a,b,c = 1, ... ,8). These coefficients have a number of useful
properties which we now describe. First, we note that the hjk of
S U(3) are analogous to the eWe of S U(2). Recall that from the
e;jk one can form a set of 3 x 3 Hermitian matrices (t;)jk =
- ie;jk that give a representation of the algebra of S U(2), as
indicated in Eqs. (1.35)-(1.37). Moreover, from the basic spinor
if; and the generators iT, one can form a set of basis functions
if;tiTif; transforming according to the regular representation of
S U(2). Analogous results hold for S U(3).
We introduce the idea of the regular representation in a context
more general than that of SU(3). Suppose we have a set of N
independent matrices Ma (a = 1, ... N) subject to the following
commutator structure

(2.96)

(a,b,c = 1, .. . ,N). The structure constants Cabc specify the charac-

teristics of the algebra exhibited in this equation. When the Ma
are generators of some continuous group of transformations, the
Cabc completely determine the local properties of that group. The
structure constants Cabc are antisymmetric in the lower indices:

(2.97)

We now show that the C's themselves can be used to construct a

set of N x N matrices satisfying the algebra of the M's. This set
of N matrices built from the structure constants is called the
regular representation of the algebra defined by Eq. (2.26).
The proof hinges on Jacobi's identity:

o (2.98)
50 INTRODUCTION TO UNITARY SYMMETRY

Using Eq. (2.96) to evaluate the commutators and making use of

the independence of the M j gives

(2.99)

If we now multiply by ( -1), permute c and a in the second term

and transfer the third term to the right-hand side, we obtain

(2.100)

If we now define a set of N matrices Ra by

(Rahc == -Cabe (2.101)

Eq. (2.100) can be rewritten as

(Ra)em( Rb)mn - (Rb)em( Ra)mn = Cab m( Rm)en (2.102)

showing that a particular instance of Eq. (2.96) is Ma = R a, with

Ra defined in Eq. (2.101).
We apply this result to the structure constants iJabe of SU(3).
Since the Jabe are real and completely antisymmetric, the 8 x 8
matrices Fa defined by

(Fahe == (Fa)be == - iJabe (2.103)

are Hermitian. They naturally satisfy (2.26):

[Fa,Fb] = iJabeFe = -FabeFe (2.104)

(The writing of superscripts and subscripts on these matrices is

only a matter of notational convenience.)
The" vector" matrix F is the analog of the matrix t encountered
in S U(2), the latter generating the rotations of the ordinary
position vector r [see Eq. (1.38)]. This analogy leads us to define
a vector operator in S U(3) as follows. Suppose we have a set (of
operators or N x N matrices) Va (a = 1, ... ,8) whose commu-
tators with the generators Ge (operators or N x N matrices)
satisfy
(2.105)

Then V is said to be a vector operator. This definition shows that

SU(3): SYMMETRY IN STRONG INTERACTIONS 51

under the transformation (9" = exp (iex· G), V transforms accord-

ing to the regular representation; we consider infinitesimal
transformations:

(2.106)
= [8ij + i8ex· Fij] Vj
i.e., V ---7 exp (iex·F)V in analogy to r ---7 exp (iex·t)r under spatial
rotations.
The 3 x 3 generators Ai can be used as Clebsch-Gordan
coefficients to construct a set of functions (or operators) trans-
forming as the regular representation. Suppose we have triplets
</>' rjJ transforming as </> ---7 (1 + iSex· ")../2)</>. Then experience with
S U(2) suggests that the eight quantities

(2.107)
constitute a basis for the regular representation with matrices F.
The anticlimactic details follow:

V k ---7 rjJt(l - i8ex · ")../2)Ak(1 + i8ex· ")../2)</>

= V k + i8a rjJt[ Ak,AM/2
1 (2.108)
= (8 km + i8a l Fkml) Vm
The V k can be written in the form

(2.109)

which displays the role of the (Ak)ij as Clebsch-Gordan coefficients,

which combine 3 and 3* in the appropriate way to obtain the
eight-dimensional regular representation. (The one-dimensional
representation is found by constructing the inner product rjJt</>.)
The above result can be generalized considerably. Consider
"vectors" P, cP transforming via 1 + i8ex· G. Then the scalar
product of a vector operator V automatically transforms as

ptvcp -7 (1 + i8ex·F)ptVCP (2.110)

52 INTRODUCTION TO UNITARY SYMMETRY

An especially important case occurs when V is just F, and lJ' and

cP are eight-component objects transforming as V k in Eq. (2.108).
In fact, the result holds whether or not the components of lJ', cP
have their complex conjugate taken, as we are instructed to do
in Eq. (2.110). This is true because the regular representation is
equivalent to its complex conjugate. The particular regular
representation D = exp (icx· F) is even identical to its complex
conjugate since D* = exp ( - icx· F*) = D, because F* = - F.
Thus consider the" octets" A a, Ba (a = 1, ... ,8) transforming
as
Aa -7 (Sab + iSacFabC)Ab
(2.111)
Ba -7 (Sab + iSacFabC)Bb
The linear combination V/
(2.112)
also transforms as
(2.113)

Therefore the coefficients Jabc are Clebsch-Gordan coefficients in

the r egular representation.
We now recall the symmetric real coefficients dabc ' Do these
have any useful properties? From them we can form the real,
symmetric 8 X 8 Hermitian matrices Da:
(2.114)

We now learn that D is a vector operator in the regular repre-

sentation, i.e.,
(2.115)

The proof of this and similar identities given below is based on a

method described by Cutkosky (24). The appropriate D's and F's
can be generated by the identity
(2.116)

One readily finds

Db/Fa/ (2.117)
SU(3): SYMMETRY IN STRONG INTERACTIONS 53

which can be written in the form (2.115) by permuting various

indices.
According to the previous discussion the quantities

(2.118)

where Aa and Bb transform as indicated in Eq. (2.111) also give

a basis for the regular representation:
VD-;.-(1 + ioa.F)VD (2.119)

In the usual manner, the "dot product" of one vector with

another is a scalar; whether these " vectors" are functions or
vector operators is irrelevant. Thus the quantities F2, D2, and
F · D comm~te with the F j , and by Schur's Lemma are multiples
of the unit matrix in the regular representation. In particular, we
can show that
Tr FaF b = F20ab

Tr DaDb = j)2oab (2.120)

Tr FaDb = 0
where F2 and ])2 are the eigenvalues of the matrices F2 and D2.
(Later we show that F2 = 3 and ])2 = 5/3.) The proof consists of
rewriting
Tr FaF b = (Fa)jj( F b)jj = (F2)ab = F20ab
(2.121)
Tr DaDb = (DaLj(Db)jj = (D2)ab = j)2oab

Tr Fa Db = (Fa)j,( D b)ji = 0 (2.122)

the latter following from the symmetry of D and the anti-

symmetry of F in the ij indices.
The following trace properties are also useful:
Tr pa = Tr Da = 0 (2.123)

The first of these relations is trivial. To prove the other, set

a = b in the anticommutator {A'a,A b } and sum on a. This gives

2)..2 = Jill + 2 L DaacAc

a
(2.124)
54 INTRODUCTION TO UNITARY SYMMETRY

Since ),.2 is proportional to I (it commutes with the A,), we see that

Ac Tr DC = kl (2.125)

where k is a constant. Multiplying by Aa and taking the trace gives

2 Tr Da = k Tr Aa = O.
In addition to the commutators [Fa,FbJ, [Fa,DbJ it is sometimes
useful to know [Da,DbJ. To obtain this one can begin from the
identity
(2.126)

Proceeding in a now familiar manner, one obtains

(2.127)

References
1. R. H. Dalitz, Ann. Rev. Nucl. Sci., 13, 339, (1963).
2. A. H. Rosenfeld, A. Barbaro·Caltieri, W. H. Barkas, P. L. Bastien, J. Kirz,
and M. Roos, Rev. Mod .. Phys., 36, 977 (1964).
3. P. Roman, Theory of Elementary Particles, North Holland Pub!. Co., Amster·
dam (1960), Chap. 12.
4. S. Sakata, Progr. Theoret. Phys., 16, 686 (1956).
5. M. Ikeda, S. Ogawa, and Y. Ohnuki, Progr. Theoret. Phys. (Kyoto), 22, 715
(1959); Suppl.19, 44 (1961).
6. M. Cell-Mann, Phys. Rev., 125, 1067 (1962).
7. Y. Ne'eman, Nucl. Phys., 26, 222 (1961).
8. R. Behrends, J. DreitIein, C. Fronsdal, and B. W. Lee, Rev. Mod. Phys., 34,
1 (1962).
9. M. Hamermesh, Group Theory, Addison-Wesley, Reading, Mass., 1962.
10. S. Okubo, Progr. Theoret. Phys., 27, 949 (1962).
11. C. A. Levinson, H. J. Lipkin, and S. Meshkov, Phys. Rev. Letters, 1, 44 (1962).
12. H. J. Lipkin, Lie Groups for Pedestrians, Interscience, New York, 1965.
13. M. Cell-Mann, Phys. Rev. Letters, 8,214 (1964).
14. C. Zweig, CERN Rep. No. 8182/Th. 401 and 8419/Th. 412 (1964).
15. F. Cursey, T. D. Lee, and M. Nauenberg, Phys. Rev., 135, B467 (1964).
16. J. J . de Swart, Rev. Mod. Phys., 35, 916 (1963).
17. A. J. MacFarlane, E. C. C. Sudarshan, and C. Dullemond, Nuovo eimento, 30,
845 (1963).
18. P. Tarjanne, Carnegie Tech. Rept., NYO-9290, 9290A.
19. P. McNamee, and F. Chilton, Rev. Mod. Phys., 36, 1005 (1964).
20. L. C. Biedenharn and E. C. Fowler, unpublished.
SUP) : S YMMETRY IN STRONG INTERACTIONS 55
21. G. E. Baird and L. C. Biedenharn, Proceedings of the Coral Gables Conference
on Symmetry Principles at High Energy, ed., B. Kursunoglu, W . H. Freeman
and Co., San Francisco, Calif., 1964.
22. C. R. Hagen and A. J. Macfarlane, J. Math. Phys., 5, 1335 (1964).
23. S. Okubo, C. Ryan, and R. E. Marshak, Nuovo Cimento, 34, 759 (1964).
21.. R. E. Cutkosky, Ann. Phys., 23, 415 (1963).
 Chapter 3

Construction of Representations of 8U(3)

3.1 Introduction
The irreducible representations of S U(3) are found by decom-
posing Kronecker products of 3 and 3* representations into
irreducible constituents. Consider a product state composed of p
functions if/' transforming like 3, and q functions <Pk transforming
like 3*:

The labels 1,2, ... , p + q can be regarded as independent

coordinates. This set of 3P +q functions is a (reducible) tensor
under SU(3) transformations; we shall often denote the (reducible)
representation generated by (3.1) by

(3)P x (3? (3.2)

For later reference, we note that the highest weight contained in
the set (3.1) is·

M(p,q) = pmH
+ q' -
+ qmH* = ( p2V3 p - q)
6- (3.3)

which occurs for a = f3 = ... = v = 1, a = b = ... = k = 2,

where mH and mH* are the highest weights of the representations
3 and 3*, respectively. Note that this state, which is the analogue
of the" stretched" configuration occurring in the vector addition
of angular momentum, is non degenerate. The irreducible repre-

• To find the state of highest weight, first find the states with greatest H , ;
among these choose the one with greatest Hz.
56
CONSTRUCTION OF REPRESENTATIONS OF 8U(3) 57

sentation to which this function belongs is called D(p,q), or

simply (p,q). When no confusion can arise, the representations are
oft en referred to by their dimensionality. In the D(p,q) notation
the identity representation 1 is D(O,O), 3 is by convention D(I,O),
and 3* is D(O,I). The importance of this notation is revealed by
the following fact: by letting p and q range over the nonnegative
integers the states of highest weight (3.3) in the product (3.2) run
through all the irreducible representations of S U(3).
As discussed in Chapter 2, the conservation of isospin (T) and
hypercharge (Y) corresponds to invariance under a particular
subgroup SU(2h x U(I)y of SU(3) . The third component of
isospin Ta was identified with V3H 2 • There is more freedom in
the assignment of Y but we shall adopt, subject to further dis-
cussion, the eightfold way assignment Y = 2H2 [see Eqs. (2.53)
and (2.54)]. The state of highest weight in D(p,q) has T = Ta for
that state. Hence we find that the state of highest weight in the
representation D(p,q) has isospin and hypercharge given by

TH = t(p + q), YH = t (p - q) (3.4)

the latter relation being specific to the eightfold way. In this

scheme the only permitted representations having integral Y have
p and q connected by p - q = 3n, n an integer. Hence the eight-
fold way excludes a large class of representations of S U(3).*
There are two principal means of decomposing (3.2) into its
irreducible constituents. One of these methods is a simple
generalization of the raising-lowering operator technique so
familiar in SU(2). This procedure is elementary and provides a
detailed understanding of the specific properties of the various
representations (1-4). The second method, which involves the use
of tensor analysis, is speedier but requires more subtle mathe-
matical theorems for its justification (5-7). Therefore we shall

• The group of the eightfold way is the factor group SU(3)/Z3' where Z3 is the
center of SU(3). In exact analogy, the integralj (single-valued) representations of
SU(2) correspond to SU(2)/Z2' In the latter case Z2 is composed of ± I, I the
identity matrix.
58 INTRODUCTION TO UNITARY SYMMETRY

describe the raising-lowering operator technique in some detail in

Section 3.2. In Section 4.6 we shall describe, without complete
proofs, how to use the tensor methods and the connection between
the two approaches.

3.2 The Raising and Lowering Operator Technique

Under an infinitesimal transformation the tensor (3.1) trans-

forms as
rpa'Il' 0 0 0 V' _ Ta'Il' 00 0 V'; abo 0 0 k rpallo 0 0 v (3.5)
a'b' . .. k' - a8 . .. v; a' b' ... k' ab • . • k

where the infinitesimal transformation coefficients are

(3.6)

Here the distinction between upper and lower indices has been
introduced to agree with the conventions of the tensor methods
discussed below. The label a in Aba stands for the row index
while b is the column index for the matrix A. More succinctly, the
transformation T = 1 + ilia. G where 0

1( p p+q)
G = 2" j~ )'(i) + j=~l )'C(i) (3.7)

As discussed in Chapter 2, it is convenient to introduce a

special phase convention in order that the Condon-Shortley phase
convention may be used for two of the SU(2) subgroups. Using
this convention [cf. Eq. (2.79)], we convert (3.7) to the analogous
relations for the raising and lowering operators:
CONSTRUCTION OF REPRESENTATIONS OF 8U(3) 59

T± - V6 [~E±1(i) + j:~1 E±lC(i)],

T± V6 [j~ E±l(i) + j:~1 EU(i)];

U± - V6 [~ E±3(i) + j:~1 E±3C(i)],
U± V6 [j~ E±3(i) - j:~q1' E T3 (i)];

[~ E±2(i) + j~~1 E±2C]'

(3.8)
v± - V6

V± V6 [i~ E±2(i) + j:~1 E'F2(i)];

ii=1 H(i) + Pi
q
H = H"(i),
i=p+1
p+q
H = L H(i) - .L
P

i= 1 j=p+1 H(i).

We have presented the generators in mixed form (T, U, V, and H)

Mince we find this convenient in applications. The last of Eqs.
(3.8) exhibits the familiar feature that the eigenvalues of the
Jiagonal operators (H) add in product states. Before giving a
general discussion we shall illustrate the essential features of the
procedure by a few examples. Our treatment is patterned after
lhat described by Gasiorowicz (1), with some changes in con-
ventions used.
First we consider the representation 3 x 3, whose basis may be
taken as
ifJa(J = ifJa(l)ifJ(J(2) (3.9)
the transforml:\tion matrix being exp [till' (A(I) + A(2))). The
composite weight [corresponding to H = H(I) + H(2)) for the
state ifJaa is hence
(3.10)
60 INTRODUCTION TO UNITARY SYMMETRY

. .
xp.
3 2 I 3 2

·6
(0) (b) (e)

Fig. 3.1. Fig. (3 .1a) illustrates the vector addition of the weight diagram
3 X 3; the number of dots at a site indicates the degeneracy at each site for th
corresponding reducible representation. Figs. (3.1b) and (3.1c) show how 3 X
-breaks up into the irreducible representations 6 and 3·.

The allowed weights can be found by "adding the weigh

diagrams" of the two representations (Fig. 3.1). The dots indicat
the number of states having the same weight. The state o
highest weight 1jJ1l is nondegenerate and has weight 2m1 =
(1/V3,t) corresponding to T3 = T = 1 and Y = 1 [cf. Eq
(3.4)]. The remaining members of the representation to which
1jJ1l == lJ\ belongs are found by applying T ±' U ±' V ±. Since P
is the T3 = 1 member of a T-spin triplet we can obtain the
remaining members of this triplet by applying T _ :

P 2 == Jz: T- P1 = J2 (1jJ2(1)1jJ1(2) + IjJl(1)1jJ2(2))

(3.1l
P 3 == ~ T _ P 2 = 1jJ2(1)1jJ2(2).

The various sites in the weight diagram are labeled in order a

shown in Fig. 3.1h. To obtain further members of the representa
tion we can, for instance, apply V + to P 1 , which -state is the
V3 = -1 member of a V-spin triplet. Thus we find

which is the T3 = 1/2 member of a T = 1/2 doublet, the accom

panying T3 = -1/2 state being
CONS TRUCTION OF REPRESENTATIONS OF SU(3) 61

(3.13)

The V3 = 1 state 'P6 , which clearly has T = 0, is given by

(3.14)

No new states can be generated by further operations in the group.

This is most easily seen by observing that the operators T ±' U ±'
V ± are · essentially permutation operators, and the functions
IfIi (i = 1-6) exhaust the symmetric functions that can be formed

from the nine states ifP(1)if/3(2).

The six-dimensional irreducible representation generated by
IPi (i = 1- 6) is called D(2,0) or just 6. It is composed of a triplet,
doublet, and singlet in any of the SU(2) subgroups (see Fig. 3.1b).
The remaining three states independent of 'P2 , 'P4 , 'Ps and sites
2, 4, 5 form the basis for a three-dimensional representation
iHomorphic to 3*. At site 2 we form a function orthogonal to and
I here fore independent of 'P2 :

(3.15)

The sign has been chosen by the convention that the plus goes
with the highest T3 state of particle number one. ct>3 clearly has
T = 0; it is also the V3 = - 1/2 member of a V-spin doublet. The
other member of this doublet,

(3.16)

is orthogonal to 'Ps. ct>l is the T3 = -1/2 member of aT-spin

doublet. The state ct>2 given by

(3.17)

has T = T3 = 1/2, and is orthogonal to 'P4 • The states ct>,

(i = 1- 3) have the same weights as the CPi (i = 1-3) used to define
62 INTRODUCTION TO UNITARY SYMMETRY

the representation 3*, and are transformed into each other in the
same manner as the <Pi when the group operations are performed
(see Fig. 3.1c). The numbering of states within an irreducible
representation has been chosen in accord with the convention o
de Swart (1).
The resultant decomposition
3 x 3 = 6 + 3·
or (3.18)
D(l,O) x D(l,O) = D(2,O) + D(O,l)
is seen to be accomplished simply by decomposing the second rank
tensor ljiii into its symmetric and antisymmetric parts:

ljilXB = ljirL(1)ljiO(2) = ~ (ljia(1)ljiB(2) + ljiP(1)ljilX(2))

V2 V2
(3.19)
_1 (ljia(1)ljiP(2) - ljiO(1)ljirL(2))
+ V2 V2
Next consider the more interesting case of 3 x 3*, whose basis
is
ljipa = ljia(1)<pp(2)
Addition of the weight diagrams of 3 and 3* leads to the allowed
weights shown in Fig. 3.2a. It will be noted that the only de-
generacy occurs at the center of the weight diagram. The highes
weight is M(l,l) = (1/V3,O). The state lji2 1 therefore has T =
T3 = 1, and Y = O. The remaining members of the T-spin

2· I·

r (a)
5·

8·
6·· 4

(b)
7·
·3 +

Fig. 3.2. Fig. (3.2a) illustrates the vector addition of weight diagrams in the
direct product 3 X 3*. Figs. (3.2b) and (3.2c) show how 3 x 3* breaks up into
(c)

the irreducible representations 8 and 1.

CONSTRUCTION OF REPRESENTATIONS OF 5 U( 3) 63

I riplet are found by applying T _. The triplet T = 1, Y = 0 has

I he wave functions (see Fig. 3.2b for the numbering convention)

P 3 = rplc/>2

P4 = ~2 (rp 2c/>2 + rplc/>l) Y = 0 (3.20)

'I' n is a member of a V-spin doublet with V3 = + 1/2. The V3 =

1/2 member located at site 2 in Fig. 3.2b is found by applying
" , with the result P 2 = rp 2c/>3. This state, in turn, is the T 3 =
1/2 member of a T-spin doublet. Applying T + to P 2 then gives
'1'1:
P l = rp l c/>3
(3.21)
P 2 = rp 2c/>3
In the same manner, we note that P 3 is the V3 = -1/2 member
"f another V-spin doublet, whose companion is P s = T _ P 7 :

P7 = rp3c/>2
Y= -1 (3.22)
P s = rp3c/>l

"'I. us far we have shown that all six states on the periphery belong
I" the same irreducible representation. It remains to investigate
I hc threefold degeneracy at the center of the weight diagram. We
huve seen that T + or T _ applied to P s and P 3 lead to a state at
I he center, P 4 • The three independent functions with zero weight

IIrc rplc/>l' rp 2c/>2' and rp3c/>3. The combination P 4 = (rplc/>l + rp 2c/>2)1

v2" has been shown to belong to the same representation as the
Mix states on the periphery of the weight diagram. Next note
I hat P l is the T3 = -1 member of a V-spin triplet. The state

IPa =V + P1/Yz is accordingly the V3 = 0 member of the triplet:

V3 = 0 (3.23)
64 INTRODUCTION TO UNITARY SYMMETRY

(lJ'a is also obtained by applying V _ to IJ's.) Similarly 'P2 is the

U 3 = + 1 member of a U-spin triplet. Applying U _ leads to IJ'p:

U3 = 0 (3.24)

The minus sign can be traced to the sign conventions discussed in

Section 2.5.
The functions lJ'a' 'Pp, and 'P4 are not independent. Extracting
the component of lJ'a parallel to 1J'4 defines a new function Xa
orthogonal to and independent of 1J'4:

(3.25)

A similar projection of 'Pp yields XP = X«:

(3.26)
XP = Xa
Normalizing Xa and defining a new function 'Ps we find

IJ's == ~3 Xa = ~6 (ljJlCPl - IjJ2CP2 + 2ljJ3CP3) (3.27)

V3 'IfS
IJ'a --2"IIJ'4+2 (3.28)

(3.29)

The state 'Ps , which was obtained by requiring it to be orthogonal

to the T = 1, T3 = 0 state 1J'4' has T = O. Note that (3.28) and
(3.29) show how to express the central components of the U and
V-spin triplets in terms of components of the T-spin triplet. This
connection is very useful in applications.
CO NSTRUCTION OF REPRESENTATIONS OF SU(3) 65

We have now generated (by systematic application of all the

gro up operations to the state of highest weight) an irreducible
representation with eight basis functions 0/1 (i = 1-8). This
eight-dimensional representation will be shown to be the regular
representation of S U(3). It plays a central role in physical appli-
c:ations, as discussed in Section 3.4. The isospin-hypercharge
eomposition of this representation [8; D(I,I)] is seen to coincide
exactly with that of the collections of particles in Fig. 2.1, except
for the extra, neutral vector meson which will be the subject of
,mbsequent dispussion. This correlation justifies the physical cor-
respondence Y = 2H2 of Eq. (2.54).
One independent function remains at the center of the weight
diagram. Since it is inaccessible by the route of applying S U(3)
operations to members of the representation 8, it belongs by
definition to an inequivalent representation. The function
orthogonal to 0/4 and 0/6 is

T = Y = 0 (3.30)

This S U(3) scalar is simply the contraction of I{Ii and cPf taking
into account the extra minus sign introduced in the cP triplet to
enforce the standard phase convention. (With the extra sign the
metric tensor is no longer the Kronecker delta.)
In Table 3.1 the basis functions for the various subgroups
S U(2 ) x U(I) have been given for the representation 8. The signs
have been chosen to maintain the standard phase conventions in
all three S U(2) subgroups. In parentheses the states have been
expressed in terms of the usual baryon states, iabeled by the
quantum numbers of SU(2h x U(I)y appropriate to physical
particles.
It is frequently useful to have at hand explicit expressions for
the matrix elements of the generators in the representation 8.
These are easily found by referring to the weight diagram, which
reveals the nonzero matrix elements, and using the results of the
preceding calculation of the basis functions . Rather than write
out 8 x 8 matrices having few nonzero entries, we employ an
 TABLE 3.1
Composition of the Octet Wave Functions in Terms of the Constituents 3(.p1) and 3·(</>/)' Basis Functions May Be Labeled by the
Quantum Numbers of any of the SU(2) X U(1) Subgroups. U- and V-Spin Multiplets Are Labeled in Terms ofthe T-Spin States,
the Latter Labeled by the Familiar Baryons.

SU(2h X U(l)y SU(2)v X U(1)z SU(2)u X U(1)Q

0
T = 1/2 {'I'l = .p'</>a (p) V = 1/2 {'I'7 (8 ) U = 1/2 {'I'l (p)
Y = 1 '1'2 = .p2</>a (n) Z = 1 'I'a (E+) Q=1 - 'I'a (-E+)

a-
a- T=1
y=o
'1'3 = .p'</>2

fl" ~ \,<>".. )(E; H·••

(E +)
V=1
z=o
C''I'a
(8-)
(~;E"+ ~3A)
V=1
Q=O
C' (0)
'1'6 (-tEO+ -; A)
v-
'I's = .p </>1 (E ) '1', (p) - '1'7 (_8 0 )

T = Y = 0 '1'0 = :'6 W</>, - .p2</>2 + 2.pa</>a) (A) V= Z=O

{'I'a' = 2vi '1'4 - t'l'o
= (~3 EO - tA) U = Q= 0
{
'1'6' =

=
2
V3
'1'4

(~3 EO + tA)
+ t'l'o

T = 1/2 {'I'7 = .pa</>2 (8 0 ) V = 1/2 {'I's (E-) U = 1/2 {'I's (:~)

Y = -1 '1'0 = .pa</>, (8-) Z = -1 '1'2 (n) Q = -1 '1'. (.:\ )
CONSTRUCTION OF REPRESENTATIONS OF SU(3) 67
"uter product notation, using Ii > for Pi (i = 1-8). The following
results are easily found for the generators T _, u _,
V _, T 3 , and
Y. T +, U +, and V + are found hy transposition of T _, U _, and
V _. E ±a and H are found hy inserting the appropriate factors.

12><11+ 18><71 + Y214><31 + Y215><41;

12><51 + 13><71 + Y2Ia><81 + Y211><al,
1 1
12><51 + 13><71 + Y2 14><81 + Y2 11 ><41
+ Y!16><81 + Y!11><61;
1 1
-1 3><11 + 18><51 + Y3 1.8><21 - Y317><.8I, (3.31)
1 1
U_ = -1 3><11 + 18><51 - 2Y3 14><21 + 2Y317><41
+ 116><21 - tI 7><61
T3 = t,l,O, -1,0,1, - 1]
diag [t, -

Y = diag [1,1,0,0,0,0, -1, -1]

The states numhered 1 through 8 correspond respectively to the
particle states p, n, E+, };O, E-, 11°, EO, E-. The memhers of
the T-spin multiplets just named are related to each other hy
the standard phase convention.
The important ten-dimensional representation arises in the
reduction 3 X 3 X 3. The latter representation has the hasis
g'X(I) gO(2)e(3), where g transforms as 3. Since 3 X 3 = 6 + 3*,
the product ga(l)gD(2) can he expressed in terms of the hasis
functions of the irreducihle representations 6 and 3*. The prohlem
thus reduces to considering separately 3 x 3* and 3 x 6:

3 x 3 x 3 = (6 + 3*) x 3 = 6 x 3 + 3* x 3 (3.32)
68 INTRODUCTION TO UNITARY SYMMETRY

(0)

(c)

Fig. 3.3. Fig. (3.3a) illustrates the vector addition of weight diagrams in th
direct product 3 X 6. The breakup of 3 X 6 into 10 and 8 is illustrated in Figs
(3.3b) and (3.3c).

The weight diagram for the product representation 3 x 6 I

shown in Fig. 3.3a. The state of highest weight, which has

M(3,O)
V3
= ( T':2 1) T3 = T = 2'
3
Y = 1 (3.33

belongs to a T-spin quartet. (This state is also the V3 = - 3/2

member of a V-spin quartet.) The basis functions !fa (0: = 1, . .. ,6
of the representation D(2,O) were obtained in Eqs. (3.9-3.14)
(To avoid confusion we denote the triplets by g instead of !f.)
Labeling the sites in the composite weight diagram as in Fig
3.3b, the T-spin quartet belonging to the state (/)1 of highes
weight is
. :ON f; TRUCTION OF REPRESENTATIONS OF SU( 3) 69

(3.34)

cP = _1 T m. = ./,3 C2
4 V3 - 'l'3 'f' ~

III order of decreasing T 3. Applying V + to CPI results in a T 3 =

'/' = 1 state at site 5 in Fig. 3.3b. Applying T _ to CPs twice r esults
III the T-spin triplet:

(3.35)

II T-spin doublet is reached by applying V + to CPs, and then

npplying T_

(3.36)

lind finally a T-spin singlet CPIO is reached by applying V +

to CPa:

(3.37)

T he nature of the obtained functions is perhaps best exhibited

by expressing CPj (i = 1, ... ,10) completely in terms of the con-
stituent triplets:
70 INTRODUCTION TO UNITARY SYMMETRY

IPI = e(1)e(2)e(3)

IP2 = J3 W(l)e(2)e(3) + e(1)e(2)e(3) + e(1)e(2)e(3»

IP3 = J3 (e(l)e(2)e(3) + e(l)e(2)e(3) + e(1)e(2)e(3»

IP4 = e(1)e(2)e(3)

IP5 = J3 W(1)e(2)e(3) + e(1)e(2)e(3) + e(1)e(2)e(3»

IPs = J6 W(1)e(2)e(3) + e(1)e(2)e(3) + e(1)e(2)e(3)

+ e(1)e(2)e(3) + e(l)e(2)e(3) + e(1)e(2)e(3»

IP7 = J3 (e(1)e(2)e(3) + e(l)e(2)e(3) + e(1)e(2)e(3»

IPs = J3 (e(1)e(2)e(3) + e(1)e(2)e(3) + e(1)e(2)e(3»

IPg = J3 (e(1)e(2)e(3) + e(1)e(2)e(3) + e(1)e(2)e(3»

IP10 = e(l)e(2)e(3) (3.38)

The IP j exhaust the set of functions completely symmetrica
under interchange of the coordinates Cl of the triplet functions
g(Cl). This fact implies that we have found all the members of the
irreducible representation D(3,0), since the generators E ± a =
E±a(l) + E±a(2) + E±a(3) are symmetrical under interchange
of 1, 2, and 3, and can never change the symmetry of a function,
even though the labels k of e'
are permuted by E ± a' (Recall that
we begin with the state of highest weight, which is completely
symmetrical. )
The representation 10 (D(3,0» is completely symmetrical with
regard to T, U, and V-spin. As discussed in Section 3.4, its
isospin-hypercharge content coincides with the observed set of P3/2
meson- baryon resonances.
• UNST RUCTION OF REPRESENTATIONS OF SU(3) 71

The generators, expressed in terms of the outer product nota-

explained after Eq. (3.31), are given by
1'011

T_ = V312><11 + 213><21 + v314><31

+v216><51 + v217><61 + 19><81
U_ = V319><10I + 217><91 + v314><71 (3.39)
+v216><81 + v 2 13><61 + 12><51
V_ = v318><10I + 215><81 + v311><51
+v216><91 + v212><61 + 13><71
The rest of the reduction of 6 x 3 follows easily. Separating
"II' the sites of 10 (Fig. 3.3b) from Fig. 3.3a leaves the weight
oIingram of D(1,1), Fig. 3.3c. Taking that combination of functions
II' Hi te 5 orthogonal to (/>5 gives the state of highest weight of the
"'f.( ht -dimensional representation:

\', == vtVs1e - vtVs 4 e

= J6 (2e(1)e(2)e(3) - e(1)e(2)e(3) - e(1)e(2)e(3))

(3.40)
(:I.:arly there is an arbitrary overall phase. We have' chosen the
I'hase so that plus one goes with the left-hand state Vs with the
I,if.\hest weight. We shall not quote the remaining seven functions
II HHociated with (3.40). Note that (3.40) is symmetrical in coordi-
illites 1 and 2 [Vs k = Vs k (1,2) is symmetrical]. The symmetry in the
I'llirs (1,3) and (2,3) though" mixed" is preserved by the group
operations, and is best analyzed using Young diagrams. [The
,·if.\ ht-dimensional representation obtained from 3* x 3 in (3.32)
.M antisymmetrical in 1 and 2.]
The complete reduction is therefore

3 x 3 x 3 = 10 +8 +8 +1 (3.41)
Note that the two eight-dimensional representations, though
co mpletely indistinguishable under S U(3) transformations (action
72 INTRODUCTION TO UNITARY SYMMETRY

on the basis function labels) can be distinguished by symmetry

properties outside the group, in this case permutations of the
coordinates which distinguish one triplet g from another. We shal
often have occasion to distinguish among S U(3)-identical basis
functions in this manner.

3.3 Casimir Operators for SU(3)

Weare now in a position to discuss the so-called Casimir
operators for SU(3). For orientation, note that in SU(2) the
eigenvalues of J2 (square of the" angular momentum") are used
to distinguish the various irreducible representations from each
other. Within an irreducible representation J2 = j(j + 1)1
where j takes on anyone of the values 0, 1/2, 1, 3/2, ... J2 is the
Casimir operator for S U(2). In S U(3) there are two independent
quantities that commute with all the generators and so are pro-
portional to the identity within an irreducible representation.
One of these operators is clearly G2. However, let us proceed in a
more systematic manner. We have already noted that there are
two independent ways to couple two sets of functions transform-
ing as 8 to get another basis transforming in the same way. (That
there are only two ways will be proved in Section 3.6.) The sets
V/ and Vc D of Eqs. (2.101) and (2.107) are clearly independent
since they are orthogonal. Thus, given a third basis set Cc trans-
forming as 8 we can form a scalar from A, B, C in two independent
ways
SF = (Fc)abAaBbCc
(3.42)
SD = (Dc)abAaBbCc
In SU(2) there is only one analogous invariant, B'jkA,BfCk =
A.(B X C).
In analogous fashion we can construct from an irreducible repre-
sentation of the generators Ga satisfying (2.26) two independent
vector operators
C) Fa = (Fa)bCGbGc
(3.43)
C)D a = (Da)bCGbGc
CONSTR UCTION OF REPRESENTATIONS OF S U( 3) 73

'I'hen the quantities

C2 ' = (Fa)beGaGbGe
(3.44)
Ca = (DaheGaGbGe
" .. mmute with all the Gk and by Schur's Lemma are multiples of
I he identity within an irreducible representation of S U(3). The
"igcnvalues of these Casimir operators are used to characterize the
v II rious representations.
Previously it was noted that G2 is a Casimir operator. In fact
II coincides with C2 ' up to a constant factor. To show this,
IlUlc that in C2 ' we can replace GbGe by UGb,Gel because of the
IIlIli-symmetry of (Fahe in the lower indices. Hence

(3.45)

using Eq. (2.109), anticipating the precise value of F2. Since

1111

is of simpler structure we shall use it instead of the equivalent

(; 2
form C2 ' of Eq. (3.45). Thus we take our definition of the Casimir
operators to be:

(3.46)

The evaluation of C2 is easily accomplished by converting it to

I he basis of the E ± a' H and applying C2 to the state of highest
wcight M of the irreducible representation in question. Using the
correspondence (2.18) (tA. ~ G) we find

C= 3{H12 + H22 + t~ (EtE_ t + E_iEt)}

2 (3.4 7)

We recall that for the state .p(M) of highest weight M

E1.p(M) = E _ 2.p(M) == E _3.p(M) = 0 (3.48)

Using the commutator [Ea,E-al = r(a)·H we then find

74 INTRODUCTION TO UNITARY SYMMETRY

C2.p(M) = 3{H12 + H22 + EIE_l + E _2E2 + E_3E3}.p(M)

= 3{M12 + M22 + [El,E_lJ + [E-2,E2J + [E_3,E3]}.p(M)

= 3{M12 + M22 + [r(l) - r(2) - r(3)J·M}.p(M)

= 3{M12 + M22 + :3 Ml }.p(M) (3.49)

Substitution of Ml = V3 (p + q) /6, M2 = (p - q)/6 now yields

(3.50)
The values of C2 for various representations of moderately low
dimension can be found in Table 3.2.

TABLE 3.2
Dimensionality and Values of the Casimir Operators for Some Low-Dimensional
Representations (p,q) of SU(3).

(p,q) (0,0) (1,0) (1,1) (2,0) (2,1) (2,2) (3,0) (3,1) (3,2) (3,3) (4,0) (4,1)
Dimension 1 3 8 6 15 27 10 24 42 64 15 35
C2 o 4/3 3 10/3 16/3 8 6 25/3 34/3 15 28/3 12
C3 o 10/9 0 35/9 28/9 0 9 80/9 55/9 0 154/9 18

The evaluation of C3 is somewhat more involved. For the

details we refer the reader to the paper of Baird and Biedenharn
(9), which also considers analogous constructions for the group
SU(n). In terms of our notation their result is

C3 (p,q) = ·h(p - q)(2p + q + 3)(2q + P + 3) (3.51)

Clearly either the pair (p,q) or C2 , C3 can be used to characterize
a representation. Note that C2 is symmetrical in (p,q) while C3 is
antisymmetrical. Thus the sign of C3 distinguishes a representa-
tion D(p,q) from its complex conjugate equivalent D*(q,p). For
p = q, C3 = 0, and D(p,p) is equivalent to D*(p,p). C3 van-
ishes only for those representations which are equivalent to their
complex conjugate.
W e can now complete the evaluation of Tr Fa Fb and Tr Da Db
CONS TRUCTION OF REPRESENTATIONS OF SU(3) 75

hq:; un in Eq. (2.109). Since F2 = 3, from Eq. (3.50) with P = q=

I , we find
(3.52)

We can evaluate Tr DaDb by writing out Eq. (2.116) for [Da,Dbl

i ll full

1)"CUDbUd - DbckDakd = -i(ll ac ll bd - Iladllbc) - (Fk)ab(Fk)cd

(3.53)
Setting b = c and summing on b gives

which reduces to
(3.55)

I liking into account the fact that D has vanishing trace.

F'or reference we also express C2 in terms of the generators A/
defined in Eq. (2.27). In terms of these quantities it is easy to
".,nstruct scalars by complete contraction, e.g.,

(A,A) == A/A/
(3.56)
(A,A,A) == A/A/A,/'

lind so forth, commute with all the A/' and so qualify as Casimir
"pcrators. (Contractions of products of A's greater than 3 gives
~(; al ars that are not independent of the above.)
The connection of the A/ to the Hermitian generators Gk is

1 (3.57)
A22 = - G3 + VS Ga = - Q + Y == - Z,

2
A 33 = - VS Ga = - Y; A12 = (Al)t, etc.
76 INTRODUCTION TO UNITARY SYMMETRY

Expanding Ai' Au· in terms of its constituents leads, after appli

cation of the commutation rules for the Ga , to the result

(A,A) = 2C2 (3.58

A similar method of constructing the Casimir operators fo

SU(n) has been given by Klein (8). An essentially equivalen
procedure has been described by Biedenharn (9).

3.4 Physical Interpretation of the Representations

An unsatisfactory feature of the theory as developed up to this

point is its avoidance of a basic qllestion: why is the Sakata
model wrong? Why must we use the more complicated assign
ments of the eightfold way? One wonders why no S U(3) triplets
have been observed, since objects of this type were used to define
the group. From the point of view of bootstrap theories based on
dispersion relations, the realization of a given representation by
means of particles or resonances is purely a dynamical question
However, the poor numerical accuracy that characterizes such
calculations prevents the exclusion of the Sakata model on this
hasis. On the other hand, the traditions of field theory sugges
that one attempt to associate actual particle fields with the
defining representation 3. There are quite a number of ways to do
this; since there is no experimental hint that any of these schemes
is correct we only mention the most economical model (10, 11)
which employs only one field. Our reason for doing this is that
practical calculations are often clarified by the use of these
probably fictitious particles. The properties of this field must be
such that the properties of the observed multiplets are reproduced.
Recalling the relations

3 X 3* =8+1
(3.59)
3 x 3 X 3 = 10 +8+8+1
suggests that the baryon octet and decuplet be composed of three
spin 1/2 Dirac particles q while the pseudoscalar and vector meson
 CONSTRUCTION OF REPRESENTATIONS OF SU(3) 77

o(:lcts are made up from q and anti-q. We follow Gell-Mann (10)

ill calling these (presently hypothetical) particles quarks. The
" e~ triction to one field requires the quarks to have baryon number
1/3 . Thus instead of Sakata's (p,n,ll) triplet we have

(3.60)

where (ql>q2) still make up a T-spin doublet and qa aT-spin

Minglet . However, the joint requirements that baryons transform
n~ qqq imposes the eightfold way relations Y = 2H2 , Ta = V3 Hl
on the quark quantum numbers. The quantum numbers of qf are
listed in Table 3.3. The discovery of such bizarre objects would

TABLE 3.3
Quantum Numbers of the Components of the Quark Field (a)
and of Some Products (b).
(a)
Q y T3 B

q,
q2 I -1/3
2/3 1/3
1/3
1/2
-1/2
1/3
1/3
q3 -1/3 -2/3 0 1/3
(b)
Q T3 Y B
Q,q2 -1 -1 0 0
q2q3 0 1/2 -1 0
Q3q, 1 1/2 1 0

~u rely be an exciting event. At the time of writing no such par-

licles have been discovered with mass less than several nucleon
masses.
Whether or not quarks exist, they are extremely useful and
suggestive in calculations. We exhibit this by constructing the
appropriate field operator for the P-meson octet transforming
according to the hermitian regular representation F. The appro-
priate set of linear combinations of qf*qj transforming in this way
78 INTRODUCTION TO UNITARY SYMMET

is given by P a = "ij/t.aq/V2, a = 1, ... ,8. The eight components

P a are given explicitly by
~
P5 = Vz, (ii3ql - iilq3)

~ 1
P2 = V2 (ii2ql - iilq2) Ps = Vz, (ii2q3 + ii3q2)
(3.
1 ~
P3 = V2 (iilql - ii2q2) P7 = V2 (ii3q2 - ii2q3)

1
P4 = ~2 (iilq3 + ii3ql) Pa = VI) (iilql + q2q2 - 2q3q3)

The Pi have zero baryon number. P a is clearly an isosinglet w

zero hypercharge. P 3 has T3 = Y = 0; it has opposite symme
to P a and has isospin 1. P a and P 3 clearly correspond to 77° a
7]. Inspection of Table 3.3 suggests in addition the follow
identifications:

ii2ql = 77+, iilq2 = (77+)* = 77-

ii3ql = K+, iilq3 = (K+)* = K - (3.

ii2q3 = KO, q3q2 = (KO)* = , KO

The combination of P-meson fields transforming via the gene
tors F is therefore

(77+ + 77-)/V2
i (77+ - 77-)/V2
77°
(K+ + K-)/V2
P= P* = P (3.
i(K+ - K-)/V2

(KO + KO)/V2

i(KO - KO)/V2

7]
CO NSTRUCTION OF REPRESENTATIONS OF SU(3) 79

Notice that P is self-conjugate: Pa* = P a. When q is subject to

lhe transformation q --+ exp (iex· "j2)q, P undergoes the trans-
for mation P --+ exp (iex·F)P. In this case the complex conjugate
representation exp ( - iex · F*) is identical to the original one since
F* = - F. The underlying quark fields may now be abandoned in
favor of the" physical" fields P a' For the vector meson octet we
~ ubstitute 1T± --+ P ±, 1To --+ pO, K± --+ K* ±, KO --+ KO*, KO --+ KO*,
TJ -+ 4>0' However, the baryon octet is not self-conjugate and it is
convenient to make the following definition of the baryon octet
Ba transforming as (3.63):

(.E+ + .E-)jV2
i(.E+ - .E-)jV2
.Eo

(p - E-)jV2
B= (3.64)
i(p + E-)jV2
(n + EO)/V2
i(n - EO)jV2
A

In (3.64) E- is quite independent of p, in contrast to the pair

K+, K- in (3.63). The sign on E- has been chosen so that the
customary isospin doublet (EO,E-) enters in the S U(2)-invariant
couplings. The above fields now agree with the conventions of
Eq. (l.52). In particular, .E± = (.E1 ± i.E2 )jV2 destroy the
indicated particles, etc. As mentioned at the end of Chapter 1, the
operators in Eqs. (3.63) and (3.64) do not create multiplets having
the useful standard phase relation. This problem will be discussed
shortly.
First, however, we indicate the connection of the above
approach with the commonly used 3 x 3 matrix representation,
which arises naturally when one uses tensor analysis for the con-
struction of irreducible representations. We may define traceless
80 INTRODUCTION TO UNITARY SYMMETR

3 x 3 Hermitian matrices by contracting P or B with A

g = 2a PaAa/V2, $ = 2a Ba Aa/ V2
V37TO + 7J 7T+ K+
V6
g= 7T
-V37TO + 7J KO
V6
-2
K- KO
V6 7J
(3.65
V3,E° + AO ,E+ P
V6
$= ,E- -V3,E° + A
n
V6
-2
-8 - 8° A
V6
It is easily verified that the transformation of g by the 3 x
matrix U = exp (ia: . A/2):

(3.66
agrees with the relations

(3.67

The conjugate?} is defined to include Hermitian conjugation o

both the fields P a and the matrix A; hence g = g for the meso
matrix and 33 is given by

V3,E° + 7.I r- -8-

V6
33= ,E+ - V"3!'U + 7.I SO (3.68
V6
-2
P it 1I
V6
CONS TRUCTION OF REPRESENTATIONS OF SU(3) 81

Under charge conjugation the matrix ~ is simply changed into

iI H transpose. This operation is also the same as reflecting in the
origin of the weight diagram. The latter is not the same as charge
t'llOjugation for 33 but is sometimes a useful concept nevertheless.
We define the R-transformation by

(3.69)

This transformation (defined more generally below) does not

hdong to those of S U(3). (In particular the strong interactions are
definitely not invariant under" R-conjugation ".) To demonstrate
I his for, say, the matrix ~ we can ask whether there exists any V

which transposes ~:
?

V~V-l ~ ~T (3.70)

I f this were the case, then

(3.71)
For i = 1,3,4,6,8 we have AI· = Ai and for i = 2,5,7, AI· =
- AI' Thus V must simultaneously satisfy
[V,A I ] = 0 i = 1, 3, 4, 6, 8
(3.72)
{V,A i } = 0 i = 2,5,7
The most general 3 X 3 matrix V can be written in the form

V = aI + b'A (3.73)
where a and bl (i = 1, ... ,8) are complex numbers. We now show
lhat (3.72) implies a = bl = O. For i = 2, 5, 7 we have

(3.74)
while for i = 1, 3,4, 6, 8 one finds
(3.75)

Taking the trace of Eq. (3.74) gives bi = 0, i = 2,5,7. Eq. (3.75)

implies that bd"ikl = O. Taking successively k = 1,3,4,6, and
noting nonzero values of fi i k' shows that b3 = b1 = bs = b4 = O.
82 INTRODUCTION TO UNITARY SYMMETR

To eliminate bs we note that now bd';kl = bsiskl' so that takin

(k,l) = (6,7) completes the proof that b = o. Eq. (3.74) reduce
to 2aA i = 0 which in turn implies that a vanishes. A similar proo
can be given for the inequivalence of 3 and 3*. In that case on
proceeds as above, in showing that no 3 x 3 matrix V exists suc
that VAjV-l = -Ai*. [Cf. the discussion following Eq. (2.73).]
We now discuss the identification of state vectors with th
components of irreducible representations of S U(3). For this pur
pose, it is useful to employ phase conventions different from thos
used in the preceding discussion of field operators, as has bee
mentioned several times. Thus, in the following discussion particl
symbols always refer to states having the standard phase relation
Hence, the states 'Pi (i = 1-8) of Section 3.2 can be identifie
with the following sets of physical states:
Ip) IK+) IK+*)
In) IKO) IKO*)
127 +) 177"+) Ip+)
127°) 177"°) IpO)
IBi) = IPi) !Vi)
127-) 177"-) Ip-)
IA) 17]) 14>0)
lEO) IKO) IKo*)
IE-) IK-) IK-*)
(3.76
provided one is rather tolerant of the obvious nondegeneracy o
the masses. (See Fig. 3.4, for example.) The fact that the strongl
interacting particles (3.76) fit so neatly into the regular representa
tion of S U(3) is surely astonishing and beautiful. There are, o
course, many other implications of the presumed invariance of th
strong interactions which must be satisfied, at least approxi
mately, if the symmetry is to be believed. These other relation
are the subject of later chapters. The identification of parti
cles with the basis functions of irreducible representations is
necessary preliminary for such applications.
CO NSTRUCTION OF REPRESENTATIONS OF SU(3) 83

n p
• •

-- .
~ .~
_0
Fig. 3.4. The baryon octet is the prime example of the correspondence of observed
low-lying particles with the regular representations of S U(3) .

One can also group the particle states (3.76) into V- or U-spin
rnultiplets; for instance, the sets (.E+,8°), (n,.E-) , (p,t(.Eo +
\1'3 11),8-), and!( _.Eo + V311) give V-spin doublch, a triplet,
nnd a singlet. If desired, appropriate field operators fo r U- or
V-spin may be constructed as was done in Section 1.4 for ordinary
7'-spin_ For completeness we give the relation between the
operators appearing in (3.64) and the corresponding states in
(:3 .76). Recalling the discussion of Chapter 1, we note the
eonnection

Ip) p • IK+ ) K+ •
In) n IKO ) KO

I.E+ ) -.E+ 17T+ ) -7T+

I.E°) .Eo 17T0) 7T 0
xIO); xlO) (3_77)
I.E- ) .E- 17T- ) 7T
111) 11 17]) 7]
18°) 8° IKO ) KO
18 - ) g- IK- ) -K-

[n (3.77) 10) stands for the vacuum state and the asterisk denotes
Hermitian conjugation of the components of the column matrix.
T he set of P3 /2 baryon resonances and the so-called Q-, stable
84 INTRODUCTION TO UNITARY SYMMET

under strong interactions, fit in nicely with the representation

whose weight diagram was given in Fig. 3.3b. (Notation: P
means L J , where L is the orbital momentum of the meson a
baryon, and J the total angular momentum.) The spin and par
of Q - are not yet conclusive, but the Q- mass is so close to th
predicted by the Gell-Mann, Okubo mass formula (see Chap.
for this representation that one has considerable confidence in t
interpretation of the Q-. The identification of the N*(123
quartet, the Y1 *(1385) triplet, the 8*(1430) doublet and
Q-(l585) singlet with the sites in the 10 weight diagram follo
upon comparison of Figs. 2.2a and 3.3b. (The number in pare
theses following a particle name is the mass, in MeV.) From 1
10 the states are N*++, N*+,N*o, N*-, Y 1 *+, Y 1 *o, Y 1 *-,8*
8* -, and Q - . The reader may prefer the less cumbersome b
equally arbitrary nomenclature described by Chew et al (12).
Extremely important theoretical support for the above pict
arises from the natural way in which the existence of the P
decuplet can be understood in terms of a simple dynamical mod
The elucidation of such connections between symmetries a
dynamics is under active investigation at the moment. T
general ideas of the dynamical approach are sketched in Chapte
There are a large number of additional resonances, mainly
higher energy and higher spin than the "low-lying" states d
cussed above. There is also a T = 0 TT]; resonance at 1405 M
(apparently 5 1 / 2 ) which can be (and according to dynamical c
culations, should be) considered to be a unitary singlet. There
a DS/2 state at 1520 Me V that may also be a unitary singlet. W
shall defer discussion of the possible existence of other represen
tions to subsequent chapters. We only mention here that there
substantial, though not unambiguous, evidence for a DS/2 oc
of meson-baryon resonances including the 1512-Me V TT N ma
mum (see Glashow and Rosenfeld (13)).

3.5 The Reduction of 8 X 8

In this section we work out in detail the basis functions of
irreducible representations contained (14) in 8 X 8. The resul
CONSTRUCTION OF REPRESENTATIONS OF SU(3) 85

of prime interest in applications and is sufficiently complicated to

illuminate the general analysis of representations.
For clarity we label the first set by the baryon labels and the
Hccond by the pseudoscalar mesons. The sixty-four states BjPj
(i,j = 1, ... ,8) corresponding to the sets of Eq. (3.76) generate a
reducible representation; the state of highest weight being E+l7+.
Adding the octet weight diagrams gives rise to Fig. 3.5; the
degeneracy at a given site is indicated by the attached number.
AII the states belonging to the same representation as E + 17 + can
he obtained by: applying T ± and V± to E+l7+ and all the states
obtained thereby. (The use of U ± along with its minus signs can
he easily avoided.) Applying T _ to E+l7+ generates a T = 2
clui ntet (El7)2' The following conventions are adopted with
respect to the isospin wave functions: the baryon is particle
" Humber one" and stands on the left while the meson is particle
I wo and stands on the right. The isospin T wave function formed
fro m baryon and meson is written (BPh, where the T3 value
usually does not need to be specified. The sign convention is as

I 2 I
• • •

2 6 6 2
• • • •

6 10 6
• • • • •
2 6 6 2
• • • •

I 2 I
• • •
I'ig. 3.5. Vector addition of the octet weight diagrams gives the degeneracies
indicated. (The central site has Y = T3 = 0.)
86 INTRODUCTION TO UNITARY SYMMETRY

Fig. 3.6. This diagram shows the paths followed in separating the representatio
27 from the 64-dimensional reducible representation 8 x 8. Site a is the (non
degenerate) state of highest weight.

follows: particle number one having the highest charge is always

given a positive Clebsch-Gordan coefficient. For example, the
T3 = 1/2 state in the multiplet (N7Th/2 is (p7T O - V2 n7T+)/v3
while that in (EKh/2 is (V2E+ KO - EOK+)/V3.
E+7T+ is also the V3 = -1 member of a V-spin triplet. Apply
ing V +I'VZ to E+7T+ == o/a then gives the normalized state o/b a
site b of Fig. 3.6. We find
1 1 -
o/b = vZ V+E+7T+ = VZ(E07T+ + E+KO) (3.78)

which is the T3 = 3/2 component of a T-spin quartet

1 -
vZ [(E7T)3/2 + (EKhd (3.79)

the remaining components are found from the usual isospin con-
siderations. Proceeding to site c of Fig. 3.6 by applying V + IV2 to
IjIb yields
(3.80)

which is the T3 = 1 member of a T-spin triplet (EKh. Going now

to site d, to which belongs the V = 1 state o/d = E-7T-, we obtain
a function % with T3 = - T = - 3/2
CONSTRUCTION OF REPRESENTATIONS OF SU(3) 87

(3.81)

one member of the T = 3/2 quartet [(N1Tb2 + (EK)3 12]/V2.

Promoting o/e again by V _I V2 yields

1
0/[ = V2 V -o/e = nKo (3.82)

lind the associated triplet (NKh.

We next ask whether the other independent state at site b,
[(EK)312 - (E1Th '2]/V2 (note sign convention), belongs to the
Harne representation as o/b' To see that this is not the case we go
from a to b along the paths agh and aghb. Neglecting normalization
fuctors, along path agh we have
<Pb == U_T-o/a = [U _,T_]o/a + T _U _o/ a
(3.83)
= - V+o/a + Ooc o/b
Similarly, along path aghb we end up with the function

Xb == T + V + T -o/a = T + T _ V +o/a oc o/b (3.84)

Mince T _ commutes with V + and o/b is an eigenfunction of T + T _.

A slight extension of this argument shows that no path can
~e nerate a function at b independent of o/b' Hence, by using the
Weyl reflections, the boundary points e, k, and I of Fig. 3.6 have
only one state belonging to the same irreducible representation as
'Ila. This result can be generalized; the boundary of the weight
diagram of an irreducible representation is nondegenerate.
Thus any more functions belonging to o/a's representation must
lie within the boundary. We start at point k, where o/k =
(p1T+ + E+ K+ )/v2 has V3 = - V = - 3/2, and create a state
at site g:

1 1 1
all = V3 V+o/k = V6[PK'° + EOK+ + V2(E+1TO + E01T+)

+ Vt(A01T+ + E+7])] (3.85)

88 INTRODUCTION TO UNITARY SYMM

This function is independent of, but not orthogonal to, the T3

member of the (E7T)2 multiplet. Orthogonalization gives

a T = T3 = 1 function. By reflection symmetry, the rou

will not give a function at g independent of .Jig and <fog. Nor
route lmg independent of lkg, because V + U _ = U _ V +. Fur
more, the route kmg is equal to a combination of kg and kag,
T_U_ = V+ + U_T_.
As these results are general, we state them graphically in
of the parallelograms of Fig. 3.7. Consider going from point
2 in (a) and (b). In case (a) we have 120° turns; the ope
products corresponding to paths A and Bare T _ V + and V
respectively. Clearly, we can rotate the figure by 120° and
and obtain the same result. If, however, we turn by 240°
Fig. 3.7b, the paths A and B do not yield the same result. P
involves U _ T _ while B stands for T _ U _. Since [T _,U _] =
we see that
path A + path (12) = path B

(The latter results apply also for parallelograms rotated by

and 240°.) The T = 1 multiplet associated with <fog is

2
iSl . B
A\Z 2 B
(0) (b)

Fig. 3.7. Sites 1. 2. A. and B are occupied sites in a weight diagram.

from 1 to 2 (using the appropriate operators) through A or B gives th
result for paths of type (a) and a different result for paths of type (b).
j :ONSTRUCTION OF REPRESENTATIONS OF SU(3) 89

Next we apply V+ to the state pK+ at site l. Since pK+ is the

VJ = - 2 member of a V-spin quintet we find

= ~ [p7TO + V3pYJ + ,E°K+ + v3AK+J

2V2
(3.89)
This state does not have a definite T-spin . We find a T = T3 =
1/2 state by removing from .pm its projection onto the T3 = 1/2
,"ember of the T = 3/2 state at site m, [(N7Th/2 + (,EK)3/2J/V2.
The result is the doublet

(3.90)

Si milarly proceeding from n to 0 or from g to h by V ± and

.. rlhogonalizing to the Y = -1, T = 3/2 multiplet (3.79) we
.. htain a T = 1/2, Y = -1 doublet

(3.91)

II is n ow clear that the inner octagon (sites mghpqr) of Fig. 3.6

iH doubly degenerate with states attached to the state of highest
wcight. We now have to find out what states belong to this
""presentation at the origin 0 of the weight diagram. There is
.. Illy one more, obtained by going from the V3 = -1 state (3.89)
I" the origin. The result of applying V + to (3.89) is a superposition
.. f T = 0, 1, and 2 components. Extracting the T = 1 and 2
(,arts already found gives the Y = 0 isosinglet

(3.92)

By reflection symmetry, no other approach to the origin gives

thing n!)w.
lilly

We summarize the wave functions of the 27 -dimensional

irreducible representation D(2,2) in Table 3.4. The degeneracy of
the weight diagram of 27 is shown in Fig. 3.8.
90 INTRODUCTION TO UNITARY SYM

• • •

2 2
• • • •

2 3 2
• • • • •

2 2
• • • •

• '. •
Fig. 3.8. The degeneracy of the weight diagram for the irreducible repres
27 is shown.

Removing the number of independent states of the repre

tion 27 of Fig. 3.8 from those of Fig. 3.5 reveals that the rem
state of highest weight occurs at site k of Fig. 3.6. Orthogona
to the member of 27 at k, (pK+ + E+ K+)/V2, gives the
(E+ K+ - pK+)/V2, the topmost member of the T = 3/2 q

Going from site k to site g gives the T3 = 1 member of the

triplet

.} [V2( EOK h - V2(NKh + V3(E'l)h - V3(A7Th

v 12
+ V2 (E7T)d
Continuing in the V + direction gives a T-spin doublet
T-spin singlet

.} [V3(EKh f2 + V3(E7Th f2 + V3(E'l)h f2 - V3(AK)]

v 12
(EK) 0
The ten independent states contained in the multiplets (
CO NSTRUCTION OF REPRESENTATIONS OF SU(3) 91

(3.96) form a basis for the representation 10. As seen in Section

:L2, the weights of this representation are nondegenerate. [This
result actually obtains for all triangular representations; the
latter are ofthe form D(n,O) or D(O,n). The state of highest weight
also has highest hypercharge. The top boundary is flat and
reflection invariance implies a triangular shape. The non-
degeneracy follows from the observation that only one completely
Hymmetrical function can be formed from n g's at each site, the
generators being permutation operators working inward from the
Htate of highest weight.] One can make sure that the state at site
II does not belong to the above representation by using the rules
OIl' Fig. 3.7. The path kab, which gives zero, is equivalent to kgb.
Thus b can't be reached fr om k through g. Further considerations
nf the same nature show that no path leads from k to b.
Having removed 27 and 10 from 8 x 8, we remove the
representation attached to the state of next highest weight, at
Hite b. Again one obtains ten functions, belonging to the repre-
Mc ntation 10· [or D(O,3)], whose weight diagram is obtained from
I hat of 10 by reflection in the origin. As the calculation is essen-
I ially the same as for 10, we omit the details and give the result
in Table 3.4.
Removal of the states of 27, 10, and 10· from the weight
diagram of Fig. 3.5 leaves us with Fig. 3.9. The latter' clearly

2 2
Fig. 3.9. After the representations 27, 10, and 10· have been removed from
/I x 8, the residual independent functions have the weights shown (with degeneracy
indicated). The lines indicate the paths followed in rearranging these functions to
form two 8-dimensional and one I-dimensional irreducible representation.
92 INTRODUCTION TO UNITARY SYMMETRY

indicates that two octets and a singlet remain. We therefore have

obtained the reduction

8 x 8 = 27 + 10 + 10· + 81 + 82 + 1 (3.97
where the indices 1 and 2 label the two independent eight-dimen
sional representations, whose S U(3) transformation propertie
are identical.

TABLE 3.4
Wave Functions of the Irreducible Representations Contained in 8 X 8 in Term
of Isospin-Hypercharge Components.

Y T
----
t{l(Y,T) 27
2 . 1 (NK)l
3/2 [(N7Th'2 + (.EKh,2]/V2
1 1/2 [3(N1) + (AK) + (N7T)"2 - (EK)1/2]/V2ii
0 2 (E7T),
0 [V2(NKh + V2(8K), + V3(A7T) + V3(E1)]/VlO
0 0 [V6(NK)o - V6(8K)o + 3V"3(A1) - (E7T)o]/V40
-1 3/2 [(87Th'2 + (EK)3/2]/ V 2
-1 1/2 [3(AK) + 3(81) - (87T)1/2 + (EK')1/2]/V2ii
-2 (8Kh

t{l(Y,T) 10

3/2 [(EKh,2 - (N7Th,2]/V2

o 1 [V2(8Kh - V2(NK)1 + V3(.E1) - V3(A7T) + V2(E7T),]/V1
-1 1/2 [(87Th'2 + (EK)1/2 + (81) - (AK)]/2
-2 0 (8K)o

t{l(Y, T) 10·
2 0 -(NK)o
1/2 [- (N7T)1/2 - (EK)1/2 - (AK)1/2 + (N1)]/V2
o 1 [V2(NKh - V2(E7Th - V2(8Kh + v"3(.E1) - V"3(A7T)]/Vi
-1 3/2 [(EKh'2 - (87Th,2]/V2
(continued
CONSTRUCTION OF REPRESENTATIONS OF SU(3) 93

TABLE 3.4-conl .

Y T .p
--- ---1-----------------------------------------
.p(Y,T) 81
1/2 [- 3(L'K)1/2 + 3(N7Th/2 - (N7J) - (AK)]/2V5
o [V2(L'7J) + V2(A7T) - V3(8Kh - V3(NK)d/VlO
o 0 [- V6 (L'7T)0 + (NK) 0- (8K)0 - V2 (A7J)]/VlO
- 1 1/2 [3(L'K)112 - (871) - 3(87Th/2 - (AK)]/2V5

.p(Y,T) 82
1/2 [(L'Kh/2 + (N7T)1/2 + (N7J) - (AK)]/2
o [2(L'7Th + (NK)l - (8K)d/V6
o 0 [(NK)o + (8K)0]/V2
- 1 1/2 [(L'.Kh /2 + (87Th/2 + (AK) - (871)]/2
Y= T = 0 -I

.p = 1/4[2(NK)0 - (8K)0 + V6(L'7T)0 - 2(A7J)]

To obtain the wave functions of the two octets we begin once

,"ore at the site of highest weight, here labeled a in Fig. 3.9.
From the six independent states .E+7TO, .E 0 7T+, pj(o, ,s0.E+, .E+T), and
117T+ belonging to this site, we form two orthogonal states also
orlhogonal to the four Y = 0, T3 = 1 states belonging to 27, 10,
" lid 10*. Expressing the latter in terms of symmetric and anti-
"y mmetric combinations [here S(x,y) means x + y and A(x,y)
,"cans x - y)]

,/1(27, T = 1) = h /2S(pKo,,s°K+) + Y3S(.E+T),ih+)]/YlO

'/1(10, T = 1) = [- Y2A(pKo,,s°K+) + Y3A(.E+T),il7T+) (3.98)

+ A(.E+1To,.E°1T+)]fYlO
'/1(10*, T = 1) = [Y2A(pKo,,s°K+) + Y3A(.E+7],il7T+)
- A(.E+1To,.E°7T+)]fYlO
94 INTRODUCTION TO UNITARY SYMMETRY

we see immediately that the appropriate functions are proportiona

to V2S(.E+7],ATT+) - V3S(EoK+ ,pKO) and V2 A(.E +TTO,.E°TT +
+ A(pKO,EoK+). The two octets then have states of highes
weight

!fs 1 H = [V2(.E+7] + ATT+) - V3(EoK+ + pKO)J/VI0

(3.99
!fs2 H = [V2(L'+TTO - L'°TT+) + (pKO - EOK+)J/V6

and corresponding Y = 0, T-spin triplets

!fs , (T = 1) = [V2(L'7]h + V2(A7i"h - V3(EKh

- V3(NKhJ/VlO (3.100)

!fs2 ( T = 1) = [2(.ETTh + (NKh - (EKhJ/V6

Proceeding in a now overly familiar manner along the directions
indicated in Fig. 3.9 we obtain the wave functions for 8 1 and 8 2
given in Table 3.4.
Now only one state remains, the unitary singlet Y = T = 0
state. It must be a linear combination of (L'TT)O, (NK)o, (A7]), and
(EK)o orthogonal to the three other states of the same weight hav-
ing T = Y = O. A brief, tedious calculation gives

!f(0,0) = t[2(NK)o - 2(EK)o + V6(.ETT)O - V2A7]J (3.101)

All these results are collected in Table 3.4.

In Section 2.6 we showed how to construct two independent
octets (transforming via the Hermitian generators F t ) from two
octets Aa and B b. These functions (JabcAbBc and dabcAbBc) trans-
form in the same way under SU(3) transformations, but have
different symmetry under the exchange A ~ B. The present
analysis proves that there are no more ways to construct octets
out of two octets.
The irreducible representations obtained in the reduction of
8 x 8 have a further useful symmetry which we now discuss in
CO NSTRUCTION OF REPRESENTATIONS OF SU(3) 95

detail. We define the R-transformation by the following substitu-

Lions of states:

P +48-, n +-> go, A +--+ A,

(3.102)

Thus in this example R is essentially the operation of inver sion

in the octet weight diagram. (More generally, it is defined as su ch
un inversion.) The operation (3.102) can be used to distinguish the
representations 8 1 and 8 2 • Explicitly we have the following
Himple result

R0/ 27 ( Y, T, T 3 ) = 0/27( - Y, T, - T 3 )
R0/10 (Y,T,T3 ) = 0/10*(- Y,T,- T 3 )
R0/81 (Y, T, T 3 ) = 0/81( - Y, T, - T 3 ) (3.103)

R0/82 (Y,T,T3 ) = -0/82(- Y,T,- T 3 )

R0/1 (0,0,0) = 0/1(0,0,0)

Thus when we apply R to a component of an irreducible

representation contained in 8 x 8 we obtain a wave function
with inverted weight. Notice that R interchanges 10 and 10*. For
'he self-conjugate representations D(2,2), D(I,I), and D(O,O)
(27, 8 1 , and 8 2 , and 1) the state at the center of the weight
diagram is an eigenfunction of R. The eigenvalue is 1 for 27, 8 1 ,
li nd 1 but -1 for 8 2 • In this way we can use the R transformation,
which lies outside the group S U(3), to distinguish among repre-
Ilcntations which are otherwise equivalent.
Even though R does not belong to the group, one can ask
whether the strong interactions are invariant under R. There are
many indications that this is not the case. For instance, transitions
of the type 8 1 -+ 8 2 in PB elastic scattering would be forbidden
if R were conserved. However it is possible that in meson
processes the R-violating baryons are sufficiently unimportant
Lhat a limited R symmetry is valid. This possibility has been
discussed by Bronzan and Low (15).
96 INTRODUCTION TO UNITARY SYMMETRY

References
1. S. Gasiorowicz, Argonne National Laboratory Report ANL-6729 (1963
unpublished .
2. J . J . de Swart, Rev. Mod. Phys., 35, 916 (1963).
3. R. Behrends, C. Fronsdal, J. Dreitlein, and B. W. Lee, Rev. Mod. Phys., 34
1 (1962).
4. P. Tarjanne, Carnegie Inslitute of Technology Report NY09290, 9290A.
5. A. R. Edmonds, Proc. Roy. Soc. (London), A268, 567 (1962).
6. M. A. Rashid, Nuovo Cimento, 26, 118 (1962).
7. M. Hamermesh, Group Theory, Addison-Wesley, Reading, Mass., 1962, Chap
10.
8. A. Klein, J . Math. Phys., 4, 1283 (1963).
9. G. E. Baird and L. C. Biedenharn, J. Math. Phys., 4, 1449 (1963); 5, 172
(1964).
10. M. Gell-Mann, Phys. Letters, 8, 214 (1964).
11. G. Zweig, CERN Repl. Th. 401 and Th. 412 (1964).
12. G. F. Chew, M. Gell-Mann, and A. H . Rosenfeld, Sci. Am., 210 (No.2), 7
(1964).
13. S. L. Glashow and A. H. Rosenfeld, Phys. R ev. Letters, 10,192 (1963).
14. S. L. Glashow and J . J. Sakurai, Nuovo Cimento, 25, 337 (1962).
15. J . B. Bronzan and F. E. Low, Phys. Rev. Letters, 12, 522 (1964).
 CIUt/lli'" I

General Properties of Represenlaliull

of (/(:1)

4.1 Introduction
The previous chapter was concerned with the explio'il un"
detailed discussion of those low-dimensional representa I iUII M III
special physical interest. Although more extensive resuILH 110' 0 11111
often necessary in physical applications, some readers will wi .. I, I U
obtain a deeper understanding of the properties of the g rllup . "'"
obtain a panoramic view of the representations of SU(3) (I, ,, ' n
one might as well study SU(n) in general) there is no su Io Hli,u'o'
for the use of the symmetric group and its calculus of Y(lIlII /-I
diagrams. The symmetry properties of various specific repl'CMoHlI /I -
tions were noted in Chapter 3. We shall summarize a fcw of I" "
most useful methods and results obtained by this techniqu, ill
Section 4.6. The reader is directed to Weyl's Classical GrolLps ( ' )
for an extensive development of the subject, although it ma y h.,
casier to begin with Chapter 10 of Hamermesh's book on Croll!'
Theory (2). A series of papers by Baird and Biedenharn p , ,1,)
develops the theory of representations of SU(n) in the " cxpli ci l
manner preferred by physicists." Further references will bc giv ell
in Section 4.3.
Other topics we shall not describe in full generality concern thc
for m and multiplicity of the eigenvalue lattices corresponding to
the various representations of SU(3). However, much can b e said
fro m our previous analysis. First of all, the Weyl reflection
invariance (Chap. 2) shows that the weight diagram has to be
invariant under 120 0 rotations. Our analysis of the reduction of
97
98 INTRODUCTION TO UNITARY SYMMETRY

8 x 8 also shows generally that the boundary is nondegenerate

and convex. We also learned that for fixed Y, as one proceeds in
from the boundary one generally encounters a new independent
function for each step inwards, corresponding to a sequence of
isospin states of eigenvalue T, T - 1, T - 2, . . .. Analogous
remarks holds for the U- and V-spin directions except that if one
writes the independent functions at a site in terms of the subgroup
SU(2)T X U(l)y then these functions are linear combinations of
the U- or V-spin eigenfunctions. This detail does not affect the
counting of independent states. Thus, as one proceeds inwards
from the boundary along any of the lattice translation (root)
vectors the degeneracy of the site increases by one. What we shall
not prove is that the process terminates after the last layer
assumes a triangular shape. The eigenvalue lattice may be viewed
as a truncated pyramid, the top of which is a triangle. Consider
the weight diagram of the representation D(p,q). The state of
highest weight has Y = (p - q)/3 and T = (p + q)/2. It is
easily seen that one can take q steps in the U + direction and p
steps in the V + direction (the state of highest weight belongs to a
U-spin multiplet of dimension q + 1 and a V-spin multiplet of
dimension p + 1). This is illustrated in Fig. 4.1 for the special

Fig. 4.1. The degeneracy of the sites for the representation D(6,3) is shown to
illustrate the rule that the degeneracy increases by one as one follows contour lines
inward until a triangular contour is reached.
 GE NERAL PROPERTIES OF REPRESENTATIONS OF SU(3) 99

case p = 6, q = 3. Given this fact it is easy to compute the

dimensionality d(p,q) of the irreducible representation

d(p,q) = t(p + 1)(q + 1)(p +q+ 2) (4.1)

Later we shall prove this formula using tensor methods. For a

detailed description of the structure of eigenvalue lattices the
reader is referred to Gasiorowicz (5), Cutkosky (6) and Behrends
t:L al. (7).
In order to give a unique matrix representation of an irre-
ducible representation one has to make a fairly elaborate set of
phase conventions. These have been described in Section 3.5. In
hrief, we agree to use the eigenvalues (T, T 3 , Y) of the isospin-
It ypercharge subgroup to label the basis functions. The Condon-
Shortley phase convention is enforced for the T- and V-spin
operators. Biedenharn (4,8) and Racah (9) have given explicitly
lit e effect of applying V_to a given state </>( T, T 3 , Y) belonging to
f)(p,q )

V_ </>(T,T3'Y) = a+</>(T + t, T3 + t, Y + 1)
+ a_</>(T - t, T3 + t, Y + 1), (4.2)

(T + T3 + 1)[t(p - q) + T + Y + 1] t }1/2
/I ,. = x [t(p + 2q) + T + t Y + 2][t(2p + q) - T - t Y]
{
2(T + 1)(2T + 1)

(T - T 3)[t(q - p) + T - t Y] }1/2
II = { X [Hp + 2q) - T+ t Y + 1][H2p + q) + T- t Y + 1]
- 2T(2T + 1)

t\ long with the commutation rules and the usual matrix elements
.. I' the T ± operators, this result completely defines the represent a-
I ion matrices.

'•.. 2 SU(3) Clebsch-Gordan Coefficients

In this section we describe the properties of the S U(3) Clebsch-
(;ordan coefficients, following closely the review of de Swart (10).
100 INTRODUCTION TO UNITARY SYM

Special cases have already been given in Chapter 3. Supp

have two irreducible representations D(Pl,ql) and D
generated by the sets of basis functions </>v, (/1,) and </>V2 (/12),
tively. Here the labels f.Lj specify the representation and
particular basis function within the representation.
includes the quantum numbers T, T 3 , Y, while f.L stands
values of the two Casimir operators C2 and C3 , or equivale
pair (p,q). The dimensions of the representations are den
d 1 and d2 [cf. Eq. (4.1)]. The product </>v, (/1')</>V2 ("2) gene
reducible (except in the trivial case in which one of the
sentations is the trivial one) representation. In the usual w
splits the (d1 d 2 )-dimensional space spanned by </>v, (/1')</>V2
irreducible invariant subspaces by forming suitable line
binations:

Here f.L describes the irreducible representation thus obtai

v the composite quantum numbers (T, T 3 , Y). The labe
tinguishes independent but S U(3)-equivalent representati
example, in reducing 8 x 8 we found two eight-dime
representations which could be distinguished by their sy
properties [cf. Eq. (3.103) or Sec. 2.6]. Hence, in the C
Gordan series for S U(3),

D(PHql) x D(P2,q2) = .L n( P,Q)D( P,Q),

p.Q

the positive (or zero) integers n( P,Q) can be greater than

in contrast to the well-known vector-addition rule for S U
The analysis of the preceding chapter suggests that w
express the set <t> usefully in terms of a sum of various
wave functions. Clearly we can proceed in two steps. Fo
values of the additive quantum numbers T z = T 1z + T
Y = Y 1 + Y 2 we first construct eigenfunctions of T(T
T 2 ) using the usual S U(2) Clebsch-Gordan coefficients [th
tion of Rose (11) is used]:
(;EN ERAL PROPERTIES OF REPRESENTATIONS OF SU( 3) 101

TTzY)

L C( Tl T2 T; T lz T 2z T 3 Z )</>Vl ("1)</>V2 (" 2) (4.5)

Tl z T2z

Then we superpose the various X[v = (T, T z, Y) fix ed] to obtain

"igenfunctions of C2 , C3 :
({)(fLyv;fLlfL2)

= T~1 (~~l ,;:~J ;~)X(~~l ';:~2' TTzY) (4.6)

1'2 Y 2

In practice, the isoscalar factors (12)

fL1
( T1 Y
fL2 I
fLY)
T2 Y 2 TY
(4.7)
l

(so named because they are independent of the azimuthal

,/uantum numbers) are found by starting from the unique state
,,(' highest weight, for which (4.7) is defined to be + 1, and applying
I he r aising and lowering operators as described in Chapter 3. The
I'ocfficients of the isospin wave functions in Table 3.4 are the
isoscalar factors for the reduction 8 x 8 . We have thus found that
I he S U(3) Clebsch-Gordan coefficients factor as follows :

fL1
( VI
/1-2
v2
fLy)
V
= C(T
1
T T'T
2 ,
T T)( fLl /1-2 fLy)
1z 2z z T1 Y 1 T2 Y 2 TY
I (4.8)

As t he S U(2) Clebsch-Gordan coefficients are widely available, we

can devote our attention to the isoscalar factors. First we note
I h at they can be chosen real, since the matrices T ±' V ± ' U ± are
real (of course the S U(2) Clebsch-Gordan coefficients are real).
If we suppose that the functions appearing in Eq. (4.3) are
orthonormal (in particular the independent functions dis tin-
I!; uis hed by the label Y should be orthogonal), then the S U(3)
Clcbsch-Gordan coefficients form a real orthogonal matrix
(unitarity and reality). Thus we can invert (4.3) simply by writing

(4.9)
102 INTRODUCTION TO UNITARY SYMMETR

Here we have made use of the orthogonality relations:

v~ (~ll
fL2 fLY)(fLI fL2 fL'Y')
v' = 8~~,8yy,8yy' (4.1
V2 v VI V2

~ ell }L2
V2
fLY)(fLI,
v VI
fL2
V2
, ~Y) = 8Y1Y1,8Y2Y2' (4.1

Because of the similar relations for the S U(2) Clebsch-Gorda

coefficients (11) these equations can be simplified to the followin
orthogonality relations among the isoscalar factors:

4.3 Orthogonality Properties of the Representation Matrices

Consider the d~ basis functions IcPy(~» of the irreducible repr
sentation fL. To be definite, we consider the functions IcP'<~» to b
quantum-mechanical state ve:tors and define the representatio
matrices by analogy to Eq. (1.83).

w.<~» = (')(ex) IcPv(~» = 2:lcP/~» DA'<~)(ex) (4.1

.<
By virtue of the compactness of S U(3) the matrices D(~)(
induced by the unitary operator (')( ex) may always be taken
unitary and finite-dimensional. As in Chapter 1, we may introduc
a field operator cPy (~) whose adjoint creates the state vector IcP'<~
Thus cP'<~)t transforms as

(')(ex)cPV(~)t(')-I(ex) = 2: cP}~)tD.<y(~)(ex) (4.1

.<
that is, the column vector <p(~), composed of the cPv(~)' transform
as
(4.1
GENERAL PROPERTIES OF REPRESENTATIONS OF 8U(3) 103

If we write O( a) in the form exp (ia· G), where G is the Hermitian

unitary spin operator, then the matrix D(u) is given by

(4.17)

This result is most easily seen by expanding (4.14) for infinitesimal

a. Writing the unitary matrix (4.17) in the form (G(u) = G(u)t is
a H ermitian du X du matrix)

(4.18)

we note that the transformation property (4.16) of the field

operator becomes
(4.19)

T he usual orthogonality relations for the representation

rnatrices of finite groups hold for compact groups with an appro-
priate definition of volume element (2). The density function dp
will depend on the eight real parameters (for instance, a) specify-
ing an 5U(3) transformation. The normalization can then be
dlOsen so that

(4.20)

where d u is the dimension of the representation D(u). The unitarity

condition is, of course, expressed by

L: Di/U)(a)D,,/U)'(a) = ai"
j
(4.21)

A useful alternate form for the Clebsch-Gordan series will now

I.e proved. We subject (4.9) to the transformation O( a), obtaining

(4.22)
104 INTRODUCTION TO UNITARY SYMMETRY

and taking advantage of the orthogonality of the basis functions

gives explicitly the reduction of the direct product D(/Ll) x D(/L2):

Similar relations are well known in the study of the rotation group
Tables of isoscalar factors are given in Appendix 1.

4.4 Irreducible' Teusors; Wiguer-Eckart Theorem for SU(3)

An irreducible tensor operator T/") transforming according to

the irreducible representation (jL) is defined in the usual way hy

(')(a)T}/L)(')(a)-1 = 2: T//L)DA//L)(a) (4.24)

A

where the index v labels the various components of the tenso

T(Il). Sometimes it is useful to convert (4.24) to the equivalen
infinitesimal form, as was done for the vector operator defined in
Eq. (2.94). In that example (jL) = 8 and v runs from 1 to 8.
A Wigner-Eckart theorem similar to tha~ for SU(2) [cf. Eq
(1.84)] holds in SU(3). The proof is very similar to that given
for the former group [see, for example, Tinkham (13), Chap. 5]
Consider the matrix element of the irreducible tensor opetator Tv (Il
hetween states I4>Vl (Ill», I4>V2 (1l 2 » transforming according to the
irreducible representations D(llll, D(1l 2 ). The theorem states tha
this matrix element is given by

(4.25)

where the reduced matrix element <jLzll T(Il) IIjLI)Y depends only
on the representations involved. All the dependence on the
quantum numbers VI' vz, V is given hy the SU(3) Clebsch-Gordan
coefficient. The number of terms in the sum is equal to the
number of times the representation D("2) is contained in the
reduction of D(lll ) x D(Il). The left-hand side of (4.25) is equal to
GE NERAL PROPERTIES OF REPRESENTATIONS OF 8U(3) 105

«(O(a.)4> V2(#'») 1O(a.)T/#)O(a.)-1 IO(a.)4> v1(#1»

2:
\ll'V2'V'
(4.26)

on using (4.14) and (4.24). The matrix elements involved are

merely numbers independent of the group parameters a.. In order
to use the orthogonality theorem to advantage, we use Eq. (4.23)
to express (4.26) in terms of products of two representation
matrices:

We now integrate over the parameter-space using the group

weight function; from Eq. (4.20) we find ,\ = fL2' P = V2', a = V2
in the sum
( 4) v2 (#2) 1 T/#) I4>V1 (#1»

= 2:Y(fLlVI fL

V
fL2Y)
V2
.!.
d/1. 2: (fLI,
VI
fL2

V
, fL2,Y)
V2
\l1'V2'\I

X ( 4) V2,(/1.2)1 T,,</1.) l4>v1 ,(/1.» (4.28)

Equation (4.28) clearly has the form promised in Eq. (4.25).

Note that two independent constants are required to give all
I he matrix elements of a vector operator in the regular repre-
scntation. The reader will notice the connection with the dis-
cussion of Section 2.6. Lurie and MacFarlane (14) have given
explicit expressions for all nonvanishing matrix elements of the
vector operator.

4.5 Symmetry Properties of the SU(3) Clebsch.Gordan Coeffi.

cients
In practical applications, one often has to pay attention to
conceptually simple but time-consuming questions involving the
106 INTRODUCTION TO UNITARY SYMMETRY

symmetry properties of generalized Clebsch-Gordan coefficients.

For example, if one adds up unitary spins in different order the
result is essentially the same although some extra minus signs
may appear. Recalling the conventions made in Section 4.2, we
note that

(4.29)

where gl is ± 1 and may be found most directly by explicit

inspection of the functions. As gl is independent of "I> "2' and '"
its value may be found from the state of highest weight in the
representation jLy.
By utilizing familiar symmetry properties of the S U(2) Clebsch-
Gordan coefficient, relations between the isoscalar factors are
obtained. Employing the relation

we obtain from (4.29) the result

jLl
Tl Y l
I TY
jLy)

(4.31)
As a by-product of (4.31) we obtain the special result

T = odd
T = even
(4.32)

In addition to (4.31), the following results involving the com-

plex conjugate representations have been derived by de Swart
(10) for irreducible representations occurring in the eightfold way.
When only one representation occurs more than once (label y),
we are given the relation

jL2Y'
-1'2
*) (4.33)
GE NERAL PROPERTIES OF REPRESENTATIONS OF 8U(3) 107

where the symbol - v denotes the set (T, - T 3 , - Y) if v stands for

( T , T 3 , Y). The S V(2) relation

permits the simplification of (4.33) to an expression for the iso-

~c alar
factors

A simpler relation is given by

(4.36)

This may be reduced to the following relation between isoscalar

fac t ors

on using

C( Tl T2 T; T 1z T 2z T z)
= (-I)Tl+T2-TC(TlT2T;-Tlz,-T2z, - Tz) (4.38)

T he phase factors tl> t2, t3 have been given by de Swart (10) for
the decompositions 8 x 8, 8 x 10, 8 x 27, 8 x 10·, 10 x 10,
10 x 10·, and 10· x 10·. They are reproduced in Table 4.1.
108 INTRODUCTION TO UNITARY SYMMETRY

TABLE 4.1
Phase Factors Involved in the Symmetry Properties of the SU(3)
Clebsch-Gordan Coefficients.

27 -1 1 35· 1 -1 1
10 -1 -1 1 27 -1 -1 1
8 10·
10· -1 1 10· -1 -1 -1
8 8
81 1 8 -1 -1
8. -1 -1 -1
1 -1 1 28 1 1
35 -1 -1
35 1 -1 10 10
27 1
27 -1 1
8 10 10· -1 -1
10 -1 -1 -1
8 1 1-1
64 1 1
64 1 -1 1 27 -1 -1
35 -1 -1 10 10·
8 1
35· -1 1 1 -1 -1
271 1 1
8 27
272 -1 -1 -1 28· -1 1
10 1 -1 -1 35· -1 -1
10· 10·
10· 1 -1 27 -1 1
8 -1 10 -1 -1

4.6 Constrnction of Representations by Tensor Analysis

In Chapter 3 we described how one constructs irreducible
representations by reducing the direct product (3)1' x (3*)Q.
Although the use of the two three-dimensional representations in
this manner is economical, it is not essential since 3* is the anti-
symmetrical combination of two 3's. Hence we could obtain the
same representations (and some additional ones) by studying
(3)1'+2Q at the expense of introducing a greater number of func-
tions. We can further consider the problem of reducing (3)N as a
special case of the general linear transformation on three com-
ponent objects. In fact, it is no more difficult to study the general
linear group in n dimensions GL(n), defined by the transformation
law on n objects 1jJ!
(4.39)
l;ENERAL PROPERTIES OF REPRESENTATIONS OF SU(3) 109

for an arbitrary complex matrix a. Tensors with respect to

GL(n) are defined as usual by (r factors)
(4.40)
The representation a x a x ... x a (r times) is reducible. The
irreducible tensors, which generate the irreducible representa-
Lions, are constructed by applying the Young symmetrizers to the
Lensors T. The crucial facts that make the symmetrized tensors
irreducible tensors are two: (1) the index permutations commute
with the transformation (4.40); (2) for gener al a/, no other
oper ation commutes with the transformation. In this way the
theory of the symmetric group is seen to be of direct relevance for
the representation theory of GL(n) and its various subgroups.
This theory is developed fully in Chapter 10 of the book by
Ha mermesh (2). Only a few salient points are reviewed here. A
succinct summary of the properties of Young diagrams, tableaux,
and symmetrizers may be found in Appendix D of the book by
Messiah (15). The main point of interest is that when the trans-
for mation matrices in (4.39) are required to be unitary and uni-
modular, the representations irreducible for GL(n) remain
irreducible for SU(n). (This is not the case for the subgroup
O(n), the orthogonal group in n dimensions.)
The rth rank tensors of (4.40) span an nT-dimensional vector
space. Let us represent Eq. (4.40) symbolically by
T' = AT (4.41)
and any permutation of the indices by P. Then one easily finds
PT' = PAT
(4.42)
= APT
Hence sums and products of P's will also commute with the
transformations of GL(n). The nT-dimensional vector space is
decomposed into invariant, irreducible subspaces by the Young
symmetrizers, projection operators f!jJ a obeying
f!jJaf!jJP = 8apf!jJa
(4.43)
no INTRODUCTION TO UNITARY SYMMETRY

With each symmetrizer (not defined here) we can associate a so-

called Young diagram.
We reexamine the examples of Chapter 3, paying special atten-
tion to the symmetry properties. The simplest case of all is the
familiar process of representing a second-rank tensor as the sum of
its symmetric and antisymmetric parts (Eq. 3.19)

ifi(I )if/(2) = t( tfii(l )tfif(2) + tfif(l )tfii(2»

+ t(tfil(I)tfif(2) - tfif(IW(2)) (4.44)

which for SU(n) corresponds to the reduction of n x n into

irreducibk representations of dimension tn(n + 1) and tn(n - 1)
(for example, 3 x 3 = 6 + 3*). In the reduction of 3 x 3 x 3 it
was found that the resulting representation 10 was completely
symmetrical, while the two 8 representations have mixed sym-
metry and the I-dimensional representation is the completely
antisymmetric combination of tfiitfiftfik. The mixed symmetry of
Eq. (3.40) (and of all the associated basis functions) is that of the
operations of antisymmetrization of tfii(I)tfij(2)tfik(3) in labels 1 and
3 followed by symmetrization in labels 1 and 2.
Thus, using the antisymmetrical tensor elfk (== eifk ), we can
convert two 3 representations tfit to one 3*(ti) by contraction
ti = eijktfiltfik (4.45)

or two 3* representations (t j ) to one 3(tfil)

tfii = eiiktjtk (4.46)

Here we are distinguishing upper indices (contravariant vectors

'" 3) from lower indices (covariant vectors '" 3*). This is useful
because the operation of contraction of an upper with a lower
index reduces the rank of a mixed tensor [see Eq. (3.5)] by two.
The basic example is the formation of a scalar

(4.47)

[Note: In tensor analysis it is convenient to forgo the Condon-

Shortley phase convention; thus tj = (tt)*. Alternatively (16) one
can use the phase convention (2.79) for 3* and compensate by
GE NERAL PROPERTIES OF REPRESENTATIONS OF SU(3) III

contracting with g/ = diag (-1,1,1) instead of the Kronecker

delta; cf. Eq. (3.29).]
Using the unitary-unimodular property of the transformation
matrices a, one easily verifies that the tensors

(4.48)
are unaltered by S U(3) transformations. Alternatively, these
operations (contraction and antisymmetrization) commute with
the SU(3) transformations on the mixed tensors and can be used
to reduce the typical mixed tensor Eq. (3.1). The latter has p
upper and q lower indices. Contraction with 8a a results in a tensor
with p - I upper and q - I lower indices:
~'8 . ... _
b ... 1e -
8a a~ aB .. .•
ab ... 1e (4.49)
The antisymmetric tensors eWe and eijle are used to raise and lower
indices ; for example,
~"an . • .• co _ ecoab~aB .• .•
c . .. Ie - ab ..• k (4.50)
has p + I upper and q - 2 lower indices and
~'" Y • . • v _ e ~aB ... v (4.51)
ab . •. lel - la/l ab • •. 1e

has p - 2 upper and q + I lower indices.

The tensor ~ is surely not irreducible since the operations
(4.49)-(4.51) yield tensors having transformation properties dis-
tinct from the original tensor ~. If, however, we modify ~ to a
for m in which the operations (4.49)-(4.51) yield zero, then we
shall have obtained an irreducible tensor. The procedure is then
to add terms to ~ in order to obtain a new tensor which is totally
symmetric in the upper indices and also in the lower indices, as
well as traceless. (By virtue of the symmetry condition, the trace
is defined on only one pair of indices: Tr ~ = La ~~g ::: ~.)
We shall now prove the dimension formula (4.1) by counting
the number of independent components possessed by a tensor
having the special properties mentioned in the preceding para-
graph. First ignore the trace condition. For a fixed set of lower
indices there are t(p + l)(p + 2) independent components. To
112 INTRODUCTION TO UNITARY SYMMETRY

show this we arrange the indices of a given tensor from left to

right in order of increasing value. This avoids overcounting that
might otherwise occur due to the symmetry of the tensor. Suppose
that there are nones (n can be anyone of the sequence 0,1, ... , p).
The remaining p - n places are occupied by 2's and 3's. There
can be no two's, one two, etc., up to (p - n) two's, i.e., a total of
(p - n + 1) possibilities. The total number of independent
upper index components is thus
p

Nu = L (p -
n=O
n + 1) = Hp + l)(p + 2) (4.52)

In exactly the same way the number of independent lower index

components is
N/ = t(q + 1)(q + 2) (4.53)
The corresponding symmetrized tensor <1>s therefore has
NuN/ = Hp + l)(q + l)(p + 2)(q + 2) independent components.
Suppose <1>s is further required to have zero trace. The trace is a
symmetric tensor with p - 1 upper and q - 1 lower indices. All
the Nt = tpq(p + l)(q + 1) components of <1>s must vanish.
Hence the dimensionality of the associated representation is
d(p,q) = NuN/ - Nt = t(p + l)(q + l)(p +q+ 2) (4.54)
The regular representation 8 has components given by
(4.55)

where (1jI,1jI) = ljIiljli. We can also regard Ti j as a traceless 3 x 3

matrix (rows associated with i and columns with j).
For example, if ljIi is associated with the quark field qi and 1jI!
with iji' (4.55) is equivalent to
t(2ijl ql - ij2q2 ij2ql ij3ql
- ii3q3)
iilq2 t(2ij2q2 - iil ql q3q2
T= (4.56)
ij3q3)
qlq3 ij2q3 !(2ij3q3 - ql ql
- 9.2q2)
G ENERAL PROPERTIES OF REPRESENTATIONS OF SU(3) 113

Making the identifications of Eqs. (3.61), (3.62), and (4.56)

reduces to the matrix f!jJ of Eq. (3.65) previously derived from a
different point of view.
The ten-dimensional representation is the completely sym-
metrized product of "i(I)"i(2).pk(3) [cf. Eq. (3.38)].
Next we give a simple example illustrating the reduction of
direct products of irreducible representations. The direct product
3 X 8 has the set of basis functions

(4.57)
°
where Tii = by (4.55).
We wish to express (4.57), which is not traceless and does not
h ave definite symmetry properties in upper indices, as a sum of
terms having these desired properties. One proceeds in two steps:

</>tT\ = tWT\ + </>iTt k) + tWTik - </>iTik)

= [tWTi k + </>iTt k) - i(8 tk</>ITi, + 8i k</>ITt,)] (4.58)
+ [tWT\ - </>iTtk) - H8 tk</>ITi, - 8\</>ITi,)]
+ [i8 tk</>ITi, - t8 i k</>'Tt,]
The three terms in square brackets are traceless tensors. The first
is symmetric in its upper indices and hence gives rise to D(2,1),
or 15. The second term is antisymmetric in i and j; there are six
independent components. Denoting it by Aiik, one sees that its
independent components can be put in one-one correspondence
with the tensor </>'k = £/iiAti k + £ktiAii" a symmetric covariant
tensor of rank 2. Thus the representation generated by Aiik is
D(0,2), or 6*. Similarly, the last term corresponds to D(I,O), or 3.
This example should remind the reader that although a given
representation can be obtained by symmetrization of the primeval
t ensor If>, tensors of more complicated symmetry give rise to the
same representation. A good exercise is to reduce 8 x 8 using this
technique. The lazy reader will find the details given in Okubo's
notes (17).
A simple prescription for reducing direct products (using
tensors) has been given by Coleman (18). A graphical method has
been given by Speiser (19) [see also de Swart (10)]. Further details
114 INTRODUCTION TO UNITARY SYMM

on the analysis of representations using tensor analysis c

found in references 1-4, 7, 16, and 17.
We conclude by giving the definition of a mixed rank
tensor operator in the present context. This operator corres
to the" vector operator" defined by Eq. (2.101). The eight
generators transforming as Eq. (4.55) were given in Eq. (
The corresponding commutation rules were given in Eq. (
The mixed tensor operator of rank 2 hence satisfies (by defin

(This definition holds also if we do not require Tkl to be trac

in any event T\ commutes with the AI, and is a multiple
identity.)
Contraction of two such tensors on one index yields an
tensor of the same type; thus UI, = st" Tk, satisfies (4.59)
T --J>- U. Further ut t is a scalar within an irreducible repre
tion. Thus we note that the operators

all transform as scalars. Since only two Casimir operators ex

the group S U(3), one expects that only two of the sequence
are independent. In fact all "powers" of A higher tha
second and third of (4.60) can be expressed in terms ()f the
two (17).
An alternative form of the Wigner-Eckart theorem fo
octet operator of SU(3) is given by Okubo's formula for the
less tensor Tt,: Within an irreducible representation Tt, ha
form (17, 20)

Tt, = a(p,q)Atj + b(p,q)[At"A"j - t8 tiA.A)]

where a(p,q) and b(p,q) depend on the representation.

References
1. H. Weyl, Classical Groups, Princeton University Press, Princeton, New
1946.
2. M. Hamermesh, Group Theory, Addison-Wesley, Reading, Mass., 1962
GE NERAL PROPERTIES OF REPRESENTATIONS OF SU(3) 115

3. L. C. Biedenharn, J. Malh. Phys., 4, 436 (1963).

4. G. E. Baird and L. C. Biedenharn, J. Math . Phys., 4, 1449 (1963); 5, 1723
(1964); 5,1730 (1964).
5. S. Gasiorowicz, Argonne Nat!. Lab. Rept. ANL 6729 (1962), unpublished.
6. R. E. Cutkosky, Ann. Rev. Nucl. Sci., 14, 175 (1964).
7. R. E . Behrends, J. Dreitlein, C. Fronsdal, and B. W. Lee, Rev. Mod. Phys.,
34, 1 (1962).
8. L. C. Biedenharn, Phys. Leiters, 3, 69 (1962).
9. G. Racah, in Group Theoretical Concept. and Methods in Elementary Particle
Physics, F. Gursey, Ed., Gordon and Breach, New York, 1964.
10. J. J. de Swart, Rev. Mod. Phy•. , 35, 916 (1963).
11. M. E. Rose, Elementary Theory of Angular Momentum, Wiley, New York, 1957.
12. A. R. Edmonds, Proc. Roy. Soc. (London), A268, 567 (1962).
13. M. Tinkham, Group Theory and Quantum Mechanics, McGraw-Hill, New York,
1964.
14. D. Lurie and A. J. MacFarlane, J. Math. Phys., 5, 565 (1964).
15. A. Messiah, Quantum Mechanics, North-Holland Pub!. Co., Amsterdam, 1963,
Vo!' II.
16. N. Mukunda and L. K. Pandit, J. Math. Phys., 6, 746 (1965).
17. S. Okubo, Lectures on Unitary Symmetry, University of Rochester, 1964,
unpublished.
18. S. Coleman, J. Math. Phys., 5, 1343 (1964).
19. D. Speiser, in Group Theoretical Concepts and Methods in Elementary Particle
Physics, F. Gursey, Ed., Gordon and Breach, New York, 1964.
20. S. Okubo, Prog. Theoret. Phys. (Kyoto), 27, 949 (1962).
 Chapter 5

SU(3)-Invariant Vertices and Amplitudes

5.1 Constrnction of Invariant Couplings

Using the results of the preceding sections, it is relatively
simple to construct interactions among the various unitary multi-
plets which are invariant under the transformations of S U(3).
First we derive the Yukawa couplings of the pseudoscalar meson
octet P a to the baryon octet Bb (a,b = 1, ... ,8). (Clearly these
considerations hold for the more general P B B vertex.) In Section
2.5 we found that there are two independent sets of functions
transforming according to the regular representation, jj Flc Band
BDkB (where the Ba and . P a transform via the "Hermitian"
regular representation, and are given explicitly in Eqs. (3.63) and
(3.64)). We can form invariants by contracting these expressions
with the pseudoscalar meson octet P k • (That these are the only
invariants can be seen from the fact that the identity representa-
tion occurs only twice in 8 X 8 x 8, since 8 x 10, 8 X 10*, and
8 X 27 do not contain 1.)
Inserting the Dirac matrix Y5 to account for the pseudo scalar
character of the mesons, the interaction Hamiltonian density has
the form
(5.1)
so that only two independent coupling parameters enter. This
should be compared with the relatively unrestrictive result
obtained using isospin invariance, Eq. (1.51).
It is customary to introduce a "mixing parameter" ex which
measures the relative strength of the symmetric and antisym-
metric couplings (D and F types, respectively) by
116
S U(3)-INVARIANT VERTICES AND AMPLITUDES 117

(5.2)
g2 = 2g(1 - a)
so that (5.1) has the form
:Ie = 2gBys[aD + (1 - a)F]B.P (5.3)
The overall strength is measured by g, which turns out to be just
the usual 7T-N coupling constant g2/47T ~ 15. This result, as well
as expressions for all the coupling constants of Eq. (1.51), can be
fou nd by writing out (5.3) in detail.
As an illustration of this, we extract from (5.3) the TTN com-
ponent. From Eq. (3.36) it is clear that PI' P 2, P 3, and B 4, . . B7
are the only components giving rise to the 7T N interaction. Con-
sider the F terms in (5.3). Omitting Ys again for notational ease,
the 7T N part comes from

[F4/(B4 B 7 - B7 B 4) + FSSl(BsBs - BsBs)]Pl

+ [F4S 2(B4 B S - BsB4) + 2
FS7 (BsB7 - B7 B S)]P2
+ [F4S 3(B4 B S - BsB4) + FS7 3(BsB7 - B7BS)]P3 (5.4)
Substituting B4 --+ p/V2, Bs --+ ip/V2, Bs --+ n/V2, B7 --+ in/V2
and the values of the Fjkt from Table 2.1, (5.4) reduces to

UV2pn7T+ + V2iip7T- + (pp - iin)7TO] (5.5)

which is tiV'tN.1t with 7T+ = (7Tl ± i7T2)/V2, 7T3 = 0
7T • The D
term gives

[D4S l (B4 B S + BsB4) + DS7 l (BsB7 + B7 B S)]Pl

+ [D47 2(B4 B 7 + B7 B 4) + DS6 2(BsBs + B s B s )]P2
+ [D44 3B4B4 + D ss 3 BsBs + D SS 3 BeBe + D77 3B 7B 7]P3
(5.6)
which when worked out coincides with (5.5). Thus the TTN part of
(5.3) is simply
2g[atlV'tN.1t + t(1 - a)lV'tN.1t] = gfihN.1t (5.7)
independent of a.
118 INTRODUCTION TO UNITARY SYMMETRY

The other couplings among the various isospin multiplets can

he worked out in a similar way. The resulting PBB coupling
constants are given in Table 5.1, where the D, F, and total con-
tributions are indicated separately. (The units are chosen so that
gnNN == g = 1.)
In Fig. 5.1 we have plotted gpBB2jg2 as a function of a, in order
to reveal more clearly the relative values of the various couplings.
The value of a has to be found from other considerations:
comparison with experiment, dynamical models, or more restric-
tive symmetry requirements. The "empirical" values of the
various coupling constants are almost always found with the aid
of simple theoretical models, hopefully applied with discretion, to
reactions in kinematical domains where the simplifying assump-
tions might be true. These models are subject to quantitative
doubt, since they almost always boil down to "hindsight" per-
turbation theory. Nevertheless, the cumulative qualitative suc-
cess of dispersion theory models of particle collisions is sufficient

TABLE 5.1
Isospin Couplings of the Pseudo scalar Octet Mesons to the Octet Baryons as a
Function of the Mixing Parameter a.

Coupling Type D Contribution F Contribution Total

TrNN a I-a
'TrEB -a I-a 1 - 2a
TrAE 2a/v'"3 0 2a/v'"3
TrEE 0 2(1 - a) 2(1 - a)

KNA -a/v'"3 - v'"3(1 - a) - (3 - 2a)/v'"3

KEA -a/v'3 v'3(1 - a) (3 - 4a)/v'"3
KNE a -(1 - a) -(1 - 2a)
KEE -a -(1 - a) -1

"INN -a/v'"3 v'3(1 - a) (3 - 4a)/v'"3

"lEE -a/v'"3 -v'"3(1 - a) - (3 - 2a)/v'"3
TJAA - 2a/v'"3 0 -2a/v'"3
"lEE 2a/v'"3 0 2a/v'"3
SU(3)-INVARIANT VERTICES AND AMPLITUDES 119

Q-

Fig. 5.1. The isospin Yukawa couplings of the octet pseudoscalar mesons to the
baryon octet depend on the mixing parameter a, the latter measuring the ratio
of D to F couplings in Eq. (5.3). The values of the couplings are shown as a function
of a. The mixing parameter is not determined by pure 5U(3) symmetry con-
siderations.

enough to deserve consideration. A further difficulty in compari-

son of (possibly model-dependent) coupling constants with results
such as displayed in Table 5.1 lies in the fact that we have not
taken into account the possibly large symmetry breaking effects.
For all these reasons one can, at best, only expect qualitative
success.
The analysis of the binding of A and .E hyperons in hyper-
nuclei indicates that the 1TA.E coupling is comparable to the 1TNN
coupling but that the 1T.E.E coupling is rather weak (1, 2). From
Fig. 5.1 one sees that this already puts the acceptable value of ex
above about 0.6, and more likely about 0.75. A value in this
region is further favored by the relatively small '7 production
cross section in 1T N collisions (3). Some information about the
120 INTRODUCTION TO UNITARY SYMM

KN Il and KNE couplings can be obtained from the analy

reactions such as yN -i>- KIl, KE; 1TN -i>- KIl, KE. These rea
are rather complex and many of the calculations are not in
impressive agreement with each other. However, typical r
indicate that gKNA 2 is somewhat larger than gKNX 2 , but still
smaller than g"NN 2 • Typical analyses of this sort are tho
Kuo (4), and Hatsukade and Schnitzer (5) for the photopr
tion reactions mentioned above. For example, Kuo finds gKN
about equal to gKNx2j41T :;;; 4. Warnock and Frye (6)
analyzed low energy KN scattering in detail using disp
theory. They are able to obtain very good fits using values
couplings compatible with S U(3), with a in the range ment
above. Little evidence is available concerning the other
couplings. The results are in satisfactory agreement wit
requirements of SU(3) symmetry.
In Chapter 7 we shall discuss the "bootstrap" model
baryon octet and decuplet. This model portrays the baryon
as primarily a bound state in the P B channel, held togeth
attractive forces due to decuplet exchange. Similarly, the de
is supposed to be a composite of P and B, held together m
by B exchange. The dynamical equations of this model giv
to a self-consistency condition (7, 8) which determines a
about 0.7. In this model, the occurrence of the P 3 / 2 de
rather than the 27 -dimensional representation also restrict
a rather narrow range. [In a more complete analysis, includin
effect of vector meson exchange and actual masses, the repre
tion 27 is suppressed (10).]
Next consider the coupling of an octet of vector m
to the baryons. For purposes of normalization and refe
to easily accessible experimental information, we write
the isospin invariant couplings of the p meson to the nu
As in the electromagnetic case, there is a vector (" elec
coupling

and an effective tensor (" magnetic") coupling:

S U(3)-INVARIANT VERTICES AND AMPLITUDES 121

"t£J
,J\.,,,NNT -_ g"NN
4M N- auv't' N . (<>U
u p V - u<>v pU) (5.9)

[Notation: The y-matrices and metric are characterized as follows:

Yo t= Yo; Y/ = -YI' i = 1,2,3; Ys = Yoyly2y 3, Yst = -Ys;
auv = i[Yu,Yv]/2, goo = -gji = 1, i = 1,2,3.]. In (5.8) the
coupling constant iPNN has been accompanied by t't'
rather than
't'
so that the equality i"NN = i""", the latter defined by

(5.10)

gives a natural description of the p being coupled to a conserved

isospin current (11). In Eq. (5.9) the normalization of g"NN has
heen chosen so that in the analogous electromagnetic case g
would be Ke, where K is the anomalous magnetic moment in units
of e/2M and e2 /4TT = 11137.
The coupling constants i"NN and g"NN can be related to the
residues of the annihilation amplitudes NN ---J> TTTT, Tl and T2
used by Frazer and Fulco (12) (in the p-dominant approximation)
by computing the perturbation graph of Fig. 5.2 from Eqs.
(5.8)-(5.10). The result is (13)

i"""i"NNI4TT = - 3Yl
(5.11)
i"""g"NNI4TT = - 3MY2
where 1m Ti(t) == TTYI o(t - m" 2) in the pole approximation. Ball
and Wong (14) estimate (on the basis of their study of the nucleon
isovector form factor) that Yl ~ -1.0 and MY2/Yl ~ 1.83. The
former gives a resultanti"NNi",,,,/4TT in fairly good agreement with
the" universal" value, if one takes i""" from the p width, given
by

where k" is the center-of-mass pion momentum. Studies of low-

energy TTN scattering (15) suggest similar values ofi"NN and g"NN'
and an explicit evaluation of the residues using a particular
122 INTRODUCTION TO UNITARY SYMM

Fig. 5.2. Nucleon annihilation into two pions through a p-intermediate

shown. The associated perturhation calculation yields the connection am
constants given in Eq. (5.11).

model of the annihilation reaction (16) gives about the

result.
The SU(3) invariant VBB couplings can now be construc
analogy with the PBB interaction. For the vector (el
coupling we have

where V transforms as Pin Eq. (3.63), and for the tensor

netic) coupling

(We might remark that if unequal baryon masses are perm

then a third Lorentz-invariant coupling can be formed: see r
Concrete information about VBB coupling constants is
scarce. There is a substantial current prejudice in favor o
dominantly F-type couplings (aE = 1) for the Yll term
largely D coupling, with aM = a ~ .6 - .7, for the ma
coupling. The situation is further complicated by the
mixing problem, discussed in Chapter 6.
S U(3)-INV ARIANT VERTICES AND AMPLITUDES 123

We note in passing that there is no corresponding freedom in

the Sakata model P B B couplings. Denoting the Sakata triplet
(p ,n,A) by ~, we have the obviously unique coupling of the octet
P to the triplet:

(Sakata model) (5.15)

where g is again the 7rNN coupling constant. Writing (5.15) out

in full gives

- 1 - 2 -
:Ie = g [ NYs'f:N .1t + V3 NYsNT} - V3 AYsAT}

+ V2(.NYsKA + H.C.)] (5.16)

Simjlar constructions hold for the electric and magnetic couplings

of a vector meson octet with a Sakata triplet.
A commonly used expression for SU(3) invariant PBB coup-
lings involves the trace of products of the 3 x 3 matrices intro-
duced in Section 3.4. For example, the expressions

(5.17)

are invariant under simultaneous unitary transformation of $

and fJJ by the matrix U = exp (ia· ")../2). Expanding the matrices
in terms of the Aj and the B j , Pi as in Section 3.4 we use the
trace condition and definitions [Eqs. (2.22)-(2.24)] as follows:

Tr [3333& + $fJJ$] = 2 -3/2 Tr [BkAkBiPj(AjAj + AjAj)]

= 2-3/2BkBjPjTr[Ak(!lljj + 2d jjl A,J
(5.18)
= 2-1 /2BkBjPj djjl Tr AkAI
= V2BDB.P
For the other combination we obtain

Tr [33$9' -$9'$] = - V2ifkjjBkBjPj

(5.19)
= V2BFB·P
124 INTRODUCTION TO UNITARY SYMMETRY

Therefore, we can write the P B B interaction in one of the two

equivalent forms (Y5 implicit)
:Ie = 2gB(aD + (1- a)F)B . P

V2g[a Tr ($$fJ! + If,fJ!$) + (1 - a) Tr (3l$fJ! - $fJ!$)

(5.20)

Under the particular R transformation fJ! ---+ fJ!T, $ ---+ $T, If,---+
SF the two types of coupling behave as follows:
Tr $($fJ! ± fJ!$) ---+ Tr $T($TfJ!T ± fJ!T$T)
= Tr [± ($fJ!$)T + (fJ!$If,)T]
= ± Tr [33($fJ! ± fJ!$)] (5.21)

sInce the trace of a matrix is unchanged by transposition. As

remarked previously there is no reason to believe in R invariance
of the strong interactions. Indeed, the double occurrence of the
regular representation is one of the especially interesting features
of the eightfold way.
The coupling of a unitary singlet meson S with the baryon
octet B is clearly
EBS (5.22)
leaving the Lorentz structure unspecified.
N ext consider the interaction between an octet of vector mesons
V and two identical pseudoscalar meson octets P. In the Her-
mitian basis the octet fields are "real": pt = P and vt = V.
Thus the four-vector quantity

(5.23)

is Hermitian (reordering as usual) and antisymmetric in the par-

ticle labels b, c. The coupling is thus of the form (recalling that
Oil VIl = 0)
(5.24)

where the gbc a are real constants that have to be chosen to make
(5.24) invariant under SU(3} transformations. We have learned
SU(3)-INVARIANT VERTICES AND AMPLITUDES 125

that there are only two such sets of coefficients, iabc and dabc ' The
symmetric coupling can be eliminated by invoking a generalized
Bose symmetry for the P.S. mesons, which states that Je is
invariant under the permutation pb f---* pc. Thus gbc a = - gcb a and
+--+
Je ppv = gppviabc V" apbO" pc (5.25)
A byproduct of this argument is that a unitary singlet vector
meson cannot couple to two P octets. A similar argument leads
to the form of the V2 P coupling (here ea/JyO is the completely anti-
symmetric tensor 130123 = 1)
(5.26)
Lipkin has emphasized the interesting point that S U(3) invariant
trilinear meson interactions do not forbid any reactions not
already forbidden for other reasons (17). This circumstance
arises from the occurrence of both symmetrical (d) and anti-
symmetrical (f) types of coupling coefficients in SU(3). Com-
parison of Eqs. (5.10) with Eq. (5.25) indicates that gppv =
tip",,' An equivalent expression for (5.25) involving the trace of
the meson matrices f?l and 'lJ is

(5.27)

Similarly, the trace form of the VVP coupling is

Jevvp = g;; Tr f?l(oa'lJ/JOy'lJo + Oy'lJooa'lJ/J)ea/JY6 (5.28)

It is very useful to express the P PV interaction, Eq. (5.25) or

(5.27), in terms of isospin couplings. One finds, after a little
manipulation, the expression
+--+ +--+ +--+
Je ppv = gppv{p,,·(n x o"n + iKt-ro"K) + (iM"t-rK.o"n + H.c.)
+--+ +--+
+ V3(iM"tKo,,1) + H.c.) + V3irp"Kto"K} (5.29)
Here we have denoted the K*(888) doublet by M to prevent over-
crowding. The doublet M is composed of col (M+ ,MO), while Mt
126 INTRODUCTION TO UNITARY SYMMETRY

stands for (M- ,MO). As alluded to before and discussed in the

next chapter, the symbol cp appearing in (5.29) is to be inter-
preted as a mixture of the physical rp-w states. Similarly, working
out Eq. (5.26) or (5.28) yields

'W
<.I\'vvp 2
-_ gvvpe apyo{ VS 1t.
.J.
Pa.p'!'y.o + 1t. M a.D t 't M y.o

1 1
- VS (KtMa./J + Ma.ptK)rpy.o + VS1JPa.p·Py.o
1 1
- VS M a.fJ tM y.o1J - VS·CPaAy.o1J (5.30)

In writing (5.30) we have introduced the shorthand notation of

expressing the derivative o fJ Va as V a.fJ • The complexity of the
VVP coupling is less than it seems from the great number of
indices. The coupling is really of the form that describes the
7T Oyy vertex, which can also be written as 7T oE · H or in the" co-
variant" form 7TOeaflYOoaAi\ Ao' (AIL is the electromagnetic four-
potential. )
Another method for constructing invariant vertices is to
employ the SU(3) Clebsch-Gordan coefficients. DeSwart has con-
structed the P BB couplings in this manner; Martin and Wali (8)
have derived the coupling of a J = 3/2 decuplet to the baryon
and meson octets, expressed in terms of the isospin couplings.
(In the latter case there is no mixing parameter, since the
identity representation occurs only once in the direct product
10 x 8 x 8.)
In the preceding we have not been concerned about whether
the various couplings are "fundamental" or phenomenological.
If no such field operators exist, then one should convert the
effective vertices to momentum space, changing the coupling
constants into form factors.
There is one interesting idea that deserves special comment at
this point. We show that if the p is coupled to the (conserved)
isospin current, S U(3) invariance requires that the vector octet
be coupled to theF-spin current (conserved in this limit), i.e.,
SU(3)-INVARIANT VERTICES AND AMPLITUDES 127

ai: = 0 in (5.13) and fpNN = fM'" From Table 5.1 we see that the
pBB coupling part of Eq. (5.13) is

Je pBB = fPNNP".{Ny"t't"N + (1 - 2aE)Ey"t't"E

+ ~i (Ay,,:E + H.c.) + (1 - aE)( - i~YIl X :E)} (5.31)

Clearly, aE = 0 is required" in order that the pBB coupling be of

the form fPNNpll. JILT
and (5.32)

where the densities J o(x) are constructed as in Section 1.3.

If we add the contributions of the V P P coupling (with gppy =
t fp"" = tfPNN) we obtain a major portion of the isospin current
+--+
J"T = NYllt't"N + 8 yllt't"E + l!y"t:E + t1t x G,,1t
+ iKtt't"GILK + . .. (5.33)

Moreover, the K* doublet is coupled to the strangeness changing

current J/8 and the </> to the hypercharge current JIl Y with
definite strength:

Je YBB = fpNN{pll. JILT + V3(Mt" //8 + H.c.) + ~ </>" /Il Y

}

(5.34)
where the strangeness changing current transforms as a 2-com-
ponent spinor under S U(2) transformations, and the hypercharge
current is

(5.35)

The above considerations are closely connected with Sakurai's

generalization (11) of the Yang-Mills theory (18) in which there
are fundamental vector mesons (p,</>,w) coupled to the currents of

" From this it follows that the electric coupling of the vector mesons is given by
the" F-spin" current (cf. Eq. 5.13). The D and F currents are respectively even
and odd under R-conjugation.
128 INTRODUCTION TO UNITARY SYMMETRY

the quantItIes conserved in strong interactions (isospin, hyper-

charge, and baryon number). In addition, the strange vector
meson (K* or M) is coupled to the "partially conserved"
strangeness changing current. (For a discussion of the generali-
zation to many compact Lie groups, see Gell-Mann and Glashow
(19).) If the physical w were a pure member of an octet then the
w would be coupled to the baryon current BYJlB in the form

(5.36)

However the actual situation is more complicated than this, due

to cfo-w mixing. In the absence of a symmetry greater than
S U(3), gwBB has no relation to !PNN' Some (presently conflicting)
estimates based on vector meson exchange models of nucleon-
nucleon forces are now available.
The present time is not propitious for a detailed evaluation of
various methods of determining effective coupling constants. We
shall instead consider only one especially direct method, that of
measuring the decay widths of resonances.

5.2 Branching Ratios in Strong Decay Processes

In the limit of perfect unitary symmetry the branching ratios

for the decay of a resonance are given simply by Clebsch-Gordan
coefficients. We can illustrate by reviewing the implications of
isospin conservation for the decay of the 3-3 isobar N*. The N*
occurs in four charge states with the usual wave functions
N*++ !f++ = 7r+P
N*+ !f+ V!7r°p + VI7r+n
(5.37)
N*o !fo V!7r°n + VI7r-p
N*- !f- = 7r - n
In terms of the total width T = T(7r+p), we expect

T(NH -+ 7r 0p) = iT
(5.38)
T(NH -+ 7r+n) = IT
 SU(3)-INV ARIANT VERTICES AND AMPLITUDES 129

Because the masses of the different charge states of N* are not

exactly the same, the relations (5.38) will not be strictly valid.
At the present level of understanding, the corrections are of two
types: first, the mass differences affect the kinematical and phase
sp ace factors in a way that is easy to account for; second, the
m atrix elements differ from the ones predicted by "perfect sym-
metry" by an amount which is easily calculable only if the trans-
for mation property of the symmetry-breaking term is known. In
p ractice, this extra term has to be weak if any simple estimate is
to be made of its effect. For example, in the N* decay we expect
that the electric charge is responsible for deviations from charge
independence. The effect is small and the extra interaction is
linear in the third component of isotopic spin so that perturbation
t heory can be used. A systematic general way to handle such
p erturbations has been developed (20-22). Most applications to
date simply make the obvious kinematical corrections.
As a first illustration we use the vertex [Eq. (5.25) or (5.29)]
to correlate the p meson width (about 100 Me V) with that of the
4> meson, assuming both to belong to an octet of vector mesons.
As the masses are quite different (750 MeV for p, 1020 MeV for
4» the kinematical factors are very important. The only nonzero
iabc (c = 8) coefficients are i458 = i678 = V3j2. Thus the 4>
couples only to the K mesons in the limit of unitary symmetry,
as is seen explicitly in Eq. (5.29). The decay rates into K+ K-
and K.o KO are equal; the total rate is quickly found to be

(5.40)

The only significant allowed p decay mode is p ~ 27T. The p

width is

(5.41)

The ratio of the width of the 4> to that of the p is thus

r"'~KK _ 3 PK 3jm(j)2 3.4 MeV
(5.42)
rp~2" - 2 p,,3jmp2 = 100 MeV
130 INTRODUCTION TO UNITARY SYMMETRY

Later we shall wish to refine this value to account for the 4>-w
mixing which reduces the octet component of the physical 4>. For
a further discussion of 4> decay see the note by Sakurai (23). It is
interesting to record here the measured value r", = 3.1 ± 1.0
Me V (24). Of the other P PV couplings, the K* -+ KYJ mode is
kinematically forbidden. The decay mode K* -+ K7T has width
given by
3 PK,,3/ mK .2
(5.43)
"4 p",,3/ mp 2
where PK" (p",,) is the center-of-mass momentum of the K or 7T
(7T or 1!) in the rest system of the K* (p). Numerical evaluation
gives rK./rp ~ 0.30 as compared to the experimental value 0.47.
Next we consider the branching ratios for the decay of the
various members of the P 3 / 2 decuplet into the various isospin
states. Here we observe that most of the two particle channels
making up the SU(3) wave functions are closed. The thresholds
vary greatly and large symmetry breaking effects are to be
expected. For example, the T = 3/2 quartet in 10 is composed of
equal components of N7T and I:K in the limit of perfect S U(3)
symmetry. How is one to reconcile the successful dispersion theory
description of low energy pion-nucleon scattering (which ignores
the I:K channel altogether) with the spectacular mass formulas,
which are seemingly predicated on the smallness of the symmetry
violation? Here we only mention that the mass perturbations
involve quantities distinct from the state vectors. Thus the actual
T = 3/2, TTN state may be nearly of the form [if;27 (1 , 3/2) -
if;lo(1,3/2)]/V2 but the perturbation which lifts the degeneracy
nevertheless transform as 8.
In the limit of S U(3) symmetry the widths for the decays of
N*, Y1 *, E1/2 *, and Q would all be equal (all components of a
V-spin multiplet have the same width). Or, if the particles were
stable in the symmetric limit, the corresponding coupling con-
stants would be equal. The wave functions of these particles
can be read from Table 3.4_ We observe that the Q couples to
only one pair, EK. Of the constituents of the E*, only the E7T
.sU(3)-INVARIANT VERTICES AND AMPLITUDES 131

channel is open. Similarly, the decay of N* into EK is ener-

getically forbidden. However, the Y 1 * can decay into both E7T and
A7T. The symmetric branching ratio is

r( Y 1 * ---7E7T) 2
(5.44)
r(Yl* ---7 A7T) 3
as compared with the experimental value (25) ;5 .04! Of course
the kinematical corrections have not yet been supplied. A favorite
fudge factor is given by (7, 26)

(5.45)

where I is the orbital momentum, 11a an effective radius, and Yl

the reduced width of the resonance. The correction involving the
parameter a has to be viewed with distrust in a relativistic theory;
presumably it depends on dynamical details not always dis-
covered by inspection. Fortunately, the threshold factor p21+1
gives the major energy variation of rl' It is then supposed that
the symmetry considerations apply to Yl' although the validity of
this assumption is not yet known.
Taking 11a ~ 1 BeY (Glashow and Rosenfeld (26) used 11a =
350 Me Y) Cutkosky finds (7) r( Y 1 * ---7 };7T)/( Y 1 * ---7 A7T) = .16,
still uncomfortably larger than the experimental value. On the
basis of his investigation of the dynamics of the P S / 2 decuplet,
Cutkosky suggested that there should be some admixture of 27
into the P S / 2 resonances. Let us see if" contamination" of the 10
by the 27 gives an easy interpretation of the suppression of the
};7T decay mode of Y 1 *. We write (to lowest order)

(5.46)

From the wave functions (Table 3.4) we notice: 1) the 27 state

has no E~ component, 2) for r > 0 the A7T component is enhanced,
while the heaviest two particle states EK and};TJ are suppressed.
(For r = 1 the latter states are completely absent.) The ratio for
decay of a pure 10 state [e.g., Eq. (5.44) modified by (5.45)] is then
132 INTRODUCTION TO UNITARY SYMM

multiplied by the factor (1 + r) -2. An r of order unity the

presses the relative En decay to a value compatible with e
ment.
Suppose, for simplicity, that the Y1 * actually corresponds
mixture (5.46) with r = 1. Then the wave function (5.46) re
to

and the width of Y 1 * is [ignoring the factor (1 + a 2p2) in

which is not very important here]

r(Y1*) = [(p.E,,3 + 6pA,,3)/llpN,,3]r(N*) ~ .4r(N*)

Taking r( N*) to be 125 Me V yields 50 MeV for the wi
Y 1 *. This value compares favorably with the experimental
of 53 Me V (25).
Next consider the width of 8 1 / 2 *, Since the (8nh/2 comp
of 27 is very small we consider a pure 10 state, correct
kinematical factors in the usual way. This gives

The predicted value (9 Me V) compares well with the e

mental one (7 Me V).
We wish to emphasize that the theory of broken S U(3)
in a very primitive state, and that the above calculations m
have more than qualitative validity. Nevertheless, it is sati
that a few sensible corrections of the type just described
adequate to bring theory in line with experiment.
A valid question to raise at this point is whether some
ponents of the 27 P 3 / 2 states have observable effects. This
seem likely, especially in view of our discussion of the Y 1 * br
ing ratio. Calculations by Golowich (10) and Wali and Wa
(27) show that the 27 representation is less important than
be expected. I
An interesting situation ohtains in the decay of a baryon
resonance state into octets, e.g., a resonance B* - 7 B
S U(3)-INVARIANT VERTICES AND AMPLITUDES 133

because there are two independent decay amplitudes. [Recall that

the identity representation occurs twice in the decomposition
8 X 8 X 8 which enters into the decay matrix element (<fB.,<fBP)']
Consider the T = 1/2, Y = 1 member of the octet. From Table
3.4 we find

1
. /_ [3(N7T) - 3(EK) - (NYJ) - (AK)]
2v 5
(5.50)
1
<fa 2 = "2 [(N7T) + (EK) + (NYJ) - (AK)]

where (N7T) denotes a normalized T = 1/2 isospin wave function,

etc.
The observed state <fB' will decay into a linear combination of
<fal and <fa2' specified by the mixing angle ():
(5.51)

which gives the probability of the T = 1/2 state of (N7T), (EK),

(NYJ), and AK:
(N7T) (EK)

C2~;(} + t sin (}t (t sin () - 2~5 cos ())2,

(5.52)

(t
(NYJ)

sin () - 2 ~5 cos () t (t
(AK)

sin () + 2 ~5 cos () r
The time reversal symmetry of strong interactions has been used
in writing the probability amplitudes of <fal ' <fa 2 as relatively real.
Figure 5.3 shows the variation of the probability for a given
particle pair (in the T = 1/2 state) as a function of the mixing
angle. It will be observed that P N ,,((}) = P EK ( - (}), P AK ((}) =
P Nn ( - (}). The dominance of 7TN interactions in the Y = 1, T =
1/2 channel suggests that () is positive. Of course, when given an
octet one can find () by comparing the experimental branching
ratios of a given (Y, T) state with Fig. 5.3. This has been done by
134 INTRODUCTION TO UNITARY SYMMETRY

.75~--------------------~----------------------,

PIe)

8-

Fig. 5.3. The general octet wave function built from the pseudoscalar meson
and baryon octets has an arbitrary parameter determining the relative proba-
bility of various two particle isospin components (see Eq. (5.51». The probabilities
P(II) of finding T = 1/2, Y = ± 1 components are shown.

Glashow and Rosenfeld (26) for a D3/2 octet (including the

T = 1/2, "2nd" TTN resonance at 1512 MeV) and existing mem-
bers of a presumed FS /2 octet (including the" 3rd " TT N resonance
at 1688 Me V). As the experimental situation with regard to these
"octets" is rather fluid, we shall not give a detailed discussion.
(For example, the 1520 Me V, D3/2 Yo * is now believed to be a
unitary singlet, rather than the T = 0 member of an octet.)
Further, all members of the presumed FS /2 octet have not yet
been discovered. We mention, however, that the value of (J ~ 35°
gives a satisfactory fit to all the octets. One direct piece of evidence
is the very small (or completely absent) decay mode of the 1688
MeV state into the 7]N channel (3, 29). I
One further important remark is that the most plausible
dynamical models of the higher baryon resonances lead to essen-
tially the same ratio of 8 1 and 8 2 • If we regard the baryon octet
SU(3)-INVARIANT VERTICES AND AMPLITUDES 135

.75r----------------------r--------------------~
Y=T=O

8-
Fig. 5.4. The probability of finding the T = Y = 0 two particle components in
the octet wave function is given as a function of the mixing angle O•

.75r---------------------r-----------~------_,
(Y,T)=(O,I)

L7r

8-
Fig. 5.5. The probability of finding the T = I, Y = 0 two particle components in
the octet wave function is given as a function of the mixing angle O.
136 INTRODUCTION TO UNITARY SYMMETRY

as predominantly a bound state of Band P then the self-con-

sistency of the reciprocal bootstrap (7, 8) model connecting the B
octet and B* decuplet fixes the mixing parameter to be () :::; 33°.
A priori, there is no reason why () should be the same for every
octet, but there is apparently a kind of universality in effect. A
very interesting point is that the observed () is that which nearly
maximizes the 7T N constituent of the (Y, T) = (1,1/2) wave func-
tion. Presumably it is this fact, along with the suppression of the
high mass states when symmetry breaking is taken into account,
which explains the moderate success of .earlier models of pion and
nucleon phenomena in which the strange. particles were ignored
completely. This success does not necessarily carryover to other
channels, however. The advent of SU(3) has done much to cure
the symbiotic diseases of" piophilia" and" xenophobia".
For completeness we give in Figs. 5.3-5.5 the constituent
probabilities for the (Y, T) = (0,1), (0,0), and (-1,1/2) channels.

5.3 SU(3) Invariance and Scattering Amplitudes

The assumption that the scattering operator S commutes with
the unitary spin G leads to many connections among various
scattering amplitudes. Our language in this section is appropriate
to two-body reactions

a + b ~ c + d. (5.53)

However, the reader is invited to regard anyone of these labels

as representing a resonant state, or he may easily extend the
explicit equations to n-body reactions along the lines used in
isospin analysis (see, for example, reference 29). For a detailed
study of the latter one will need the S U(3) generalization of the
recoupling coefficients (Racah coefficients, 6-j symbols, etc.).
Those recoupling coefficients of practical importance can be ob-
tained fro~ the octet-model crossing matrices given in ChaP1 r 7.
Let us suppose that particles a, b, c, and d belong to irreducible
representations of SU(3). Using the SU(3) Clebsch-Gordan
coefficients discussed in Chapter 4 we can expand the states lab)
SU(3)-INV ARIANT VERTICES AND AMPLITUDES 137

and led) in terms of various irreducible representations contained

in the products of a x band c x d, as in Eq. (4.9).

(5.54)

A similar equation holds for the final state led) . We thus obtain

<cdl Slab)
= L L (fLa
uvy u'v'y' Va
fLb
Vb
fLy) (fLe
V Ve
fLa
Va
~' y,) <cd(fL' 1'V') I S Iab(fLyv)
V
(5.55)

<
where in writing the cd(fL' y,v') 1S Iab(fLyv) we have suppressed
the constant quantities fLa' fLb' fLe' and fLa' Invariance of Sunder
the subgroup S U(2)r x U(1)y (the actual symmetry group of the
strong interactions) guarantees that the only nonzero contribu-
tions to (5.55) occur for v' = v. Next we show that if the repre-
sentations fLy' [specifically D<uY')(p',q')] and fLy(D<Uy)(p,q)) are
<
inequivalent then the matrix element cdfL' y.v l S IabfLv) vanishes.
By assumption of S U(3) invariance the two Casimir operators
[Eq. (3.46)] commute with S:
(5.56)
Taking matrix elements of these relations shows that

(C2 (p',q') - C2 (p,q)) <cd(fL'y'V) I S lab(fLyv) = 0

(5.57)
(C3 (p',q') - C3 (p,q))<cd(fL'y'V) I S lab(fLyv) = 0
Thus the S matrix vanishes unless p' = p and In the q' = q.
usual way <cd l S lab ) is independent of the T3 component con-
tained in v. This follows from the fact that S commutes with T ±.
Moreover, this matrix element is independent of V because of the
occurrence of generators that change the eigenvalues T, Y. Thus
Eq. (5.46) reduces to

<cd l S lab) = u~, (~: ~bb ~y) (~e ~: ~Y')<Cd(fLY')1 S lab(fLY)

(5.58)
138 INTRODUCTION TO UNITARY SYMMETRY

The content of the ahove can he recognized as a restatement of

the Wigner-Eckart theorem discussed in Chapter 4.
One new feature that emerges here is the multiple occurrence
of (equivalent) irreducihle representations. For example, in the
scattering of P and B octets (8 x 8 = 1 + 8 1 + 8 2 + 10 +
10* + 27) we need seven amplitudes to descrihe all possihle
scatterings P a + Bb -7 Pc + Bd:
S(I), S(8 1 ), S(8 2 ), S(8 1 -78 2 ), S(10), S(10*), S(27). (5.59)
The amplitude S(8 1 - 7 8 2 ) is not distinct from that for 8 2 - 7 8 1
hecause of the time-reversal invariance of strong interactions.
One can invert the procedure hy starting from SU(3) wave
functions and expanding the corresponding S-matrix elements in
terms of particle (isospin-hypercharge) states, using the isoscalar
factors.
For example, from Eq. (5.50) we see that the T = 1/2, Y = I
octet scattering amplitudes are

<8 11 S 181) = to[9S(7TN) + 9S(l:K) + S(N'l) + S(AK)]

<8 21 S 182) = ![S(N7T) + S(l:K) + S(N7J) + S(AK)] (5 .60)
I
<8 21 S 181) = . r [3S(N7T) - 3S(l:K) - S(N7J) + S(AK)]
4v5
As discussed in the preceding section, a resonant or hound octet
state is, in general, a mixture of 8 1 and 8 2 , Corresponding to the
linear comhination (5.51) (with 8 fixed hy dynamics or possihly
some still greater symmetry) a pole in the S matrix <.pB I S I.pB)
is expected near the resonance energy in the appropriate partial
wave amplitude on the second sheet in the energy availahle. The
other independent state, .p B' = - sin 8.p81 + cos 8.p82' exhihits no
special hehavior there.
The assumption of SU(3) symmetry as applied to the full
reaction amplitude has not yet had as impressive suc ~ ss as
certain other predictions. At least part of the reason is that every
partial wave amplitude requires a different correction factor due
to kinematical considerations. (Only when a resonant state can
SU(3)-INVARIA NT VERTICES AND AMPLITUDES 139

be isolated is it easy to make comparisons of the type of Section

5.2.) One rough means of accounting for this circumstance h as
been proposed by Meshkov, Snow, and Yodh (30), who compurf
various reactions of the type P + B --+ B* + P at the sanl<
kinetic energy in the center-of-mass frame. Another mctilocl
commonly used in the construction of dynamical modcl!< "I'
various reactions is to use physical masses, but SV(3) cO llJllill ~
constants for the computation of forces. This approach makes U HIl
of the apparent fact that the S V(3) values of reduced cou plill ~H
seem to be less spoiled than the magnitude of the amplitlldc:H.
Moreover, in a bootstrap model one can recompute the cOllplillf.,;
constants and (hopefully) arrive at a self-consistent dynarni c~ .
We conclude this section by describing how reactions CU ll
be related by use of the V- or V-spin subgroups. This method
requires only the S V(2) Clebsch-Gordan coefficients and is very
simple and rapid in execution. Our discussion follows that of
Lipkin (31). To illustrate this technique consider the reactions :
7T- + P--+ K+ + Y l *-
7T - + P --+ 7T + + N*-
(5.61)

K- + P--+7T+ + Yl *-
The P 3 {2 resonances (N*-,Yl*-,E-) belong to the same V-spin
quartet (they all have the same charge). The pairs (7T- ,K-) and
(K+ ,7T+) also belong to V-spin doublets and the proton is the
V3 = 1/2 member of the V-spin doublet. Thus the total V-spin of
the left-hand side of (5.61) is 0 or 1, while it is 1 or 2 on the
right-hand side. Thus to the extent that V-spin is conserved, only
one amplitude (V = 1) describes the reactions (5.61). The first
two reactions have U 3 = + 1 and the last two have U 3 = O. The
amplitudes are given by the V = 1 r~action amplitude (called a l )
multiplied by the projection of the initial and final states on the

U = 1 state. Thus the amplitudes are in the ratio - tal: 2

VS al ..
-tal: tal'
140 INTRODUCTION TO UNITARY SYMMETRY

Reactions involving the emission or absorption of photons pro-

vide an especially powerful utilization of the U-spin method. This
is because the charge operator is a U-spin scalar. Hence the
initial and final stages, Ii ) and If), of any product of current
operators jJl(x) have the same value of U and U 3 :

The charge is a component of the octet vector (G 3 + Gs /'V3)

and hence is not invariant under S U(3). Thus we shall place the
discussion of electromagnetic process in the following chapter on
"broken SU(3)."

References
1. R. H. Dalitz, Ann. Revs. Nucl. Sci., 13, 339 (1963).
2. J. J. de Swart and C. K. Iddings, Phys. Rev., 130, 319 (1963).
3. P. Carruthers, Phys. Rev. Letters, 12, 259 (1964); G. Altarelli, F. Buccella, and
R. Gatto, Nuovo Cimento, 35, 331 (1965).
4. T. K. Kuo, Phys. Rev., 129, 2264 (1963); 130, 1537 (1963).
5. S. Hatsukade and H. J. Schnitzer, Phys. Rev., 128, 468 (1962); 132, 1301
(1963).
6. R. L. Warnock and G. Frye, Phys. Rev., 138, 947 (1965).
7. R. E. Cutkosky, Ann. Phys. (N. Y.), 23, 405 (1963).
8. A. W. Martin and K. C. Wali, Nuovo Cimento, 31, 1324 (1964).
9. A. W. Martin and K. C. Wali, Phys. Rev., 130, 2455 (1963).
10. E. Golowich, Phys. Rev., 134, Bl297 (1965).
11 . J. J. Sakurai, Ann. Phys. (N. Y.), n, 1 (1960).
12. W. R. Frazer, and J. R. Fulco, Phys. Rev., n7, 1603, 1609 (1960).
13. P. Carruthers, Lectures in Theoretical Physics, Vol. VUb, Dniv. of Colorado
Press, Boulder, Colo., 1965, p. 82.
14. J. S. Ball, and D. Y. Wong, Phys. Rev., 133, Bl79 (1964).
15. A. Donnachie, J . Hamilton, and A. T. Lea, Phys. Rev., 135, B515 (1964).
16. M. Der Sarkissian, Nuovo Cimento, 34, 1010 (1964).
17. H. J. Lipkin, Phys. Lett., 7,221 (1963).
18. C. N. Yang and R. Mills, Phys. Rev., 96,191 (1954).
19. M. Gell-Mann, and S. L. Glashow, Ann. Phys. (N. Y.), 15, 437 (1961).
20. C. Dullemond, A. J. MacFarlane, and E. C. G. Sudarshan, Phys. RJv. Letters,
10, 423 (1963).
21. V. Gupta and V. Singh, Phys. Rev., 135, B1442 (1964); 136, B782 (1964).
22. C. Beechi, E. Eberle, and G. Morpurgo, Phys. Rev., 136, B808 (1964).
23. J. J. Sakurai, Phys. Rev. Letters, 9, 472 (1962).
SU(3)-INV ARIANT VERTICES AND AMPLITUDES 141

24. N. Gelfand et aI., Phys. Rev. Letters, 11, 438 (1963).

25. A. H. Rosenfeld, A. Barbaro-Galtieri, W. H. Barkas, P. L. Bastien, .T. Kil'z,
and M. Roos, Rev. Mod. Phys., 37, 633 (1965).
26. S. L. Glashow and A. H. Rosenfeld, Phys. Rev. Letters, 10, 192 (1963).
27. K . C. Wali and R. Warnock, Phys. Rev., 135, Bl358 (1964).
28. F. Bulos et aI., Phys. Rev. Letters, 13, 486 (1964).
29. P. Carruthers, Phys. Rev., 122, 1949 (1961).
30. S. Mcshkov, G. A. Snow, and G. B. Yodh, Phys. Rev. Letters, 12, 11.11 (1%1\.).
31. H. J. Lipkin, Lie Groupsfor Pedestrians, Interscience, New York, 1965.
 Chapter 6

Broken SU(3)

6.1 Introduction
As we have mentioned before, one of the remarkable aspects of
unitary symmetry is its survival in the face of the large mass
differences among particles belonging to the unitary multiplets.
At present, it is not clear whether this lack of perfect symmetry
is due to the explicit occurrence of forces which are not S U(3)
invariant (e.g., as the Zeeman splitting of atomic energy levels
due to the rotationally non-invariant magnetic field), or rather to
the self-consistency requirements of (now obscure) dynamical
mechanisms possibly underlying the observed particles. In the
absence of a fundamental understanding of this situation, we
shall be fairly brief in our description of symmetry breaking.
Fortunately, the data suggest that SU(3) is broken in a very
simple way. More surprising is the success of rather naive pertur-
bation theory calculations. Despite the dubious dynamical
approximations involved in such a treatment of the strong
interactions, it is difficult not to regard the results as further a
posteriori support for the validity of the" broken eightfold way."
In Chapter 1 we discussed the violation of isospin by electro-
magnetism (neglecting gravity and the weak interactions). In that
case the strong interactions were supposed to have the sym-
metry SU(2h x U(I)y except for the small electromagnetic cor-
rections. In a Lagrangian formalism we would write I

(6.1)
where Lstrong is invariant under SU(2h x U(I)y but L e - m, con-
142
BROKEN 8U(3) 143

taining the charge Q, is not invariant but nevertheless has a simple

transformation property in the invariance group of Lstrong. (Le-m
transforms as the sum of a scalar and the third component of a
vector. In particular, the fields entering into Le - m have simple
transformation properties under the group.)
The reader is by now no doubt aware that Lstrong is approxi-
mately invariant under SU(3). Therefore one might consider the
division
(6.2)
where Ls is the "very strong" part (S U(3) invariant) and the
"medium strong" part LMS conserves isospin and hypercharge.
Thus LMS commutes with Y, T2, and T 3 • The structure of LMS
has to be taken from experiment.
To. see what is involved we consider some specific models.
Suppose we consider deviations from isospin conservation as re-
flected in the energy levels of nuclei. Imagine that the sole cause
of these deviations comes from the neutron-proton mass differ-
ence. (This is unrealistic because of the 170 - 17+ mass difference.)
Then if the forces holding the nucleons together are charge
independent the perturbation to the kinetic energy term is non-
relativistically, approximately L (Pl j2M)(,1 Mjj M) where ,1 M j =
M - M j , with M the average mass and M j = t(Mp - Mn) +
Ta'(Mp - Mn) where T3 j is the third component of the isospin for
the jth nucleon. Thus each particle contributes a perturbation pro-
portional to T3 j of order of magnitude E k1n X (,1 Mj M) so that the
splitting is easily and legitimately estimated by perturbation theory.
In the Sakata or quark model a similar treatment, which
recognizes the mass difference between the isospin doublet and
singlet, is quite natural. The mass term in Lagrangian theory is
(cf. Eq. 3.60)
(6.3)

Introducing the average mass mo = j md + tms we can write

(6.3) in the form

(6.4)
144 INTRODUCTION TO UNITARY SYMMETRY

The role of ..\a in distinguishing the iso-doublet and iso-singlet has

been noted previously. In this model, the symmetry breaking
term transforms as the eighth component of the unitary spin, Ga.
Moreover, in the eightfold way Ga has the physical significance of
the hypercharge.
There is little evidence bearing on the validity of the specific
form (6.4). However, such detailed information is not necessary
for many purposes. In fact we can proceed in several stages of
increasingly restrictive assumptions on the nature of CMS' The
main virtue of the form (6.4) is in suggesting the hypothesis that
CMS transforms as Ga (rather than, say, some function of Ga). This
hypothesis is in striking analogy to the fa.ct that the electro-
magnetic term Ce - m transforms as the component of G, Ga +
(1/V3)G a•
Following an introductory section on mass splitting neglecting
electromagnetism, we proceed on the following path of increasing
specialization. First, we investigate the consequences of the
hypothesis that the total Lagrangian breaks up into three pieces
of decreasing magnitude with the indicated symmetry:

Cs + CMS + Ce - m (6.5)
symmetry SU(3) SU(2h x U(1)y SU(2)u x U(1)Q
Here we follow the approach of Feldman and Matthews (1). This
method exhibits clearly which results depend on the" conserva-
tive" assumptions of (6.5) in distinction to the more specific ones
that CMS transforms as Ga and Ce - m as Ga + (1/V3)G a•
As we do not wish to solve the dynamical problem posed by
(6.5), we convert (6.5) to an effective mass operator. The construc-
tion of effective Hamiltonians by means of eliminating certain
dynamical variables is a standard procedure. We leave the details
to the reader (see the paper by Marshak et a1. (2) for one method).
The mass operator corresponding to a specific type of P/ rticle is
of the form
M = Mo + MT + Mu (6.6)
where Mo is SU(3) invariant, MT is an isospin scalar, and Mu is
BROKEN SU(3) 145

a U-spin scalar [cf. (6.5)]. Experimentally one notes that MT

is about an order of magnitude smaller than Mo; in turn, M u is
about an order of magnitude smaller than M T • For mesons, we
expect an expression similar to (6.6) involving the squares of the
masses.

6.2 Mass Formulas (Neglecting Electromagnetism)

Let us assume explicitly that the medium strong mass operator
MT transforms like the hypercharge Y (ex: Gs ). If we take the
expectation value of Mo + MT in the state v, belonging to the
irreducible representation J.L, a mass formula relating various
isospin multiplets can be obtained. (If the particles belong to a
reducible representation of SU(3) then mass inequalities can be
obtained (3).) The mass is given by
M(J.L,v) = a(J.L) + <J.L,vi MT iJ.L,v) (6.7)
where a(J.L) is a constant within the irreducible representation.
We can, if we choose, use the Wigner-Eckart theorem [Eq.
(4.25)] to express the matrix element in (6.7) in terms of a number
of arbitrary constants. That number is equal to the number of
times J.L occurs in the decomposition 8 x J.L. Recall that the mul-
tiple occurrence of representations sometimes makes the SU(3)
Wigner-Eckart theorem more complicated than the S U(2) version.
Thus when J.L = 8 (an especially important case) two arbitrary
constants occur in (6.7) in addition to a(J.L). Hence we cannot
assert that the matrix element of MT is, in general, proportional
to Gs . However, we can do just as well and also avoid looking up
S U(3) Clebsch-Gordan coefficients if we examine the behavior of
L MS under the subgroup of transformations S U(2)u x U(l)Q.
By assumption, MT transforms like Y. Recall that the operator
Y is related to the diagonal generator H by Y = 2Y3r'(1).H
[Eq. (2.55)]. By resolving r'(l) into orthogonal vectors r(3) and
r'(3) we observe that Y is composed of U 3 and Q. From the
definitions of Section 2.3 we find r'(l) (V3/2)r(3) - }r'(3),
from which we derive the relation
Y = U3 + tQ (6.8)
146 INTRODUCTION TO UNITARY SYMM

Eq. (6.8) will be recognized as a "rotated" version of the re

Q = T3 + t Y.
Thus under SU(2)u x U(I)Q the operator MT transforms
sum of a U-spin vector and a U-spin scalar. Applying the
Wigner-Eckart theorem to this subgroup leads to the resul
within a U-spin multiplet the mass is given by the equal-s
rule
M(U,U 3 ) = A + BU3 ,
or M(U,U 3 ) - M(U, U3 - 1) = constant.
As is well known, this expression is very. well satisfied w
spacing of about 150 Me V for the baryon decuplet. Here
two constants enter in (6.7) since 10 x 8 contains 10 only
Moreover, the weight diagram is nondegenerate so that the U
eigenfunctions have definite isospin as well. From (6.9) we g
result
m(N*) - m(Y*) = m(Y*) - m(S*) = m(S*) - m(Q)
For the baryon octet, the non-empty relation following
(6.9) is

where I;uo = -tI;° + ('V3 j2)11° is the U3 = 0 member

U-spin triplet (n,I;uO,SO). Since m(I;uO) = tm(I;) + !
(6.11) can be rewritten in the form (4)
.,
Um(N) + m(S)] = Hm(I;) + 3m(I1)]
This formula, which closely resembles an old one in the
symmetry scheme (5) (11 ~ I;), is also well satisfied.
average multiplet masses the left-hand side is 1128 Me V as
pared to 1134 MeV on the right-hand side.
The pseudoscalar mesons also form an octet; however, (6
not at all valid. One can note that the perturbation is sc
small in this case since the deviations of the actual masses
the average mass exceed the latter in some cases. Clearly
cannot rely on perturbation theory in such a situation. lIow
the following speculation enormously improves the agreeme
BROKEN SU(3) 147

relativistic field theories of mesons the mass m always enters as

m 2 • Therefore we might try the relation
( 6.13)
(by charge conjugation invariance, K and K have the same ma sH).
The most lenient validity condition on the derivation of (6.13),
i.e., 8m 2 « m 2 , is flagrantly violated. Nevertheless, (6.13) is
satisfied to within 5%. At present, there is no basic understanding
of this question.
Some confusion arises if we consider the vector mesons; the W
(750 Me V) and cp (1020 Me V) both have the same quantum num-
bers (J = 1 -, G = -1, T = Y = 0). Indeed, neither mass fits
the formula
(6.14)
which requires mx = 930 Me V. It has been emphasized,
especially by Sakurai (6), that there is likely to be a substantial
mixing of such states, so that the observed wand cp may be con-
sidered to be mixtures of a unitary single wO and the isosinglet
member of the unitary octet, cpo. In this case, the violation of
unitary symmetry is aggravated by the unstable dynamics of
nearly degenerate states. Applying ordinary degenerate perturba-
tion theory leads to the following picture.
We write the physical cp and w states as linear superpositions
cp = cos Acpo + sin AWo
(6.15)
w = - sin Acpo + cos AWo
The (mass)2 matrix is written in the two-dimensional space of cpo
and wo:
2 2
M2 = (m (cpO) m (cpw») (6.16)
m2 (cpw) m 2 (wO)
and A is determined by requiring col (cos A, sin A) to be an eigen-
vector of (6.16). The eigenvalues of (6.16) are

m2(cp)} = t[m 2(cpO)

{ m 2 (w)
+ m2(w O)]
± t{[m 2 (cpO) - m 2 (w°)]2 + 4m 4 (cpw)F/2 (6.17)
 148 INTRODUCTION TO UNITARY SYMM

while ,\ is determined from

Taking the product of the eigenvalues we obtain an equat

eliminate m2 (c/>w)

To determine '\, we take the c/> and w masses from experime

'I
, I calculate C/>0 to be 930 MeV from the mass formula (6.14).
is determined from the sum rule

m2(c/» + m2(w) = m2(c/>0) + m2(wO)

to be 885 MeV. The resulting mixing angle is ,\ = 39°. The
values are shown in Fig. 6.1. For this calculation we have

cp
/
1000 /
/
/
/
/
cpo /

96b K*, R* K* ,R*

>.,
::;: WO \
III
\
III \
0 \
~
\
800
\
\ , w

p P

700
Fig. 6.1. The effect of rfo - w mixing on the masses of the unitary sin
and the T = Y = 0 member of the octet (rfoO) are indicated. For refere
masses of the remaining members of the vector meson octet are shown.
BROKEN SU(3) 149

m(</» = 1020 MeV, m(K*) = 888 MeV, m(w) = 780 MeV, and
m(p) = 750 MeV.
According to this simple model, the physical</> spends (cos 39°)2
= 60% of its time as </>0 and 40% as the singlet woo (For the w,
interchange </>0 and wo.) Of course (6.15) is only a rough approxima-
tion since many particle configurations are expected to be
important constituents of the </> and w ,mesons. However, it is
interesting to follow up the implications of (6.15). To do this we
suppose that, apart from simple kinematical corrections, the
only violations of SU(3) are due to </>-~ mixing.
With these assumptions let us reconsider the vertices analyzed
in Chapter 5. First consider the decay </> --'3>- KK. Since wo is pre-
sumably an SU(3) singlet, it cannot decay into two octet par-
ticles. Thus the rate previously calculated [Eq. (5.42)] should be
decreased by the factor (cos ,\)2 = .6. We thus get T(</> --'3>- KK) =
2 MeV. The mode </> --'3>- P + 7T seems to be infrequent (7) « 10%
of </> --'3>- KK). Of course the physical w is too light to decay into
KK.
Consider next the isoscalar nucleon electromagnetic form
factor, whose momentum transfer dependence is generally
believed to be dominated by the wand </> mesons (8). Since the
photon couples through Q, which is a component of a vector
operator, the SU(3) singlet component wo does not couple to the
photon. This leads to a reduction factor cos'\ and - sin '\,
respectively, for </> and w, both given in terms of the </>0y coupling.
Consider the </>BB and wBB coupling, where B is the baryon
octet. The singlet wO couples to BB independently of the V octet
coupling.
Further discussion of various applications may be found in the
paper by Dashen and Sharp (9).
The phenomenon of </>-w mixing is very interesting and pro-
vides a fertile source of speculations. We shall not describe these
here, despite their great interest. Only one further point of view
with regard to </>-w mixing will be mentioned. The preceding
discussion presumes that the effect is due to the breakdown of
SU(3). However, we could start from the point of view that the
150 INTRODUCTION TO UNITARY SYMM

symmetry is greater than that of S U(3), since we have nine n

degenerate vector mesons. The following simple model (10)
trates such a circumstance. From the quarks q and ij w
construct nine J = 1 - bound states. If the quark compone
are degenerate and the forces are independent of the S U(3)
i for this partial wave, the nine states with wave functions
(8)

ijq (1)
are degenerate (neglecting exchange forces). The T = 0 m
of the octet' (</>0) is given by Eq. (3.61), with an extra (arbi
minus sign for convenience
1
</>0 = - V6 (ijlql + ij2q2 - 2ij3q3)

while the normalized S U(3) singlet ijq is

1
wo = VS (ihql + ij2q2 + ij3q3)'

If we now let the mass m3 of the S U(2) singlet q3 be differen

~
that (ml = m 2) of the doublet (ql,q2)' </>0 and Wo no long
physical eigenstates. Instead, the T = Y = 0 eigenstates a
</> = ij3q3

[For the signs in the w eigenfunction, refer to Eq. (1.31)] Ex

sing the physical states in terms of </>0 and wo gives

</> = vI</>o + vtwO

w = - vt</>o + vlw o \
The mixmg angle ,\ is now 35°, in good agreement wit
"experimental" value given before.
Although this model should perhaps not be taken seriou
exhibits some features of general interest. The symmetry
BROKEN SU(3) 151

model is U(3), with basis q;qj, reducible to 8 + 1. No mass formula

results unless assumptions are made about the nature of the for ces
(her e independent of the unitary spin). However, in the ge neral
case mass inequalities can be derived (3). The present model is
so specific that more explicit mass r elations can be given. Sincc
the p triplet is composed of the same components as w (qHq2) we
hav e
m(p) = m(w) (6.26)
F urthermore, since the usual mass formula holds for 4>0, which is
~ 4> and !w, we obtain the relation

(6.27)
Although these relations are in fair agreement with experiment,
t he reader is advised to regard with distrust all mass formulas
based on simplified models.
To conclude this section we give the Okubo mass formula (11)
for an arbitrary, irreducible representation of S U(3). It is sup-
p osed that the mass operator for baryons [or (mass)2 for mesons]
t ransforms as the sum of an SU(3) scalar and the T 3 3 (Y = T =
0) member of the octet. From Eq. (4.61) we know that the t ensor
operator T 3 3 has within an irreducible representation the effective
form
(6.28)
where a and b are representation-dependent constants. We next
express (6.28) in terms of physical quantities. A priori, these are
expected to be T and Y because (6.28) is invariant under
S U(2)r x U(l)y. Further we know that (A,A) = 2G2 is a con-
stant within an irreducible representation. The correlation (3.57)
can be used to express the A/ in terms of the Gi • We have A 3 3 =
- Y. For A,,3A3~ we find
A,,3 A 3~ = + iG 5 )(G4 - iG 5 ) + (G 6 + iG7 )(G6
(G 4 - iG7 ) + tG S 2
= G4 2 + G5 2 + G6 2 + G7 2 + GS 2 + !G S 2
(6.29)
+ i[G 5 ,G4 ] + i[G 7 ,G6 ]
152 INTRODUCTION TO UNITARY SYM

recalling that T2 = E?= 1 G 2

j and that Gs is ('V3/2) Y. Thus w

as an alternate form of (6.28). The mass formula is accord

M(or M2) = A + BY + C[T(T + 1) - lP]
For mesons the term linear in Y is absent by charge conju
invariance. For triangular representations, T = 1 + t Y
(6.31) reduces to a linear relation
M(triangular representations) = A' + B' Y
also derivable from V-spin considerations. [Combine (6.9
(6.8); for triangular representations there is no degenerac
the V-spin eigenfunctions are also T-spin eigenfunctions.)
We note here the "rotated" forms of (6.30):

Tll = A' + B'Q + C'[V(V + 1) - tQ2]

T22 = A" + B"Z + C"[V(V + 1) - lZ2] (
Equation (6.33a) is useful in the study of electromagnetic eff
systems of strongly interacting particles (12) (tr e charge op
Q transforms like Tll).
For further information on the conversion of tensor ope
to useful forms the reader will wish to consult papers by Di
Ginibre (14), Rosen (15), and Goldberg and Lehrer-Darned

6.3 Mass Formulas (Including Electromagnetism)

We now discuss what can be learned from (6.6) includi
electromagnetic term M u and not making an explicit assum
on the form of M T • For orientation consider the parallel
formed by four neighboring points in a weight diagram (Fig
supposed to be nondegenerate. If M u were absent then i
conservation would imply the vanishing of the equality
m(l) - m(2) = m(4) - m(3) \

On the other hand, if Mr were absent the equality

m(l) - m(4) = m(2) - m(3)
B ROKEN SU(3) 153

3 4

Fig. 6.2. The parallelogram of occupied sites (nondegenerate) illustrates the maBS
relation of Eqs. (6.34) and (6.35).

would also vanish. Now (6.34) and (6.35) are algebraically the
same, and so it seems plausible that they be valid in the presence
of both terms Mu and MT [but the values of m(I) - m(2), etc.,
are no longer zero]. That this is true follows on writing out (6.34-)
in detail

MT(I) + Mu(I) - MT(2) - Mu(2)

= MT(4) + Mu(4) - M T(3) - Mu(3) (6.36)
where MT(I) = <11
MT 11),
etc., and the states Ii ) belong to an
irreducible representation of SU(3). The states 11) and 12) are
transformed into each other by T ±' which commute with M T •
Hence MT(I) = MT(2), and similarly M T(3) = M T(4). In the
same way we see that Mu(I) = Mu(4) and Mu(2) = Mu(3),
establishing the desired equality.
This" parallelogram law" can be generalized to include the case
of degeneracy in the weight diagram. Before turning to this, we
apply (6.34) to the baryon decuplet, obtaining
m(N*-) - m(N*O) = m(Yl*-) - m(Yl*O)
m(N*O) - m(NH) = m(Yl*O) - m(YIH) (6.37)
m(Y1 *-) - m(Yl*O) = m(E*-) - m(E*O)
The great width of these P3 /2 resonances relative to the electro-
magnetic splittings makes the relations (6.37) difficult to test.
The baryon octet provides a good example of how to handle a
case of degeneracy. The sites are labeled as in Fig. 6.3. The center
154 INTRODUCTION TO UNITARY SYMMETRY

n p

Fig. 6.3. Parallelograms in the baryon octet weight diagram useful in derivin
mass relations are shown.

is degenerate and so the U-spin eigenfunctions differ from the

isospin eigenfunctions. (The U = 1 and 0 eigenfunctions ar
Euo = _tEO + (V3/2)AO, Auo = ('V3/2)EO + tAo.) The mas
operators have matrix elements related by

mT(n) = mT(p) mT(E+) = mT(EO)

mu(n) = mu(EuO) mu(p) = mu(E+)
(6.38
mT(E-) = mT(EO) mT(E-) = mT(EP)
mu(E-) = mu(E-) mu(EO) = mu(EuO)
We thus find

m(n) - m(p) = mu(n) - mu(p)

v'3
= tmu(EO) + !mu(AO) - ""2 mu(EA) - mu(E+)
\
= mu(EO) - mu(E+) + ![mu(AO) - mu(EO)]
v'3
- ""2 mu(EA) (6.39
BROKEN SU(3) 155

Since mT(EO) = mT(E+) we find

m(n) - m(p) + m(E+) - m(EO) = ![mu(AO) - mu(EO)]
V3
- 2 m(EA) (6.40)

The "transition mass" m(EA) is defined by

(6.41)
The right side of (6.40) can be further simplified by noting the
relation

(6.42)

This follows immediately from the equality

<Euol Mu IAuO) = 0 (6.43)
Thus Eq. (6.40) now reads

m(n) - m(p) + m(E+) - m(EO) = - V3m(EA) (6.44)

The quantity m(EA) depends entirely on the isospin-violating
electromagnetism. Equation (6.44) gives a measure of the isospin
impurity in the EO-Ao system, discussed below.
First, we notice that a similar analysis carried out for the other
parallelogram of Fig. 6.3 yields the relation

Addition of (6.44) gives the relation first derived by Coleman and

Glashow (17)

m(S-) - m(SO) = m(E-) - m(E+) + m(p) - m(n) (6.46)

The right side of (6.46) is 7.0 ± 0.5 Me V, while the left side
has been measured to be 6.1 ± 1.6 Me V, 6.8 ± 1.6 Me V, and
5 ± 3 MeV. (The experimental sources are given in reference 18.)
This very satisfactory result is especially important because it
provides support of S U(3) independent of any special assumptions
about the transformation properties of MT and M u'
156 INTRODUCTION TO UNITARY SYM

For octet mesons the extra symmetry associated with

conjugation reduces the equality corresponding to (6.46)
empty identity. However, there is an analogue of the tran
mass formula (6.44). For the pseudoscalar meson octet it i

Let us now restrict our attention to the case in which Mr

forms as Y, as in Section 6.2, but without restricting M u. T
order, in the medium strong interactions, we again hav
formula
M = A + BU3

within a U-spin multiplet. The argument of Section 6.2 work

because Mu (as well as M o) is constant in a U-spin multip
Therefore Eq. (6.10) is true (now including electromagr
for the negative particles:

m(N* - ) - m(Y1 *-) = m(Y1 *-) - m(S*-) = m(S*-) - m

,

while (6.11) is still valid in the form originally given. In te

particle states, (6.11) becomes

2[m(n) + m(SO)] = m(EO) + 3m(A) - 2v'3m(EA)

m(EA) can now be eliminated using either or both of Eqs.

and (6.45). We choose the latter alternative, obtainin
formula

m(n) + m(p) + m(SO) + m(S-)

= 3m(A) + m(E+) + m(E-) - m(EO)

The corresponding results for the pseudoscalar meson octet

4m 2 (KO)

2[m 2 (K+) + m2 (KO)]

If there were no electricity these formulas would collapse

forms given in Section 6.2.
B ROKEN SU(3) 157

The implications of the predicted value of m(EO A) have been

studied by Dalitz and von Hippel (18). The existence of m(Eil)
implies that the physical particles are not pure S U(2)r eigen-
states . The resulting mixing problem is very similar to that in the
cp-w discussion. We write
A phys = + sin ,\. EO
cos ,\ . A
(6.54)
E phyS = - sin ,\ . A + cos ,\. EO
T he mass operator has the matrix form

M = (mo(il) m(EA») (6.55)

m(EA) mo(E)
T he eigenvalues of (6.55) give the actual mass m(A) and m(E)

m(~) = l{mo(il ) + mo(E) ± [[mo(E) - m o(A)]2 + 4m2(EA)F /2}

(6.56)
Since the effect is very small [m(Eil) = 1.5 ± 0.4 Me V] the actual
masses
2m 2 (EA)
m(E) ~ mo(E) + mo(E) _ mo(A)
(6.57)
2m 2(Eil)
m(A) ~ mo(A) - mo(E) _ mo(A)

differ by a negligible amount from the unmixed ones.

The mixing angle ,\ is given by
2m(EA)
tan 2,\ = mo(A) _ mo(E) (6.58)

and to good approximation this reduces to

,\ ~ ~ [m(p) - m(n) + m(EO) - m(E+)]

(6.59)
- V3 m(A) - m(E)

numerically one finds ,\ = - 0.019 ± 0.006 radians, small but

larger than the order-of-magnitude parameter e2 j"Trnc character-
izing second order electromagnetic corrections.
158 INTRODUCTION TO UNITARY SYMM

Of the several applications discussed by Dalitz and von H

we review the induced AA7TO coupling and its influence on
binding energy in hypernuclei. Here we must note tha
physical 7T O state contains some 7] component because of m
given by Eq. (6.52). According to reference 17 the numerical
of m 2(7T 0 7]) is -2960 ± 365 (MeV)2. The physical7To is then
by

where tan 2fL = 2m 2(7T°7])[m 2(7T O) - m 2(7])]-1 reduces to

0.0105 ± 0.0013. Evaluation of the physical AA7TO v
<Aphysl 7TphYSO IA phyS ) then gives the coupling constant

gAAn = 2 sin ,\ cos ,\ cos fL gEAn + sin fL cos 2 ,\ gAAn

+ sin2'\sinfLgEEn
Since ,\ and fL are quite small, we can set the cosines eq
unity and approximate the sines by their arguments. Thu
gUn term in (6.61) can be dropped. According to S U(3), g
- gEAn and so (6.61) is approximately
2
2m(.EA) m (7T 0 7 ] ) ]
gAAn ~ [ m(A) _ m(.E) + m2(7]) _ m2(7TO) gUn
gAAn ~ (- 0.0485 ± 0.012)gEAn
The mirror hypernuclei AHe4 and AH4 have A binding en
2.33 ± 0.10 MeV and 2.03 ± 0.09 MeV, respectively. These
bers are close enough to support the idea of charge symme
the lambda-nucleon interaction but reveal a small discre
B = 0.30 ± 0.14 MeV apparently outside experimental err
Although gAAn 2 in (6.62) seems to be minute, the effect m
of disproportionate size because the existence of the AA7T v
permits the long range [~ljm(7T)] one-pi on-exchange
between A and N. On the other hand, the longest range c
independent force involves the exchange of two pions and s
range Ij2m(7T). Numerical estimates (with gEAn ~ gNNn)
LI B A = 0.21 ± 0.05 Me V arising from the one-pion-exc
III{OKEN SU(3) 159

potential. Evaluation of other charge-dependent effects bolster

I he conclusion that the S U(3) prediction (6.62) is essential in
bringing theory in line with experiment.

6.4 Magnetic Moments

I n the previous section the influence of electromagnetic forces
un the masses of the strongly interacting particles was assessed.
We now examine the implications of S U(3) for magnetic moments,
photoproduction reactions, and photon decay vertices. Eecall
that in the eightfold way the electric charge Q is G3 + (l/V3)G e,
or Tll, a component of a traceless tensor operator Tv". Some
results may be derived without assuming the trace condition.
Since the current operator jp.(x) commutes with U, the matrix
clements <al ja(Xl)" .j.(xn ) la> are the same for all members a of
a U-spin multiplet. An especially important case is the magnetic
mo ment. The magnetic moment of a quantum system is generally
defined to be the maximal projection of the operator

M = t Jr X j(r) d 3 x (6.63)

Since the M; commute with the U j we can derive relations like

where fL(.EA) is the transition magnetic moment, which can be

estimated from the decay .Eo ~ A + y.
From isospin conservation alone we know (Section 1.4)

(6.65)

Further relations emerge if we restrict M to be a traceless

operator. We can then write for the magnetic moment of the
baryon octet, using Eq. (6.32)

fL = aQ + b[(tQ + 1)2 - U(U + 1)] (6.66)

This formula exhibits the two independent constants expected in

160 INTRODUCTION TO UNITARY SYMMET

the matrix element <BI fLop IB ) (since 8 x 8 x 8 contains

twice).
We apply (6.66) to I:uo and Auo.

(6.6

Adding, we obtain
(6.6

one new relation. For the neutron we have fL(n) = fL(I:uO) = -

Using Eq. (6.65) and

(6.6

we find
(6.7

The transition moment is given by [use (6.67), (6.68), and (6.7

fL(I:A)
V3
= -2 b = --
V3
fL(n) (6.7
2

Using the U-spin equalities (6.64) we can convert (6.65) to t

form
(6.7

All these relations hold in the 8 U(3) symmetric limit for t

electromagnetic form factors of the baryons. The effect of sy
metry breaking on the magnetic moments and form factors h
not been studied, to the author's knowledge.
The most recent measurement (19) of the A magnetic mome
gave the result fL(A) = (-0.77 ± 0.27) en/2Mc, in reasonab
agreement with the 8U(3) value fL(A) = tfL(n) = -0.95 en/2M
(Here en/2Mc is the nuclear magneton.)
BROKEN SU(3) 161

Consider the magnetic moments of the baryon decuplet. The

moments are equal along the V -spin direction

fJ-(N*-) = fJ-(Y1*-) = fJ-(E*-) = fJ-(Q-)

fJ-(N*O) = fJ-( Yl*O) = fJ-(E*O) (6.73)
fJ-(NH) = fJ-(Y1H)

Actually, since 10· x 8 x 10 contains 1 only once, we expect that

all moments can be expressed in terms of a single constant. For
10, G2 = 6, tQ + V = 1, and Eq. (6.32) reduces to JL = const. Q,
or
(6.74)
We can also consider the transition magnetic moment fJ-T =
<B*I fJ- IB). This matrix element can be derived from photopro-
duction data; for example, the 3-3 resonance in pion photo-
production is excited mainly through a magnetic dipole transition.
Again only one independent constant occurs, since the matrix
element involves 10· x 8 x 8. To analyze this problem we use
V-spin invariance. The transitions permitted by charge and
hypercharge conservation are

P...".N·+

(6.75)

The baryons are grouped into the V-spin multiplets (p, -E+),
(n, Euo = -tEO + ('V3/2)AO, _EO), Auo = (V3/2)EO + tAO and
(E- ,E-). The decuplet V-spin states do not have any problems
of tricky minus signs or degeneracy. Note that since E- and E-
have V = 1/2, while Y 1 • - and E· - have V = 3/2, we obtain the
prediction
(6.76)
162 INTRODUCTION TO UNITARY SYMMET

Next we use the U-spin raising and lowering operators

<NU[ fL [p> = -<YI U [ U_fLU+ [E+> = -<Y1 U[ fL [E+>
(6.7
since U _ U + = U2 - U3 (U 3 + 1). Similarly, one finds t
equality
<N*O[ fL In> = <Yl*O[ fL [Euo> (6.7
= -t<Yl*O[ fL [Eo> + ~3 <Yl*O[ fL [A>
This result can be shortened by noting that <Yl*O[ fL [Auo> =
or
(6.7
Using this result, (6.78) can be re-expressed in the form '

Further motion along the N*o, U-spin triplet yields

One further relation which allows one to express all six nonze
matrix elements in terms of one number follows on applying t
"reflection" operator R = exp (i71'T2). Notice that in the matr
element <Nu [ fL [p>, L1 T = 1 so that only the isovector pa
of the magnetic moment operator contributes. Further RT [p
= - In> and RT [Nu> = [N*o> since <fL[ exp (i71'T2) [v>
(-l)T+uS u._ v (cf. Section 1.3). The isovector moment fLv tran
forms as RTfLvRT -1 = - fLv and so
(6.8
These results are summarized in the equations
<NU[ fL [p> = <N*O [ fL In> = _<8 *[ fL [8°>
0

-<Y1U[ fL [E+> = ~3 <Y1*0[ fL [Ao> (6.8

-2<Yl*0 [ fL [Ao>
Identical relations hold for the decay amplitudes B* --+ B +y
BROKEN SU(3)

For spinless mesons there are no magnetic momclltH. 'I'It<l

charge-form factors of the pseudoscalar octet arc all proporl iOllnl
to each other since the term in TIl linear in Q vuuiSit cH hy
charge conjugation invariance. We can write

(6.84.)
where F,,+ (q2) is the 7T+ form factor [F,,+ (0) = 1] and q2 the
invariant momentum transfer squared.
Little concrete evidence exists which bears on the validity of
most of these relations.
We have mentioned before that it is more difficult to correct
for symmetry breaking in 4-particle amplitudes than for 3-particle
vertices, unless resonances occur among pairs of the particles. For
this 'reason we give only a few examples of the use of U-spin to
analyze production amplitudes.
For Compton scattering one has obvious relations such as

(ypl s Iyp) = (yL'+ IS lyL'+)

(6.85)
(ynl S Iyn) = ( yEO I S lyEO)
etc. In lowest order in the electromagnetic interaction the process
y + a - 7 b has matrix element
(6.86)

where [j~,U] = O. Thus states la) and Ib) must have the same
value of U. Examples of this sort have been given by Lipkin (20).
The presentation of this section has been based on the work of
references 1, 2, 12, 17, and 20-24. For further results the reader
is referred to these papers.

References
1. G. Feldman and P. T . Matthews, Ann. Phys. (N. Y.), 31, 469 (1965).
2. J. H . Wojtaszek, R. E. Marshak, and Riazuddin, Phys. Rev., 136, BI053
(1964).
3. S. Coleman, S. Glashow, and D. Kleitman, Phys. Rev., 135, B779 (1964).
4. M. GeII-Mann, Phys. Rev., 125, 1067 (1962).
5. M. GeII-Mann, Phys. Rev., 106, 1296 (1957).
164 INTRODUCTION TO UNITARY SYMMETR

6. J. J. Sakurai, Phys. Rev. Letters, 9, 472 (1962); Phys. Rev., 132, 434 (1963
7. P. L. Connolly et aI., Phys. Rev. Letters, 10, 371 (1963).
8. R. R. Wilson and J. S. Levinger, Ann. Rev. Nuc. Sci., 14, 135 (1964).
9. R. F. Dashen and D. H. Sharp, Phys. Rev., 133, B1585 (1964).
10. F. Gursey, T. D. Lee, and M. Nauenberg, Phys. Rev., 135, B467 (1964).
11. S. Okubo, Prog. Theor. Phys. (Kyoto), 27, 949 (1962).
12. S. P. Rosen, Phys. Rev. Letters, II, 100 (1963).
13. B. Diu, Nuovo Gimento, 28, 466 (1963).
14. J. Ginibre, J. Math. Phys., 4, 720 (1963).
15. S. P. Rosen, J. Math. Phys., 5, 289 (1963).
16. H. Goldberg and Y. Lehrer-Ilamed, J. Math. Phys., 4, 501 (1963).
17. S. Coleman and S. L. Glashow, Phys. Rev. Letters, 6, 423 (1961).
18. R. H. Dalitz and F. von Hippe!, Phys. Letters, 10, 155 (1964).
19. D. A. Hill, K. K. Li, E. W. Jenkins, T. F. Kycia, and H. Ruderman, Ph
Rev. Letters, IS, 85 (1965).
20. H. J. Lipkin, Lie Groupsfor Pedestrian., Interscience, New York, 1965.
21. A. J . Macfarlane and E. C. G. Sudarshan, Nuovo Gimento, 31,1176 (191i4).
22. S. Okubo, Phys. Letters, 4, 14 (1963).
23. R. J. Oakes, Phys. Rev., 132, 2349 (1963).
24. N. Cabibbo and R. Gatto, Nuovo Gimento, 21, 872 (1961).
 Chapter 7

Crossing Symmetry in Strong Interactions

7.1 Crossing Matrices for Arbitrary Isospin

"Crossing symmetry" or, as it is sometimes known, the "sub-

stitution rule," is one of the most far-reaching consequences of
quantum field theory (1-3). In this chapter we discuss the rela-
tions among reaction amplitudes which follow from the invariance
of strong interactions under transformations of the internal
symmetry groups S U(2) (isospin) and S U(3). The development
follows reference 4 closely.
Since the spin structure of the amplitudes is not of concern
here, we consider only the case of spinless bosons. To be definite,
we begin with isospin amplitudes and consider "two-body"
reactions of the type

(s channel) (7.1)

among particles of isospins a, b, c, and d having third components

IX, fl, y , and 0, respectively. The four-momentum of particle a is
denoted by Pa' etc. When no confusion can arise the labels a, b, c,
and d are supposed to include the specification of the component
labels. By convention, we shall call (7.1) the s-channel reaction,
using Mandelstam's kinematical invariants (5). The amplitude for
(7.1) is physical only when s = (Pa + Pb)2 ~ max [(ma + mb)2,
(me + md)2]. The antiparticles of particles a, b, c, and d, denoted
by a, Ii, c, and iJ, have isospins a, b, c, and d and third components
- IX, -fl, - y, and - o.
165
166 INTRODUCTION TO UNITARY SYMMETR

c d c b b d

a b
(a) (b) (e)

Fig. 7.1. The three processes a + b->- c + d, a + d->- c + b, and a + c->- ;; +

are related by crossing symmetry.

In addition to (7.1), two varieties of crossed reactions must b

considered

a+d---'?c+b (u channel) (7.2

(t channel) (7.3

The three reactions under consideration are sketched in Fig. 7.1

The reactions (7.2) and (7.3) are physical in the domains of th
"crossed channels" u == (Pa + Pd)2 ~ max [(ma + md)2, (me + mb)2
and t == (Pa + Pe)2 ~ max [(ma + m e)2, (mb + md)2], respectively
The center-of-mass energy variables in the s, u, and t channel
are V;, V-;;, and vi, respectively. Of course the scattering angl
() has to satisfy (I cos () I ~ 1) for the reaction to be physical. Th
amplitudes for reactions (7.1)-(7.3) are denoted by

(cdl Ms lab):a + b ---'? c + d (7.4

(cbl Mu lad) :a + d ---'? c + b (7.5

(bdl M t lac):a + c---'? b + d (7.6

"Crossing symmetry" asserts that if the four momenta Pa' Pb

PC' and Pd (more precisely the invariants s, t, and u formed from
them) are continued to the physical domain of a crossed channel
CROSSING SYMMETRY IN STRONG INTERACTIONS 167

the continued amplitude actually describes the crossed reaction

in question. Explicitly, we have the connection

(cdl Ms lab)c = gsu(c~1 Mu lad) (7.7)

(cdl Ms lab)c = gst(~dl M t lac) (7.8)

where the subscript c denotes "continued" and the phases gsu

and gst, which depend on the isospin quantum numbers, will be
determined subsequently. We shall not describe the actual
method of performing the analytic continuation and shall
generally omit the label" c" in (7.7) and (7.8) which specifies that
the continuation has been performed. The results (7.7) and (7.8)
will be demonstrated in Section 7.2. Particular methods of per-
forming the continuation are given in references 1 and 6.
We now employ isospin conservation to express the amplitudes
(7.4)-(7.6) in terms of the isospin amplitudes for the various
channels. As explained in Chapter 1 we shall always use the
Condon-Shortley phase convention (7) to define the relative phase
of states within an isotopic multiplet. Using the notation of Rose
(8) for the S U(2) Clebsch-Gordan coefficient gives

(cdl Ms lab) = L M s(s')C(abs';a,(3)C(cds';y,o)

s'
(7.9)

(chi Mu lad) = L M u(u')C(adu';a,-o)C(cbu';y,-(3)

u'
(7.10)

(bdl M t lac) = L M t(t')C(act';a,-y)C(bdt';-(3,8)

t'
(7.11)

In these expressions, the amplitude Mv(v') denotes the amplitude

for scattering in a state of isospin v' in channel v.
Comparison of the crossing assumption (7.7) and (7.8) with Eq.
(7.9) shows that the continued isospin amplitudes can be ex-
pressed in terms of crossed-channel amplitudes by the use of
standard recoupling techniques. Employing the usual symmetry
properties of the Clebsch-Gordan coefficients one can rearrange
168 INTRODUCTION TO UNITARY SYMM

standard formulas (8) involving the recoupling (Racah) coeffi

to desired form. We find

C( abs' ;a(:3)C(cds' ;yo)

= L(2s' + l)(-l)Y - a+a-cW(abcd;s'u').C(adu';a - 0)
u'
.C(cbu';y - (:3) (
C( abs' ;af3)C( cds' ;yo)
= L (2s' + 1)(-1)6-a+a+dW(as't'd;bc).C(bdt'; - f30)
t'
.C(act';a - y) (

where W is the Racah coefficient as defined in reference 8.

The crossing matrices are defined by

Mu(u') = L Xu's,Ms(s')
s'

Mt(t') = L Xt's,Ms(s')
s'
(

We shall refer to Xus and Xts as the u-channel and t-ch

crossing matrices, respectively. From Eqs. (7.9)-(7.13) one
immediately

Xus = tsu- 1 (-1)Y-a+a-c(2s + l)W(abdc;su) (

Xts = tst -l( _1)6-a+a+d(2s + 1) W(astd;bc) (

The physical interpretation of these relations is easily un

stood from Fig. 7.2. Suppose that in the s channel particles a
b unite to form an intermediate state of isospin Ts (Fig. 7.2a
the t or u channels the kinematics is such that the" particle
is an exchanged "particle." To obtain the magnitudes o
various isospin amplitudes in the t or u channels we hav
project the s-channel states onto the appropriate states. Th
given particles a and b with isospin Ts and projection T3 = a
(similarly, c and d combine to give isospin Ts and proje
T3 = Y + 0 = a + (:3), what is the amplitude for a and il to
isospin Tu and projection a - 0, while c and b have is
CROSSING SYMMETRY IN STRONG INTERACTIONS 169

c d b c b d

o b
(a) (b) (c)

Fig. 7.2. Suppose particles a and b coalesce toform a state of isospin T. In the
crossed channels it appears that a particle of isospin T has been exchanged. The
amplitude for isospin T' in the crossed channel is then given by the appropriate
crossing matrix.

Tu and projection y - S? The answer is given by the recoup-

ling expressions (7.12) and (7.13).
In employing our results it is sometimes necessary to rearrange
the particle labels. Besides the usual symmetry properties of
S U(2) Clebsch-Gordan coefficients, we need a property which
follows from time-reversal invariance. It is well known that there
is a special phase convention in terms of which the S matrix is
symmetric (9). But since we are defining all amplitudes in terms
of the Condon-Shortley convention we have to insert the appro-
priate phase factor to correct for this fact. The result is easily
found:

(cdl M lab) = (_l)a+b-C-d(ab l M Icd) (7.18)

In Section 1.3 field operators were constructed for isospin

multiplets in accordance with the Condon-Shortley phase con-
vention. In order to discuss the crossing conditions in a systematic
way, we find it very useful to use the "reduction formalism" of
Lehmann, Symanzik, and Zimmerman (10) which relates S-matrix
elements to Fourier transforms of matrix elements of various
field operators. The field operators describing the isospin multi-
plets are those of Section 1.3. Recall that particles were there
170 INTRODUCTION TO UNITARY SYMMETRY

classified according to whether or not the set of antiparticles wer

the same as the original set of particles. The "self-conjugate
particles were said to he of type <p while" pair-conjugates" wer
of type .p.
Three essentially distinct types of crossing must he considered
the particles crossed are either (1) hoth of type <p, (2) hoth of typ
.p or (3) one <p and one .p. Even with the conventions of Eqs
(7.2) and (7.3) the lahels t and u are largely a matter of convention
For convenience we give results for cases (1-3) regarded as hot
u- and t-channel crossing. Thus consider the following pairs o
reactions.

Iu {a + ~b -+c +!a (7.19a

a+ <Pa -+ C + <Pb (7.19h

It
{a + <Pb -+ <Pc + d (7.20
a+ ;Pc -+ ;Pb + d (7.20h

lIu ta++ .pa ++

~b -+ C
-+ C
~a
.pb
(7.21
(7 .21

lIt ta++ -+:fc ++

~b
.pc -+.pb d
d
(7.22
(7.22

Illu {a + ~b -+c +!d (7.23

a+ .pa -+ C + <Pb (7.23

lIlt {a + ~b -+ !c + d (7.24
a+ .pc -+ <Pb + d (7.24

In order to discuss the crossing condition for (7.19) it is usef

to use Eq. (1.57) to invert Eq. (1.55):

(7.2
+-+
au*(k) = TJ-/i fd3x<p(-~)(x) ao fk(X) (7.2

One can then USe the reduction formalism to contract the pa

ticles band d in reaction (7.19a). (We follow the conventions
CROSSING SYMMETRY IN STRONG INTERACTIONS 171

Barton (11).) If particle b has momentum k and T3 component f3,

while particle d has momentum k' and T3 component the S a,
matrix for reaction (7.19a) is
S = ( c;k' da out I a;kbf3 in> (7.27)
Ignoring the unit operator which occurs for elastic scattering one
finds

S = i27]_/ f dxdy fk(X)fk,*(y)K}K/(cl T<?d(6)(y)<?b(-B)(x) la>

(7.28)
where Tis the time ordering operator and K} = Ox + mb • We 2

may also consider reaction (7.19b), where b has momentum q and

T3 component f3', and d has momentum q' and T3 component a'.
The S matrix for this crossed process is
SX = ( c;q'bf3' out I a;qda' in>
= i 27]_6,d f dxdy fq(x)fq,*(y)KxdKyb ( cl T<?b(B')(y)<?d(-6') (x) lit>
(7. 29)
Disregarding the (2W)1/2 factors that have to be extracted ."
obtain the Lorentz-invariant amplitude we see that the integ ra 11.1
of (7.29) coincides with that of (7.28) if we setq = -k',q' = - Ir ,
f3' = - f3, and a' = - a. Thus
7]_/S(c,k'da;a,kbf3) = 7]6dSX(c,-kb - f3; a, -k'd - a) (7.30)
Comparison of Eq. (7.30) with Eq. (7.7) indicates that for reac-
tions of type (7.19) the crossing phase is
ts} = 7]_/7]6 d = (_IV- a - b- d (7.3])
Similar calculations lead to the following phases:
ts/ = 7]-/7]/ = (_1)6-a-b-c (7.32)
tsull = ~/~6d = (_lv-a+b-d (7.:n)
ts/I = ~/~/ = (_1)6-a+b-c (7.:ltJ,)

tsu III = 7]_/~6d = (_IV- a- b- d (7. :1;;)

ts/ II = 7]-/~/ = (_1)6-a-b - c (7.:36)
172 INTRODUCTION TO UNITARY SYMMETR

The corresponding crossing matrices are therefore

Xu'! = XUS III = (_I)a+b-c+d(2s + 1) W(abdc;su) (7.37

X t,! = XtsIII = (_I)a+b+c+d(2s + 1) W(astd;bc) (7.38
Xus ll = (_I)a-b-c+d(2s + 1) W(abdc;su) (7.39
Xtsll = (_I)a-b+c+d(2s + I)W(astd;bc) (7.40
The 6j symbols, defined by

a
{c db f e} = (-I)a+b+c+dW(abdc;ef) (7.41

possess simpler symmetry properties than the Racah coefficient

Occasionally it is useful to express the crossing matrices in term
of the former. In addition extensive tables of 6j symbols exist. )
In applications, the relative phases exhibited in (7.37)-(7.40
are not always significant, but should be observed when severa
channels are being considered simultaneously in order to avoi
sign errors.
Before giving practical examples we make a few miscellaneou
remarks of interest for applications. First consider the importan
special case of elastic scattering (a = c, b = d). As is well know
(3) crossing symmetry plays an essential role in determining th
sign and magnitude of forces in the direct (s) channel due t
crossed (t and u channel) processes. A bound on these force
follows from the fact that (8) the Racah coefficients W(abcd;e
form a unitary matrix if supplied with the normalizing facto
[(2e + 1)(2f + 1)]1/ 2. Hence
IXusl = (2s + 1) IW(abdc;su) I '"~(2S+1)1/2
2u + 1
(7.42

IXtsl = (2s + 1) IW(astd;bc) I

= (2s + 1) IW(abcd;st) I """" (2S + 1)1/2
2t + 1 (7.43

From these bounds one learns that a high isospin state can lea
to a larger crossing matrix element for low isospin states I
crossed channels than in the opposite situation (high ~ low).
CROSSING SYMMETRY IN STRONG INTERACTIONS 173

This does not, of course, imply that the exchange of low isospin
states is unimportant since the resulting "forces" depend on
other variables such as mass, spin, and angular momentum. (For
example, the usual theory of the 3-3 resonance (N*) in 7rN scat-
tering relies on the attractive force in the T = J = 3/2, p-wave
due to nucleon (T = J = 1/2) exchange. The relevant isospin
crossing matrix elements are 2/3 for the amplitude in the N*
channel due to nucleon exchange and 4/3 in the nucleon channel
(T = J = 1/2, p-wave) due to N* exchange. In this example.
Eqs. (7.43) read 2/3 ~ (1/2)1 /2 (0.67 ~ 0.71) for X3/21 / 2 and
4/3 ~ 2112 (1.33 ~ 1.41) for X 1 / 23 / 2 • For this case the upper
bounds are closely approached by the actual values.)
A second important property of crossing matrices for elastic
scattering is that the square is unity:

L Xss,Xs's" =
s'
8ss" (7.44)

We shall prove an important generalization below.

This relation is a very good check on calculations and can often
be used to complete a crossing matrix if only part of it is known.
The physical content of (7.44) is, of course, that two crossings
return one to the same situation. The proof follows from the
orthogonality theorem for Racah coefficients (8). Consider the
u-channel crossing matrix:

L (2s' + 1)(2s" + l)W(abdc;s's)W(abdc;s"s')

s' (7.45)
L (2s' + 1)(2s" + l)W(abdc;s's)W(adbc;s's")
s'

the second equality following from a standard symmetry relation.

Now if b = d, i.e., if the crossed particles have the same isospin,
the orthogonality theorem applies and X2 = 1. The same result
holds for Xts if b = c, i.e., if the crossed particles have the same
isospin.
If all four particles have the same isospin then the t- and u-
channel crossing matrix elements are identical except for sign. In
 TABLE 7.1
Isospin Crossing Matrices and Their Inverses.

Process Xu, (X-I),u (X,,) (X-I),I

NK-..NK
NK-..NK (-1/2 3/2) (-1/2 - 3/2)
X,u XI'
NN-..KK
1/2 1/2 -1/2 1/2
N 1T -.. N1T
NiT-.. NiT (-1/3 4/3) X,U
(V6/3 2V6/3) (V6/6 1)
NN-.. ii1T 2/3 1/3 2/3 - 2/3 V6/6 -1/2
N1T-" N'1T
..... NiT-.. N';;; (2/3 - ViO/3) X,u C/3 ViO/3 ) C/2 5V3/6 )
....
-l
NN'-..ii1T - VlO/6 - 2/3 V3/3 - V30/15 VlO/4 - V30/12

N'1T-" N'1T eV3/3 -2V3/3 -V3 ) eV3/6 - viii/. _5V6/12)

( 1/6 -'/3 3/')
N'7i-.. N'7i -1 /3 11/15 3/5 X,u - VlO/6 - 2VlO/15 3VlO/I0 - V3/6 - VlO/I0 V6/3

C
N'R'-.. 7i1T 1/2 2/5 1/10 - V6/6 4V6/15 - V6/10 - V3/6 3VlO/20 - V6/12
1:1T-..1:1T ( 1/3 -I 5/3)
-1/3 1/2 5/6
I 5/3)
1:ii -..1:ii X,u 1/3 1/2 - 5/6 XI'
1:I: -.. 7i1T 1/3 1/2 1/6 1/3 -1/2 1/6
N'K-..N'K
N'K-..N'j( (-1/4 5/4) X,U
(-3V2/4 -5V2/4) ( - V2/4 - VlO/4 )
3/4 1/4 - y'lO/4 VlO/4 - V2/4 3viO/20
N'R'-..KK
CROSSING SYMMETRY IN STRONG INTERACTIONS 175

all cases the factor is (_I)S+I. A familiar example is provided by

the crossing matrix for 7T7T scattering. (See the I:7T ---+ I:7T reaction
in Table 7.1.) Moreover, for large values of isospin the crossing
matrices can be approximated by Legendre polynomials (6). This
follows by using well-known approximate formulas for Racah
coefficients with large arguments (8, 12). This latter application
has not yet assumed any practical importance.
In Table 7.1 we have listed crossing matrices for many cases
likely to arise in practice. To be definite we have labeled isospin
states by the names of appropriate real particles. Although the
reactions are listed as baryon-meson reactions, so that particles
a and c are always of the type .p, other reactions involving the
same isospins can be described by our results with simple modi-
fications. In practice the inverses of the crossing matrices defined
by Eqs. (7.14) and (7.15) are of importance and so are listed. In
using these results, it is important to adhere to the phase and
labeling conventions described previously.
To employ these results recall that the physical interpretation
of the crossing matrices is as follows. Suppose we have scattering
in the s channel in isospin state Sf. In the u channel this" crossed
scattering" can be looked upon as the "exchange" of a state
with isospin Sf which gives rise to certain forces in the various
allowed isospin values u. To find these coefficients one reads down
the column Sf; the force in state u f is then proportional to the
entry in the ufth row. Our tables include examples of the type
(ab ---+ cd): (t t ---+ t t); (t 1 ---+ t 1); <t 1 ---+ ~ 1); (! 1 ---+ ~ 1);
(11 ---+ 11); (~t ---+ ~ t). Explicit crossing matrices for reactions
involving particles of still higher isospins can be found in
reference 4.
The fact that crossing matrices are essentially the recoupling
coefficients was realized as early as 1954 by Dyson (13). The
present exposition (4) is a generalization of the work of Mandel-
starn et al. (14). The first general expression for arbitrary isospin
was apparently given by Barut and Unal (15). Isospin crossing
matrices for many-particle processes have been considered by
Kotanski and Zalewski (16).
176 INTRODUCTION TO UNITARY SYMMETR

7.2 8U(3) Crossing Matrices

In order to derive crossing matrices for reactions of the typ
(7.1) involving particles interacting through S U(3)-inv;uia
forces, we need the standard phase factor of Eq. (2.91) relating
representation N and its complex conjugate N*. Rather than u
this general form we shall specialize to the eightfold way. Th
representations of the latter have the especially simple phas
factor (-I)Q with integral Q and so requires less vigilance wi
signs than the general case. [The more general case has bee
analyzed by Nieto (17), who also works out in detail the crossin
matrices for 3 + 8 -+ 3 + 8, required for the study of Sakaton (
quark)-meson octet scattering.] The crossing matrix for 8 + 8-
8 + 8 was first worked out by Cutkosky (19) using projectio
operator techniques . Neville (20) approached the same proble
using tensorial methods. Here we use the generalization of th
method of the preceding section which emphasizes the connectio
between crossing matrices and the group recoupling coefficient
This is the method used by de Swart (21) and Nieto (17). [See al
the work by Krammer (18).]
Rewriting Eq. (5.58) for the Lorentz-invariant transition matr
M rather than the full S matrix gives for the s-, u-, and t-chann
amplitudes [the three channels are defined in Eqs. (7.1)-(7.3)].

(cdl Ms lab) = ll~' c: fLd

Vd
fLY')(fLa
V Va
fLb
Vb
~Y)Ms(fL'Y'Y) (7.4

~Y) M u(fL,y'y)(7.47
ll~'C:
fLb * fLY') (fLa fLd*
(cbl Mu lail)
Vb V Va vd

- M lac) =
(bdl t L (fLbVb_* fLd
Vd
fLY') (fLa
V ,Va
fLe *
ve
~Y)Mt(fL'Y'Y) (7.48
llVYY'

Here the notation is as follows. The particle symbol a (similar

for b, c, and d) includes specification of the representation fLa an
component Va = (Ta, T az , Ya). Similarly, Ii belongs to the repr
sentation fLa* and component Va = (T a,- T az ,- Ya)' The indic
CROSSING SYMMETRY IN STRONG INTERACTIONS 177

y(y') distinguish among equivalent representations in the initial

(final) states.
The crossing phases tsu and tst of Eqs. (1.7) and (1.8) are (in the
octet model) simply
tsu = (-I)Qb±Qa (7.49)
tst = (-I)Q b±Qc (7.50)

Combining Eqs. (1.7) and (1.8) with (7.46)-(7.48) we can use

the orthogonality relations for the Clebsch-Gordan coefficients to
obtain the crossing matrices Xus and Xts' defined by

Mu(fL,S'S) = 2: (fLS'SI Xus IfL'y'y)Ms(fL',y'y)

U'y'y
(7.51)

Mt(fL,S'S) = 2: (fLS'SI Xts IfL'y'y)Ms(fL',y'y)

"yy
(7.52)

The elements of the crossing matrices are given by:

x (fLC
Vc
fLa
Va
~'y,) (fLa
V Va
fLb
Vb
~'Y) (7.53)

(fLS'SI Xts IfL'y'y)

vav~v.v' (-I)Q (~b *

fLc *
b
±Q
c
fLa
Va
fLO') (fLa
V Va Vc
~o)
~' Y')(fLa ~'Y)
/La fLb (7.54)
X (fLC
Vc Va V Va Vb

The multiple summations occurring in these formulas are very

inconvenient if brute force is used in their evaluation. One
method which can be used to reduce the labor is to factor the
S U(3) Clebsch-Gordan coefficients into their constituent S U(2)
and isoscalar parts as in Eq. (4.8). Then some of the summations
can be performed, with resulting S U(2) Racah coefficients in the
178 INTRODUCTION TO UNITARY SYMMETR

summation. (It is scarcely surprising that the S U(3) crossin

matrices can be expressed as a sum of S U(2) crossing matrices.
We omit the details of this calculation (see reference 17). On
finds the results

where Zus is defined by

Zus( Ta Tb Tc Ta; T' T) = (-IVc -Ta(2 T' + 1) W( Ta Tb Tc Ta; T' T)

== (-IVb+2Tc+Td(2T' + 1)(Ta Tb TT')

Tc Ta
(7.56
and
(fLS'SI Xts IfL'y'y)

L
TaYaTbYb
TcYcTdYd

fLc fLa I
fL' y' ) (fLa
x ( Tc Y c Ta Y a T'Y'
fLb IfL' y )
Ta Y a Tb Y b T'Y'
(7.57

with Zts given by

CROSSING SYMMETRY IN STRONG INTERACTIONS 179

The quantities Zus and Zts are the same as the S U(2) crossing
matrices, up to a phase.
Since the crossing matrices (7.55) and (7.57) are independent of
T and Y, the latter may be chosen so as to take advantage of
special properties of the isoscalar factors to reduce the number
of terms in the sums.
In the especially important case of scattering of octet particles
Xus and X ts are identical up to a phase factor, the latter given by

where the gl factors are given in Table 4.l.

The 8 x 8 crossing matrices are given in Table 7.2. The upper
sign refers to the Xus and the lower to X ts . In each of these cases

TABLE 7.2
Octet Crossing Matrix (8 X 8 -+ 8 X 8). The Upper (Lower) Signs Refer
to the U-Channel (T-Channel) Crossing Matrices.
,
/L y'y

27 10 10· 8 11 8 12 8 21 822 1
27 7/40 ± 1/12 ± 1/12 1/5 0 0 ±1/3 1/8
10 ±9/40 1/4 1/4 ± 2/5 ± 1/V5 1/V5 0 + 1/8
10· ±9/40 1/4 1/4 ± 2/5 + 1/V5 -1/V5 0 + 1/8
811 27/40 ± 1/2 ± 1/2 -3/10 0 0 +1/2 1/8
/L6'6
8 12 0 ± V5/4 + V5/4 0 -1/2 ±1/2 0 0
821 0 V5/4 - V5/4 0 ±1/2 -1/2 0 0
822 ±9/8 0 0 + 1/2 0 0 1/2 +1/8
1 27/8 + 5/4 + 5/4 1 0 0 +1 1/8

X2 = 1 and so X = X-l. As in the previous section the column

labels refer to the "exchanged state ' and the row labels to the
"output force."
The result found for isospin crossing matrices, that X2 = 1
when the crossed particles have the same isospin, has a simple
generalization in S U(3). If the crossed particles belong to repre-
sentations that are complex conjugate to each other then X2 = 1
180 INTRODUCTION TO UNITARY SYMMETJ

TABLE 7.3
U-Channel Crossing Matrix Relating 8 + 10 ..... 8 + 10 to 8· + 10 ..... 8· + 1
8 10· 27 35*

8 115 -1/2 -9/ 20 714

10 -2/5 314 -9/40 71 8
Xu.
27 -2/15 -1/12 37/40 7124
35 2/5 1/4 9/40 1/8

follows from the orthogonality relations for the Clebsch-Gorda

coefficie~ts. For example, if JLa = JLb then Xus 2 = l.
The crossing matrices (20,21) for a = c = 8, b = d = 10 a
given in Tables 7.3 and 7.4.

TABLE 7.4
T-Channel Crossing Matrix Relating 8 + 10 ..... 8 + 10 to 8 + 8* ..... 10* + 11
1 81 82 27

8 VS/20 V2/S vIOlS 9v'i/20

10 VS/20 3V2/10 vIO/IO -9v'i/20
27 VS/20 -3V2/10 vIO/30 - V'i/20
X,.
35 VS/20 V2/10 -vIO/IO 9v'i/140

7.3 Dynamical Relation of the Baryon Octet and Decuplet

It has been emphasized previously that group theory, unaide
by dynamical considerations, is quite unable to predict whic
representations should occur iu nature. Muah progress has bee
made in understanding the dynamical relations among th
strongly interacting elementary particles in recent years. Th
understanding, mostly of a qualitative or semiquantitative natur
is based on imposing 5U(3) symmetry on the dynamical fram
work (3) of the so-called dispersion relations. Since our unde
standing in this area changes rapidly with time, only one especial
important example is given in this book. Further references an
a detailed discussion of the baryon resonance spectrum may b
found in reference 22.
CROSSING SYMMETRY IN STRONG INTERACTIONS 181

We first review the Chew-Low theory (23) of the 3-3 resonance .

The nucleon is regarded as fixed so that only p-wave mesons
interact. States are labeled by L 2T • 2J where L is the orbital and J
the total angular momentum. The four states P ll , P 13 , P 31 , uud
P33 are labeled by an index a which takes on the values 1, 2, :3,

and 4 corresponding to the four P states in the order list ed above.

Each of these states is described by the amplitude h,. ( w):

(7. 60)

where 8a is the phase shift in state a and v(p) is the cutoff fun ction,
which makes the theory exist. The amplitudes ha obey the
equation

h ( ) = Aa ~ foo dw' 1m ha(w') ~ A fOOd_w_'-,I,-m_h-"..A,--w-,-')

a W
W
+ 7r" W
,- W -
. +
Le
L.,
/J=l
aD
" W
,
+ w
(7.61)

"", /
-----,-----
/ /

~ (a) (b)

(e)

Fig. 7.3. Significant contributions to the low-energy baryon-meson scattering

are shown. Graph (a) is the "direct" baryon pole; graph (b) is the baryon "ex-
change" amplitude; (c) is the baryon resonance exchange amplitude; and (d) the
vector meson exchange amplitude.
182 INTRODUCTION TO UNITARY SYMMETR

Here w is the pion energy (p2 + 1'.2)1/2 and I'- the pion mass. Au
the coefficient of the sum of the renormalized Born approxima
tion terms (Fig. 7.3a,b)

r
_8)
-2
['J ~ .. (::
(7.62

where f2 = .08 is the unrationalized 71' N coupling constant an

AutJ is the p-wave crossing matrix

-4 -4

A
,e'
9 -2
4
-1
8
2
8
-1
2 D
The p-wave static theory crossing matrix is identical with th
(7.63

isospin crossing matrix

X = ~ (~1 ~) (7.64

of Table 7.1 and (7.63) is the direct product of two crossin

matrices of the form (7.64).
The direct pole term Fig. 7.3a contributes only to P l l in th
static model; the - 8 in (7.62) is made up from - 9 (direct pol
plus + 1 (crossed pole, Fig. 7 .3b). In (7.61) the contribution
crossed 71' N scattering (Fig. 7.3c) in state f3 on the state a is foun
by reading off the element in column f3 and row a. In particula
note that P l l crossed scattering gives the same set of coefficien
as that part of (7.62) due to "nucleon exchange" Fig. 7.3b.
Eq. 7.62 shows that the Born terms are repulsive in all state
except P 33 • Supposing the scattering in the former to be small, w
have for h33:

h 33
_
-
tr + -1 foo dw', 1m h33(W'). +-1 foo dw' 1m h33(W')
-
W tru W-W-le 971' u W' +w
(7.65
CROSSING SYMMETRY IN STRONG INTERACTIONS 183

Let us neglect the last term in (7.65) temporarily on the basis of

its small coefficient. Then assuming elastic unitarity, the trun-
cated equation has the standard N j D solution (3) (normalizing so
that D33(O) = 1)

D = 1 _ ~ f ao dw'k'3 v 2(k')N33 (w')

(7.66)
33 7T tL w'(w' - w - ie}

Fixing the cutoff so that Re D 33 (W r ) = 0 and replacing the real

part of the integral in (7.66) by a constant (effective range
approximation) gives the Breit-Wigner form

e
to
33
. J '"
su. °33 = tr 1. 'r'
Wr - W - 2~

For a small width we see that h33 has a pole near Wr with residue
-Y33 = - tJ2· (Y33 is also the reduced half-width.) More generally,
we define Y33 by
(7.68)

From (7.67) we see that in the "sharp resonance approximation"

(7.69)
Using the approximation (7.69) to perform the last integral in Eq.
(7.65) gives a two-pole problem which is easily solved. Again
assuming a linear D function, one finds Y33 = 1:f2. [Note that in
using (7.69) we did not assume Y33 = tJ2 but instead used (7.68)
to find Y33 self-consistently. Alternatively, one can obtain the
same answer by iteration.]
In the preceding discussion we showed how to build a 3-3
resonance from the attractive forces due to nucleon" exchange."
One is thereby able to understand the existence of, and the
numerical value of the reduced width of the 3-3 resonance. (The
position is not predicted, however.)
In the preceding discussion we regarded the nucleon and f2 as
"given" and derived the existence and half-width of the N*.
Next we investigate the Pll channel, which contains the nucleon
184 INTRODUCTION TO UNITARY SYMMETR

direct pole. Notice that N* exchange gives a large attractive for

in the P ll channel. hll obeys (approximately)

hll = _Yll + Yll + 16 ~ +~ Jao dw; 1m hd~') (7.7

W 9w 9 W + Wr 7T Il W - W- te

where Yll 3f2 is the negative residue of the direct pole, Fi

7.3a.
Res hll(O) [direct pole == -Yll (7.7

Comparing (7.71) and (7.68) one notes that the reduced ha

width plays the role of a coupling constant. (The small numeric
gain of distinguishing the" true" direct pole from the" exchang
pole, which really is a cut in the partial wave projection of th
relativistic nucleon exchange amplitude of Fig. 7.3b, is a luxu
perhaps undeserved by the crude static theory. We retain this di
tinction because the results attain thereby a certain simplicity
Since Y33 = iYll we see that the Y33 term in Eq. (7.70) is typical
larger than the attractive force (pole term) in (7.65) that gave ri
to the 3-3 resonance. Thus the first term in (7.70) might be
bound state pole, analogous to that which results from the secon
term of (7.65) if (7.69) is used to evaluate the integral. By arran
ing for Dll to have a zero at W = 0 instead of putting a pole
Nll one can obtain a relation between Yll and Y33 (tho latter no
regarded as given). Neglecting rlYll/W) and setting D ll ( -wr) =
gives
(7.7

By using a slightly different cutoff from that needed to obta

Re D 33 (W r ) = 0, as can be justified by appeal to the relativist
case, one can make D1(0) = o. Again approximating Dll(W) b
a straight line, Dll(W) ~ -w/w r gives from (7.71)

(7.7
which value is close to lYll = Y33 obtained before. In fact com
plete agreement can be achieved if we include the i(Yll/W) pie
of Eq. (7.70). To avoid confusing this term with the direct po
CROSSING SYMMETRY IN STRONG INTERACTIONS 185

one can use the trick of displacing the exchange pole to - e by

wntmg iYll/(W + e) and finally taking e --+ O. One then finds
(Dll ~ -w/wr )
(7.74)

so that the ratio of the coupling constants is consistently deter-

mined.
Although the quantitative validity of this calculation is ques-
tionable, a most intereS 'iing physical idea has been made plausible:
the nucleon can be regarded as a hound state of a pion and a
nucleon, held together principally by N* exchange, in precisely
the same way that the N* can be regarded as a composite particle
made up from a pion and a nucleon, held together mainly by N
exchange. Chew (24) has described this situation as the "reci-
procal" bootstrap, in analogy to the p meson bootstrap (25)
wherein the forces due to p meson exchange in 7T-7T scattering give
rise to the p. Although the N-N* bootstrap is not so simple, the
relation of N to N* is clearly reciprocal. According to this point
of view there is no way to tell which of N or N* is the more
fundamental. Likewise, Yll and Y33 are on a completely equivalent
footing. To be sure, the argument has a definitely circular charac-
ter of the chicken-egg type and does not explain, as yet, why
either particle exists. It will be noticed that only the ratio Yll/Y33
emerges from the preceding calculation. Moreover, the ratio is
independent of the various masses in the problem.
Improved calculations (with relativistic kinematics) of the
N-N* bootstrap have been made by Abers and Zemach (26) and
Ball and Wong (27). These models include the p-meson exchange
term (Fig. 7.3d) in addition to the N, N* exchange forces already
discussed. (We might remark here that even this force should be
regarded as part of the bootstrap since the sign and magnitude
of the NN --+ p --+ 27T amplitude depends essentially on N, N*
exchange forces. Thus we do not have to regard this contribution
as something extraneous to the bootstrap, which has to be put in
simply because it is there.) These authors conclude that the
observed 7T N coupling constant, N* width, and N, N* masses are
186 INTRODUCTION TO UNITARY SYMMET

compatible with the bootstrap model to within the expec

accuracy of such calculations (~ 20%).
One of the most impressive achievements of S U(3) is the sim
way in which the mutual existence of the baryon octet and P
decuplet can be understood dynamically (19, 28). We sketch h
the salient features of this theory, emphasizing the grou
theoretical aspects. In the degenerate mass approximation (M
baryon mass and fL = meson mass) one can classify the scatteri
amplitude according to the irreducible representations occurri
in 8 X 8. However, since 8 occurs twice, there will, in general,
off-diagonal elements corresponding to transitions 8 1 -+ 8 2 a
8 2 -+ 8 1 , corresponding to the F or D matrices occurring at t
two vertices in Fig. 7.3a,b. F2 and D2 are respect~vely 31 and
in the octet states so that the projection operators for a meson
state i going to state j (i, j = 1, 2, ... , 8 run over the appropri
octet) the projection operators are

(.9S2 )ji = tFjFi

(7.7
(.9S Jji = 1,Dj D j

The appropriately normalized operator Q for 8 1 ~ 8 2 is (19)

(7.7

Thus the direct pole contributions to the P 1 / 2 matrix amplitude

are given by

0a2
f2 ( 23 4V5a(1 - a))
h= (7.7
w 4V5a(1 - a) 12(1 - a)2

Only one eigenamplitude of (7.77) has the pole at w = 0 since t

determinant of the Born matrix (7.77) is zero. This state is

(7.7

where N is the normalization constant.

CROSSING SYMMETRY IN STRONG INTERACTIONS 137

To find the contribution of baryon exchange to th e v u .. ioU H

irreducible representations of the 8U(3) baryon- boson sca tt ~ .. ill ";
states one needs to combine the 8U(3) crossing matrix wit.h 1': /1 .
(7.77).
In the static model the crossing relation reads

ht#(w) = L etjX~"vs)hjv( -w)

;v
(7.79)

where i and j refer to the J = I =+= t states and fL and II to the

e
irreducible representations. tf is the p-wave crossing matrix
[identical to Eq. (7.64)] and X~"vS) is given in Table 7.2. For the
exchange of the baryon octet, the resulting force in J = 3(2
is measured by the coefficients Ft(a), defined by h(3/2)t =
(2f2(3w)F t (a). (For the T = J = 3(2 state in TTN scattering
F = 2.) The Fi are computed to be:
F(l) = 4[!a 2 - 3(1 _ a)2]
F27= ![3(1 _ a)2 + a2]
FlO. = ~[a2 - 3a(1 - a))
FlO = ~a(3 - 2a) (7.80)
FB2 = 6(1 - a)2 - l.a!J. a2

FBI = - [6(1 - a)2 + 2a 2]

FBI-B 2 = 0

These contributions are plotted in Fig. 7.4 as a function of the

mixing parameter a. It will be observed that for a > t, a P3/2
decuplet is quite likely.
The value of a can be obtained from the self-consistency require-
ment that the decuplet exchange force gives rise to the baryon
octet. From the crossing matrix one sees that decuplet exchange
gives the matrix for the octet subspace:

D = (tlv'S Iv'5)
0
(7.81)
188 INTRODUCTION TO UNITARY SYMMETR

-12
Fig. 7.4. The strength of the (octet) baryon exchange amplitude is given a
function of the mixing parameter a for the irreducible representations occurr
in 8 X 8. For comparison, note that F, would be 2 for nucleon exchange in T
3/2, TTN scattering.

dropping the easily calculable kinematical factors. (For th

spatial part the problem is the same as for the N-N· bootstrap
We can diagonalize D by MDM-l where

M = [2V6(1
1+ V6)]1/2
(1 +vSvI6
_
vS ) (7.8
1 + V6
CROSSING SYMMETRY IN STRONG INTERACTIONS 189

The eigenvalues of D are HI ± V6) of which only the positive

one gives attraction in the P3/2 state. Thus (7.82) corresponds
to an eigenamplitude N[l + V6)!fos 1 + V5!fos2' i.e., a definite
mixing parameter [cf. Eq. (7.78)]

3
a = ao == 2 + V6 = 0.674 (7.83)

a value very near that which gives a maximal TTN constituent in

the T = 1/2, Y = 1 states (see Fig. 3.10) and also agrees well with
the" experimental" values found by Glashow and Rosenfeld (29).
Moreover, this value requires the P 3 / 2 decuplet, as is seen in
Fig. 7.4. One still has to verify that the residues obtained in
the bootstrap are in rough agreement. This is roughly true in the
model used. (To improve the agreement one has to include the
effect of 10 exchange on 10 and octet exchange on the octet
amplitudes, in analogy to the TTN problem.)
Thus to a first approximation it appears possible to regard the
baryons and their P 3 / 2 excited states as a dynamical consequence
of 8 U(3) combined with conventional dispersion-theoretic
notions. Golowich (30) has made a detailed study of forces for
many states in meson-baryon scattering. The reader is referred
to this work, and also reference 22, for a fuller picture of the
dynamics of the meson- baryon interaction.
This example gives only a glimpse of the many interesting
problems in the region where symmetry and dynamics overlap.
Many believe that ' some day it will be possible to deduce the
existing symmetries from dynamical considerations. Capps (31)
and Cutkosky (32) have presented interesting models in which
this is the case. At present, however, broken 8U(3) is much more
precise than our imperfect understanding of dynamics.

References
1. J. M. Jauch and F. Rohrlich, Theory of Photons and Electrons, Addison-
Wesley, Reading, Mass., 1955, Chap. 8.
2. M. Gell-Mann and M. L. Goldberger, Proceedings of the Rochester Conference
on High Energy Physics (1956).
190 INTRODUCTION TO UNITARY SYMMETR

3. G. F. Chew, S·Matrix Theory of Strong Interaction., W . A. Benjamin, N

York, 1960.
4. P. Carruthers and J. P. Krisch, Ann. Phys. (N. Y.), 33, 1 (1965).
5. S. Mandelstam, Phys. Rev., 112, 1344 (1958).
6. T. L. Trueman, and G. C. Wick, Ann. Phys. (N. Y.), 26, 322 (1964).
7. E. U. Condon and G. H . Shortley, The Theory of Atomic Spectra, Cambrid
Univ. Press, London, 1957, Chap. 3.
8. M. E. Rose, Elementary Theory of Angular Momentum, John Wiley, New Yor
1957.
9. A. M. Baldin, V. I. Goldanskii, and I. L. Rosental, Kinematics of Nucle
Reactions, Oxford Univ. Press, London, 1961, p. 112.
10. H. Lehmann, K. Symanzik, and W. Zimmerman, Nuovo Gimento, 1, 2
(1955); 6, 319 (1957).
11. G. Barton, Introduction to Advanced Field Theory, Interscience, New Yor
1963, Chap. 4.
12. A. R. Edmonds, Angular Momentum in Quantum Mechanics, Princeton Un
Press, 1960.
13. F. J. Dyson, Phys. Rev., 100, 344 (1955).
14. S. Mandelstam, J. E. Paton, R. F. Peierls, and A. Q. Sarker, Ann. Ph
(N. Y.), 18, 198 (1962).
15. A. O. Barut and B. C. Unal, Nuovo Gimento, 28, 112 (1963).
16. A. Kotanski and K. Zalewski, Acta Phys. Pol., 26,117 (1964).
17. M. M. Nieto, Phys. Rev., 140, B434 (1965).
18. M. Krammer, Acta Phys. Austriaca, Suppl. 1, 183 (1964).
19. R. E. Cutkosky, Ann. Phys. (N. Y.), 23, 415 (1963).
20. D. Neville, Phys. Rev., 132, 844 (1963).
21. J. J. de Swart, Nuovo Gimento, 31, 420 (1964).
22. P. Carruthers, Lectures in Theoretical Physics, Vol. VIIb, Univ. of Colora
Press, Boulder, Colo., 1965, p. 82.
23. G. F. Chew and F. E. Low, Phys. Rev., 101, 1571 (1956).
24. G. F. Chew, Phys. Rev. Letters, 9, 233 (1962).
25. F. Zacharias en, Phys. Rev. Letters, 7, 112 (1961).
26. E. Abers and C. Zemach, Phys. Rev., 131, 2305 (1963).
27. J. S. Ball and D. Y. Wong, Phys. Rev., 133, Bl79 (1964).
28. A. W. Martin and K. C. Wali, Phys. Rev., 130, 2455 (1963); Nuovo Gimen
31, 1324 (1964).
29. S. L. Glashow and A. H. Rosenfeld, Phys. Rev. Letters, 10, 192 (1963).
30. E. Golowich, Phy•. Rev., 139, B1297 (1965).
31. R. H. Capps, Phys. Rev. Letters, 10, 312 (1963).
32. R. E . Cutkosky, Phys. Rev., 131, 1888 (1963).
 Appendix I

Isoscalar Factors

The SU(3) Clebsch-Gordan coefficients are given by the

product of isoscalar factors and S U(2) Clebsch-Gordan coefficients
as discussed in Section 4.2. Tables A-I-A-V are taken from the
review of de Swart cited in the preface. Tables A-I-A-V corre-
spond to the decompositions 8 x8 = 27 + 10 + 10* + 8 1 + 8 2 +
1, 8 X 10 = 35 + 27 + 10 + 8, 8 x 27 = 64 + 35 + 35* + 271 +
272 + 10 + 10* + 8,10 x 10 = 35 + 28 + 27 + 10*, and 10 x
10* = 64 + 27 + 8 + 1, respectively.

191
 TABLE A-I

¥=2 1=1

!~
II, ¥1; 12 , ¥2 fL y 11> ¥1; 12 , ¥2 10· fLy

t, l', t, 1 t, 1; t, 1 -1

¥=1 1= %
11> ¥1; 12 , ¥2 27 10 fL y 11> ¥1; 12 , ¥2 27 81 1 82 fL y

t, 1; 1, 0 V2/2 - V 2/2 t, 1; t, -1 V15/10 V lO/ lO 1/2 V 2/2

1, 0; t, 1 V2/2 V2/2 t, -1; t, 1 -V15/10 - V lO/10 -1/2 V 2/2
1, 0; 1, 0 -VlO/20 - V 15/5 V6/4 0
0, 0; 0, 0 3V30/ 20 -V5/5 -V2/4 0

¥=II=t ¥= - II=%
11> ¥1; 12 , ¥2 27 81 82 10· fL y II' ¥1; 12 , ¥2 27 10· fL y

t, l', 1, 0 V 5/10 3V 5/1O 1/2 -1/ 2 t, -1; 1, 0 V 2/2 - V 2/2

1, 0; t, 1 - V5/10 -3V 5/ 10 1/2 -1/ 2 1, O·, t, -1 V 2/2 V 2/2
t, 1; 0, 0 3V 5/1O - V 5/10 1/2 1/2
0, 0; t, 1 3V5/10 -V5/ 10 -1/2 -1/ 2

¥=O 1=2 ¥=-II=t

11> ¥1; 12 , ¥2 27 fL y 11> ¥ 1; 12 , ¥2 27 81 82 10 fLy

1, O·, 1, 0 1 t, -1; 1, 0 - V 5/10 -3V 5/10 1/2 1/2

1, O·, t, -1 V 5/10 3V 5/10 1/2 1/ 2
t , -1; 0, 0 3V 5/10 - V 5/10 -1 /2 1/ 2
0, 0; t, -1 3V 5/10 - V 5/1O 1/2 -1/2

¥ =O 1=1 ¥= -2 1=1
II' ¥1; 12 , ¥2 27 81 82 10 10· fL y II' ¥1; 12 , ¥2!27 fLy

t, -1; t, -1 1
t, 1; t, -1 V 5/5 - V 30/10 V 6/6 - V 6/6 V 6/6
.... t, -1; t 1 V 5/5 - V 30/10 - V 6/6 V 6/6 - V 6/6
'-0
0 0 ¥= -2 1=0
'" 1, O·, 1, 0 V 6/3 V 6/6 - V 6/6
1,
0,
O·,
0;
0,
1,
0
0
V 30/10
V 30/ lO
V 5/5
V 5/5
0
0
1/2
-1 /2
1/2
-1/ 2
11> ¥1; 12 ,
¥2 1~ fLy
t, -1; t, -1 1
 TABLE A-II

Y=2 1=2 Y=II=t

11 , Yl ; 12 , Y2 3 /-" 11 , Yl ; 12 , Y2 27 8 /-"

!, 1; ..
.l!c 1 1 : 1,
!,
0;
I-,
.,
.l!c

I,
1
0
VS/5 -2VS/5
-2VS/5 -VS/5
Y=2 1=1
I" Yl ; 12 , Y2 /-"

.... !, 1; .,
.l!c
127
1 -1
I" Yl ; 12 , Y2 35 27 /-"
""... t
11 ,
Y=I 1=
Yl ; 12 , Y2 1 35 /-"
1, 0;
!, -1; .,
1,
.l!c
1
0
V3/2 1/2
1/2 - V3/2
1, 0; 3
2, 1 I 1

Y=I 1= t 35 27 10
11 , Yl ; 12 , Y2 8 /-"
11 , Yl ; 12 , Y2 35 27 . 10 /-"
1, 0; 1, 0 -V3/6 - 3VS/IO V3/3 V30/I5
1, 0-, 3
2, 1 -1/4 -VS/4 vIO/4 0, 0-, 1, 0 V2/2 - V30/IO 0 -VS/5
0, 0; 3
2, 1 VS/4 -3/4 -V2/4 !, 1-, !, -1 V3/3 VS/5 V3/3 V30/I5
!, I-, I, 0 vIO/4 V2/4 1/2 !, -1; 3
2' 1 -V3/6 VS/IO V3/3 -2V30/15

Y=-2I=1

I" Yl ; 12 , Y2 27 8 /-" I" Yl ; 12 , Y2 35 27 /-"

1, 0; 1, 0 VIO/5 -V15/5 1, 0; 0, -2 1/2 V3/2

!, I-, t. -1 - V15/5 -VIO/5 !, -1; !, -1 V3/2 -1/2

Y=-II=t Y=-2I=0
11 , Yl ; 12 , Y2 35 27 /-" I" Yl ; 12 , Y2 35 10 /-"

1, 0-, !, -1 V2/2 V2/2 0, 0-, 0, -2 V2/2 V2/2

t. -1; 1, 0 V2/2 -V2/2 !, -1; t. -1 - V2/2 V2/2
Y=-II=! Y= -3 1= !
11 , Yl ; 12 , Y2 35 27 10 8 /-" I" Yl ; 12 , /-"
Y2135
1, 0; !, -1 -1/4 - 7VS/20 V2/4 VS/5 !, -1; 0, -2 1
0, 0; t, -1 3/4 - 3 VS/20 V2/4 -VS/5
!, 1; 0, -2 V2/4 3vIO/20 1/2 VIO/5
!, -1 ; 1, 0 -1/2 VS/IO V2/2 -VS/5
 TABLE A-III

Y=3 1= f
II' Y1 ; 12 , 110 Y1; 12 , Y2 64 35· 271 272
Y2164 1'-, 1'-,

1, l' 1, 2 1 1,
1,
l ,'
1;
.,
!'-

1,
1
1
V14j21
2V 70j21
-2 j3 V70j14 V6 j6
VSj6 -V14j28 V30j12
1, 0; 1, 2 -V21j21 - V 6j6 - V I0SjI4 I j2
0, 0; 1, 2 V 14j7 -l j2 -V70j28 -V6j4

Y=3 1=1
110 Y1 ; 12 , Y2!3S. 1'-, 110 Y1; 12 , Y2 35· 10· 1'-,

1, 1; 1, 2 -1 1, 1; 1; 1 - V 30j6 - V 6j6
1, O', 1, 2 V 6j6 -V30j6

Y=2I=2
II, Y1; 12 , Y2 64 35 110 Y1 ; 12 , Y2 64 35

.,
1'-, 1'-,

1, 1; 3
1 V6 j3 -V3j3 1, 1; 2, 0 V3 j3 - V 6j3
1, 0; 1, 2 V3j3 V6j3 1, O', .,
!'- 1 V6j3 V3j3

II, Y 1 ; 12 , Y2 64 35 35· 271 272 10 1'-,

1, l ,' 2, 0 V7 j21 -l jl2 -V5j6 SV42jS6 VIOj8 -SV 2j12

1,
1,
1,
l',
O',
0;
.,
1,
3

1,
0
1
1
5V7j21
-V21j63
5V42j63
-5j12
7V3j36
SV6j18
VSj6
-Vlsj9
V30j18
-3V42jS6
-SV 14jS6
V7j7
VIOj8
V30j8
0
- V 2j12
SV6j36
V3j9
0, 0; .,
!'- 1 VI0Sj21 -Vlsjl2 -V3j3 V70jS6 -V6j8 V30j12
1, -1; 1, 2 V3sj21 VSj6 -l j3 -V210j28 -V2j4 -VIOj6

Y=II=1

110 Y1 ; 12 , Y2 64 35· 271 272 10· 8 I'-y

1, 1; 1, 0 V3sj21 -VIOj6 3VI05j70 Ij2 -l j3 2VSjlS

1, l', 0, 0 2V3sj21 VI0j6 -VI0Sj70 I j2 I j3 VSj15
1, 0; !'-
2, 1 -V42j63 V3 j9 -V14j7 0 -V30j9 2V6j9
1, 0; 1, 1 -V210j63 -5Vlsj36 -19V70j280 V 6j8 7V6 j36 V30j4S
0, 0; 1, 1 V210j21 -Vlsjl2 -13 V 70j280 -V6j8 -V6j12 -V30j lS
1, -1; 1, 2 -2V7j21 V2j12 -V21j28 VSj4 -VSj6 -2j3

(continued)
 TABLE A-III ( continued)
Y=O 1= 3
110 Yl; 12, Y2 64 35 35" 271 272 fL y 11• Y 1 ; 12• fLy
Y2164
i. 1; 3
•• -1 2v'21/21 -v'3/3 v'3/6 - v'210/28 v'2/4 I. 0; 2. o 1
i. -1; !!.
2. 1 2v'21/21 v'3/6 -v'3/3 -v'210/28 - v' 2/4
0; 0 v' 210/21 v'30/12 v'30/12 0
I.
I.
O.
.
0-
0;
I.
2.
2.
0
0
0
v''7/7
v'2/4 -v'2/4
-1/2 -1/2
v'21/14
0
v'70/14
v'3/2
0

Y=O 1=1
11• Y1; 12• Y2 64 35 35" 271 272 10 10" 8 fLy

i. 1; •• -1
..
3
2v'3'5/63 -1/9 -5/18 3v'14/28 v'30/12 -2v'5/9 - v'5/9 4/9
i. -1; !!. 1 -2v'3'5/63 5/18 1/9 -3v'14/28 v'30/12 -v'5/9 -2v'5/9 -4/9
i. 1; i. -1 10v''7/63 -v'5/9 2v'5/9 -3v'70/70 v'6/6 -1/9 4/9 2v'5/45
i. -1; i. 1 10v''7/63 2v'5/9 - v'5/9 - 3v'70/70 - v'6/6 -4/9 1/9 2v'5/45
O. 0; I. 0 5v''7/21 - v'5/6 - v'5/6 -v'70/70 0 1/3 -1/3 -2v'5/15
I. 0; O. 0 10v''7/63 2v'5/9 2v'5/9 4v'70/70 0 2/9 -2/9 - v'5/45
0; 0 0 v'30/12
I.
I. .
0-
I.
2. 0 - v'14/63 v'1O/36
- v'30/12
v'1O/36
0
- v'3'5/14
1/2
0
v'6/6
5v'2/18
v'6/6
-5v'2/18
0
2v'1O/9

110 Y 1 ; 12• Y2 64 271 272 8 fL y

i. I-
i. -1;
. i.i. -1 2v'21/21 v' 210/70 v'2/2 2v'15/15
1 -2v'21/21 - v' 210/70 v'2/2 -2v'15/15
I. 0; I. 0 - v' 21/21 -4v' 210/70 0 2v'15/15
O. 0; O. 0 2v''7/7 -4v'70/70 0 - v'5/5

Y=-11= t
11• Y 1 ; 12• Y2 64 35" fLy

i. -1; 2. 0 v'3/3 -v'6/3

I. 0;
••
3
-1 v'6/3 v'3/3

(continued)
 TABLE A-III (continued)

II, Y1 ; 12 , Y2 64 35· 35 272 10·271 }l-y

t, -1; 2, 0 - V7/21 1/12 V5/6 - 5V4'2/56 v IO/8 - 5V2/12
t, -1; 1, 0 5V7/21 -5/12 V5/6 -3V4'2 /56 - v IO/8
1, O·, J!.2 , -1 V21/63 -7V3/36 V15/9 V2/12
1, O·, t, 5V14/56 V3O/8 5V6/36
-1 5V4'2/63 5V6/18 V3O/18
0, 0; J!.2 , V7/7 0 - V3/9
-1 V 105/21 - V15/12 - V3/3
t, 1; 1, V70/56 V6/8 - V3O/12
-2 V35/21 V5/6 -1/3 - V210/28 V2/4 vIO/6

II, Y1 ; 12 , Y2 64 35 272 10 271 8 }l-y

t, -1; 1, 0 -V35/21 vIO/6 -3VI05/70
t, -1; 0, 0 2V 35/21 v IO/6 - VI05/70 1/2 -1/3 - 2V5/15
1, 0; J!.2, -1 - V4'2 /63 -1/2 -1/3 V5/15
V3/9 - V14/7 0
1, 0; t, -1 V210/63 5V15/36 19V70/280 V6/8 V3O/9 2V6/9
0, 0; t, -1 V 210/21 - V15/12 -13V70/280 V6/8 7V6/36 - V3O/45
t, 1; 1, - 2 V6/12 - V3O/15
2V7/21 -V2/12 V21/28 V5/4 - V5/6 2/3

Y=-2I=2 Y=-2I=0
110 Y1 ; 12 , Y2 64 35· }l-y II' Y1; 12 , Y2 35 10 }l-y

t, -1; 2,
3
-1 V6/3 -V3/3 !. -1; t, -1 V30/6 - V6/6
1, 0; 1, -2 V3/3 V6/3 1, 0; 1, -2 V6/6 V 30/6

Y - -2 1=1 Y= -3 I=f
N
00 II' Y 1; 12 , Y2 64 35 271 272 }l-y 110 Y1; 12 ,
Y2164 }l-y
...... -
t , -1 ; J!.-1 - V I4/21 2/3 - V70/14 V 6/6 t , -1; 1, -2 1
t, - 1; " -1 2V70/21 V5/6 - V14/28 - V3O/12
t,
1, 0; 1, -2 V21/21 V6/6 VI05/14 1/2 Y= -3 1= t

Y2 1~
0, 0; 1, -2 V14/7 -1/2 - V70/28 V6/4 110 Y1; 12 , }l-y

t , -1; 1, -2 1
 TABLE A-IV

Y=2 1= 3

I
11 , Yl ; 12 , Y2 28 JLy

.,
110 Yl ; 12 , Y2 35 27 10' JLy
.:l. 1; .:l.
2, 2, 1 3
1; !, -1 -V6/6 -V2/2 -V3/ 3
Y =2 1= 2
t, -1 ; .:l.
2, 1 - V6/6 V2/2 -V3/ 3
1, 0; 1, 0 -V6/3 0 V3/3

., I~
110 Yl ; 12 , Y2 JLy
N> 3
Y=o 1= 0
0 1; .:l. 1 -1
N> 2,

I~
110 Yl ; 12 , Y2 JLy
Y=2 1=1 1, 0; 1, 0

I
110 Yl ; 12 , Y2 27 JLy
Y = - 1 1= %
.:l. 1; .:l.
2, 2, 1 1
11 , Yl ; 12 , Y2 28 35 27 10' p.y
Y=2 1=0 .:l.
2, 1;
.,
O. -2 V5/1O 1/ 2 3V5/ IO 1/ 2

., I~
110 Yl ; 12 , Y2 0, -2; 3
1 V5/lO - 1/ 2 3V 5/ IO -1/ 2
JLy
3
1, 0; !, -1 3V5/1O 1/2 - V5/ IO -1/2
1 ;, .:l.
2, 1 -1 !, -1 ; 1. 0 3V5/IO -1 /2 -V5/IO 1/2

Y=-II=t
110 Yl ; 12 , Y2 28 35 p.y 110 Yl ; 12 , Y2 35 27 p.y

3
i -,
1,
l',
O·, .,
1,
.:l.
0
1
V2/2 V2/2
V2/2 -V2/ 2
1, O·,
!, -1;
!, -1 -V2/2 -V2/2
1, 0 -V2/2 V2/2

Y=-2I=1
11 , Yl ; 12 , Y2 35 27 p.y 110 Yl ; 12 , Y2 28 35 27 p.y

.,
.:l.

1,
1;
O·, .,
1.
3
0
1
-V2/2 -V2/2
-V2/2 V2/2
1,
0,
0;
-2;
0,
1,
-2
0
V5/5 V2/2
V5/5 -V2/2
V30/10
V30/1O
!, -1; !, -1 V15/5 0 - V IO/5
Y=II=!
11 , Yl 12 , Y2 27 10' Y= -2 1= 0
; JL y

..
.:l.

1,
l',
0;
1,
.:l.
2,
0
1
V2/2 V2/2
V2/2 -V2/2
110

!.
Yl

-1;
; 12 ,

!.
Y21 35 p.y
-1 -1

Y=O 1=2 Y= -3 1= !
11 • Y1; 12 • Y2 28 35 27 p.y 11 • Yl ; 12 • Y2 28 35 p.y

.:l.
2. 1; !. -1 V5/5 V2/2 V30/10 !, -1 ; O. -2 V2/2 V2/2
!. -1;
••1.
3
1 V5/5 -V2/2 V30/10 O• -2 ; t .. -1 V2/2 - V2/2
1. 0; 0 V15/5 0 - V IO/5
Y= -4 1=0
110 Yl ; 12 •
Y2'1~ JLy

O. -2 ; O. -2 1
 TABLE A-V

Y=3I=t Y=O 1=1

I" Y,; 12 , Y2 64 fL y I" Y , ; 12 , Y2 64 27 8 fLy
-
.,
~
1; 0, 2 1 ~
2, 1; ~., -1 V21/21 VM/7 V6/3
1, O·, 1, 0 V210/21 3V35/3S -2viS/iS
t. -1; t, +1 V2IO/2I -4V35/3S VIS/IS
I" Y, ; 12 , Y2 64 fLy
- Y=O 1=0
.,
~
1; t, 1 1
I" Y1 ; 12 , Y2 64 27 8 1 fL y
.,
~ 1; ~ ., -1 V35/3S V21O/3S v IO/s v IO/s
Y=2I=I 1, 0; 1, 0 2VIOS/3S V70/I4 0 - V30/10
I" Y,; 12 , Y2 64 27 fLy t , -1; t, 1 3V70/3S - VIOS/3S - V5/S V5/S
0, -2; 0, 2 2V35/3S -3V2IO/70 v IO/s - VIO/IO
.,
~
1; t, 1 V21/7 2V7/7
1, 0; 0, 2 2V7/7 -V21/7 Y=-II=t
I" Y, ; 12 , Y2 64 fL y
Y=1 I=t -
1, 0; .,
~
-1 1
I" Y , ; 12 , Y2 64 fLy
- Y=-II=t
.,
~
1; 1, 0 1
I" Y , ; 12 , Y2 64 27 fLy

1, 0; ~., -1 VM/7 V35/7

t, -1; 1, 0 V35/7 - VM/7

Y=II=t Y=-1I=t
I" Y, ; 12 , Y2 64 27 fL y I" Y, ; 12, Y2 64 27 8 fLy
§.
2 , l', 1, 0 V14,f7 V35/7 1, 0; 3
-1 V7/7 4V35/3S v IO/s
1, 0; t, 1 V35/7 - VM/7 t, -1; "1, 0 2V7/7 V35/3S - v IO/s
0, -2; t, 1 VMj7 -3V70/3S V5/S
Y=II=t Y=-2I=2
I" Y, ; 12 , Y2 64 27 8
_Y:_21~~
fLy

.,
;t

1,
1; 1,
0; t,
0 V7/7 4V35/3S v IO/s
1 2V7/7 V35/3S - v iOlS
1-11_'_Y_'_;_1_2_'
t, -1; t, -1 1

t, -1; 0, 2 VI4/7 -3V70/3S V5/S Y=-2I=I

I" Y , ; 12 , Yz 64 27 fLy
Y=O 1 = 3
t, -1;
"1, -10 2V7/7
V21/7 2V7/7
3

1" Y, ; 12 , Y2164 fLy

.,
;t
1; ;!.2, -1 1
0, -2;
Y= -3 1= t
- V21/7

Y=O 1=2 I" Y, ; 12 ,

Y21~ fLy
I" Y , ; 12 , Y2 64 27 0, -2; 3
-1 1
fL y "
.,
;!.

1,
l',
O·,
.,
;!.

1,
-1 V7/7 V42/7
0 V42/7 -V7/7
 Appendix II

Vector and Axial Vector Currents;

Algebra of Currents

In Chapter 1 a general prescription was given for the construc-

tion of the isospin operator for Bose fields. The isospin was given
as the space integral of the time component of the isovector
current L:
f-+ f-+
ti4NJut</> or iifNJut'" (All-I)

for self-conjugate and pair-conjugate fields, respectively. The

analogous construction for spin 1/2 particles is clearly

(AIl-2)

where the 2t + 1 components of ",t transform as In Eq. (1.58).

Eq. (AIl-2) can also be written in the form

(AIl-3)

(Here ~ = "'tyo and the four Dirac matrices Yu are chosen to obey
Yu t = YoYuYo and {yu,Yv} = 2gu, . The metric tensor guv is diagonal
withg oo = -gjj = 1, i = 1,2,3.) The isospin current of the Dirac
field '" is then given by the second expression in Eq. (AIl-3).
These results have been used in constructing Eq. (5.33) .
The isospin current is of importance not only in electromag-
netic ap.d strong interactions but also in the description of
strangeness-conserving leptonic weak interactions. It has been
established that the latter are of the current-current type, where
the strong current is 'tJ u + Au where 'tJu is the combination
206
VECTOR AND AXIAL VECTOR CURRENTS 207

'U lU+ i'U 2/l of the isospin current 'U/l and A/l the Alu + iA 2u
component of the isovector axial vector current Au- Gell-Mann
has suggested that the vector and axial vector currents are of
prime importance for the description of the strong interactions
themselves (1).
In order to bring out the interesting mathematical relations
suggested by such an approach we shall restrict our attention to
a model in which only one basic spinor field is needed for the
construction of the vector and axial vector currents. At first we
consider isospin as the relevant internal symmetry. Denoting the
isospin doublet (comprised of p and n, if desired) by 1/;, we
construct

(AII-4)

where Y5 = iYOYlY2Y3, Y5 2 = 1, and Y5 t = Y5' In Eq. (AII-4) we

have extended T m so that TO = I; Tl, T2' and T3 as usual comprise
the Pauli spin vector. It is convenient to break up the 32 operrJ-
tors in Eq. (AII-4) into space and time components

(AlI-S)

Here m runs from 0 to 3 and a from 1 to 3. We have used the

relation aa = YOYaY5 to simplify these expressions. aa is the usual
4 x 4 spin matrix, defined by aa = abc (a, b, and c in cyclic
order; a, b, and c take on the values 1 to 3). The abc in turn are
208 INTRODUCTION TO UNITARY SYMMETRY

the space components of the matrix auv = i/2[yu'yv]' Note that

2CU' 00 is the baryon number density while COo is the isospin density.
First consider the current densities CU'om and Aom. The space
integrals of these 16 quantities

VOm = Jd 3xCU'om(x)
(AII-6)
Aom = Jd3xAom(x)
generate on equal-time commutation, the algebra of the form
U(2) X U(2). The following general formula is useful in this and
more complicated problems. Consider the linear combination of
components of !{It and !{I by two commuting matrices M and N:
!{It MiNi!{l. Using the canonical commutation rules

{!{I,,(x),!{Iu t(x')}t =t' = 0"uo(x - x')

(AII-7)
{!f,,(x),!{Iu(x')}t=t' = 0
one easily establishes the identity

[!{I t( x) MiNk!{l( x),!{1 t( x') M iNZtP(x')]t = t'

o(x - x')!{It(x){t{M!,Mi}[Nk,Nzl + t{Nk,Nz}[MhMi]}tP(x)
(AII-8)

(The reader is warned (2) that the uncritical use of (AII-7) can
lead to incorrect results arising from the singular nature of the
field operators.) Thus one only needs to know the multiplication
law for the Mi and N k , to obtain the commutator of the densities
!{It M!NktP.
For the quantities Vom and Aom of Eq. (AII-6) the explicit
forms of (AII-S) yield (with M = I or Ys and Nk = Tk/2)

(AII-9)

where Emnr is the usual antisymmetric tensor for the 1, 2, and 3

values of the indices but vanishes if any of the m, n, or r takes on
VECTOR AND AXIAL VECTOR CURRENTS 209

the value O. The eight Hermitian operators generate the algebra

of the group U(2) X U(2) as can be seen by making a change of
basis
(AII-I0)
The J ± m comprise two independent angular momenta for m = 1,
2, and 3. For all m we have

(All-ll)
[J ±m.JTn] = 0
The four hermitian operators J +m correspond to the generators of
the (reducible) unitary group U<+>(2) while the J -m correspond to
an independent group U<-)(2). Because of the second equation of
(AIl-ll), the group structure is simply the direct product
U<+)(2) X U<-)(2).
Since Voo and Aoo commute with the Yam' Aam (a = 1,2,3), we
can consider the latter six operators separately. The combinations
(AIl-I0) then generate two independent 5U(2) algebras. For
m, n, r = 1, 2, 3 (AIl-9) is well known as the algebra of the four-
dimensional real orthogonal group 0(4). Eq. (All-ll) exhibits the
isomorphism of 0(4) and 5U(2) X 5U(2).
Note that under space inversion (ljJ(x) -+ yoljJ( -x» YOm trans-
forms into YOm while Aom changes sign. Hence the parity opera-
tion P interchanges J + and J _.

PJ±P-l = J", (AIl-12)

The currents corresponding to J ±m have the form

The 1 ± Ys can be regarded as acting on both spinors; defining

the left- and right-handed spinors in the usual way by IjJL =
HI + ys)1jJ and IjJR = t(1 - ys)1jJ we see that J +/l involves only
IjJL' while J -/l is composed of IjJR' This fact accounts for the
adjective" chiral" often applied to U< + )(2) x U- (2) or 5 U< + )(2)
x 5U<-)(2).
210 INTRODUCTION TO UNITARY SYMMETRY

We next stress the crucial point that the group associated with
the algebra U< + )(2) x U< ->(2) cannot be a symmetry group of the
Hamiltonian. The known nonconservation of the axial vector
current implies that

i d~~t) = [A(t),H] =1= 0 (AII-13)

Even the mass, term (mif;«p) of a typical Lagrangian does not com-
mute with A. (14 = if;L«PR + 'ifiR«PL transforms as the representa-
tion (j+,j_) = (1/2,1/2) of SU<+)(2) x SU<-)(2), j± being the
magnitude of the" angular momenta" J ±.)
For this and other reasons there seems little hope that the
algebra of U< + )(2) x U< - )(2) can be a useful approximate sym-
metry algebra (i.e., an algebra whose operators commute with the
Hamiltonian) as was the case with SU(3). Nevertheless, one can
speculate that the algebra (AII-9) is true and investigate the con-
sequences of imposing these operator conditions on the usual
structure of field theory. The time of writing is not suitable for an
evaluation of the success of such efforts.
A glance at the operators comprising the vector and axial
vector currents (Eq. AII-S) shows that other algebras may be
formed. The space integrals of these densities are

(AII-l4)

It should be noticed that YOm and Aam are even under space
inversion while Yam and Aom are odd. Thus the commutator of
VECTOR AND AXIAL VECTOR CURRENTS 211

two odd operators is even and no closed algebra can be formed of

odd operators. However the 16 even parity operators have the
same commutation rules as the matrices Tm/2, UaTm/2. The com-
mutation rules are given by

(All-IS)

The first relation is old, while the second merely says that Aam is
an isoscalar (n = 0) or an isovector (n = 1,2,3). Hence the
possibly interesting dynamical information is contained in the
remaining commutator. The sixteen independent Hermitian
generators Yom' Aam (or Tm/2,UaTm/2) generate the algebra of the
four-dimensional unitary group U( 4).
Can the group U(4) be a symmetry group? First, note the
meaning of the transformations. In addition to the usual isospin
transformations (YOm) the operators Aam of the algebra correspond
to changes of the spin variables. Clearly Aam can coml!lute with
H only if the latter is independent of the spin (exclusive of the
orbital angular momentum) as well as the isospin. Such a spin-
isospin independent force was considered for nuclear physics,
with fair success as long ago as 1937 by Wigner (3). The idea has
been applied to particle physics recently by Sakita (4), and
Gursey and Radicati (5). The second point concerns the essen-
tially noncovariant nature of the separation of spin and orbital
momentum. Indeed, nearly all successes of this theory have been
confined to the low-energy domain. It may be useful to discuss
the U(4) mixed symmetry from the point of view of dynamical
models based on dispersion theory. In such models (for instance
the reciprocal bootstrap model of Section 7.3) one certainly
expects a definite correlation of angular momentum, parity and
isospin. In fact there is obviously a strong spin-orbit force opera-
tive in the four p states (PIH P 13 , P 31 , P 33 ) of the Chew-Low
theory. Thus the U(4) symmetry cannot be expected to hold for
212 INTRODUCTION TO UNITARY SYMMETRY

a Lagrangian, or for the S matrix. Nevertheless, in the Chew

N, N* bootstrap only the P ll and P33 states have such attractive
forces as to give binding. The difference in energy, of order 11M
relative to the nucleon mass, can then be regarded as due to a
small-spin isospin-dependent force within the states belonging to
the 20-dimensional representation to which the Pll(N) and
P33(N*) particles would be assigned in the U(4) scheme. The
forces in the remaining states P l3 and P 31 are extremely spin and
isospin dependent but are, after all, quite unimportant relative
to the Nand N* effects. It therefore appears that the U(4) (or
U(6» symmetry is of dynamical origin and has only a limited
validity.
If we recall that the projection operators a, and ii have the
properties
a = i(I + Y5)
(AII-I6)
a2 = a, a2 = a, aa = 0

then it becomes clear that two other U(4) algebras can be con-
structed from the set of Eq. (AII-l4). The two sets of 16 genera-
tors, labeled by ( +) and ( - ), are

(AII-I7)
K ± am -- fd 3 Xl!1(,(- Cl"r) _ fd3 x'1'.I.ta a (1 ±2 Y5) "2
~am ± Vam -
Tm '1'
.1.

Any operator labeled with plus commutes with any operator

labeled with minus. The sets (J +m,K+ am ) and (J -m,K- am ) have
the same commutation relations as (Vom,Aam) in Eq. (AII-I5) and
hence correspond to two independent U(4) groups labeled
U<+)(4) and U<-)(4). We therefore see that the space integrals of
the 32 operators making up the vector and axial vector currents
(AII-5) generate the algebra of U<+)(4) X U<-)(4) on equal-time
commutation (6).
The group U<+)(4) X U<-)(4) is not expected to be an approxi-
VECTOR AND AXIAL VECTOR CURRENTS 213

mate symmetry group describing the hadrons. However certain

of its subgroups may be good symmetry groups. For example the
U(4) defined by Eq. (All-IS) may be an approximate static sym-
metry group. The YOm and Aam are the positive parity components
of U<+)(4) x U<-)(4). Of course the isospin-baryon number group
U(2)y is contained in any of the groups under consideration.
Notice the following independent subgroup decompositions

U<+)(4) x U<-)(4) J U(4) J U(2)s x U(2)y

(AII-I8)
U<+)(4) x U<-)(4) J U<+)(2) x U<-)(2) ) U(2)y

Here S stands for the group containing spin transformations, as

discussed following Eq. (All-IS).
The preceding analysis, based as it was on the vector and axial
vector currents, may seem insufficiently general. After all, the
starting point was rooted in the weak interactions. Therefore, we
generalize by considering all 16 possible linear combinations of
~aif;b by means of the Dirac matrices: fj = 1, iys, Yu' YuYs, auv
(S,P, V,A, T). By means of the 16 fi and the 4 T m we can construct
the 64 densities D(fiT m):

(AII-I9)

We now proceed to examine the algebra obeyed by the space

integrals

(AII-20)

First note that the 16 linearly-independent Hermitian matrices

Yofj generate the algebra of U(4). (The quadratic form if;tif; is left
invariant by the arbitrary infinitesimal unitary transformation
if; ~ (1 + ie jYof j)if; where the ej are 16 real infinitesimal param-
eters.) It is then clear that the 64 Hermitian operators A(fjTm)
generate the algebra of the group U(8).
In order to investigate the sub algebras contained in this algebra
we write out the 64 densities, along with appropriate specializa-
tions to space and time components.
214 INTRODUCTION TO UNITARY SYMMETRY

(AII-21)

In Eq. (AII-21), the Roman letters a and b take on the values

1,2, and 3.
The addition of the scalar and tensor terms permits the con-
struction of two new U(4) algehras similar to U(4) considered
previously, with Yo replacing ),5. The matrix content of the linear
comhinations is
1 ± Yo Tm
(AII-22)
--2-2'

with a = 1, 2, 3. Note that (1 ± yo)/2 are energy (Casimir) pro-

jection operators in the rest frame, and so one might follow
Marshak and Okuho (7) in calling these suhgroups "Casimir suh-
groups" U c ±(4) in contrast to the "chiral" groups U(±)(4) hased
on the helicity projection operators (1 ± Y5)/2.
Thus the group U(B) contains, in addition to the chiral suh-
group U<+)(4) x U<-)(4), the Casimir suhgroup U/+)(4) x
U c <-)(4). Note that the latter group also contains U(4) as a suh-
VECTOR AND AXIAL VECTOR CURRENTS 215

group. We have previously noted the essentially low-energy

character of U( 4). Therefore it is of some interest (8) that the
Casimir subgroup contains another U(4) subgroup based on the
relativistic spin operator Sa with components 1, yoa x' yoa y, and
a z . The space components of Sa clearly obey the same algebra as
the Pauli spin matrices. The U(4) algebra based on the 16
matrices SaTm/2 (rather, the U(6) counterpart) was proposed by
Barnes, Carruthers, and von Hippe! (9) as a phenomenological
symmetry of the vertex function. A peculiar feature of this group
is its selection of a special direction (z), the direction of motion of
a particle. It appears then that this symmetry cannot be fulfilled
in the traditional sense but only for a selected class of matrix
elements. In fact the generators do not commute with the kinetic
energy, although the mass term is invariant. Indeed subsequent
applications (10) of this idea to scattering reactions have not (11)
reproduced the mysterious success of the collinear group [as
named by Dashen and Gell-Mann (8)] for the vertex function
(9, 12).
In practice it is quite useful to know how the 64 densities
D(r!Tm) transform under the operation of G conjugation. This
operation was introduced in Section 1.4. It is G = e exp (i7TT 2 )
where e is defined for Dirac fields by
(AII-23)

T denoting the transpose and the matrix C defined by CYuTC-1 =

- Yw Under C, the isoscalar covariants transform as
(AII-24)
where Wi = + 1 for r = S, A, and P and Wi = - 1 for r = V,
and T. Next consider the (2t + I)-component field transforming
in the standard way [Eq. (1.58)] under isospin rotations O(a) =
exp (ia·T)
(AII-25)
216 INTRODUCTION TO UNITARY SYMMETRY

It is then elementary to show that the isospin densities transform

as
(AII-27)

For example, the isovector vector current is even under G, while

the isovector axial vector current is odd. This transformation
corresponds to the invariance of the SU(4) algebra of (All-IS)
under the substitution Aa ---?- - A a, Vo ---?- + Vo, Aao ---?- A ao .
It will be noted that under charge conjugation the densities
corresponding to the different components of the relativistic spin
operator transform differently, since Sz comes from the axial
vector density and Sx, Sy from the tensor density:

(AII-28)
(Sx,Sy) = f d xtjho(ux,uy)if;
3
---?-
~
-(Sx'Sy)

Finally, we note the changes of notation necessary to extend

the previous results to the case in which the internal symmetry
group is SU(3) instead of SU(2). In place of the isospinor if; we
introduce the quark field q transforming according to the funda-
mental representation 3.
In addition to the eight 3 x 3 matrices Ai of Chapter 2 it is
convenient to introduce the unit matrix. Defining Ao = V2 /3 I
we see that the commutation rules may be expressed as

[Ai,Aj] = 2ij;j/c Ak
(AII-29)
Pi,Aj} = 2dijk Ak
where i, j, and k run from 0 to 8. j;jk is defined to be zero if any
index takes on the value zero. The coefficient dijk is defined as
before except that it is equal to (2/3)1/2 when one index is zero
and the other two · are equal.
In place of the 16 x 4 = 64 densities of Eq. (AII-21) we now
have 16 x 9 = 144 Hermitian generators

(AII-30)
VECTOR AND AXIAL VECTOR CURRENTS 217

These (12)2 operators generate (8) the algebra ofthe unitary group
in twelve dimensions U(12). Notable subgroups of U(I2) include
the chiral and Casimir subgroups U<+)(6) X U<-)(6) based on the
structures (1 ± Y5)i\m/4, ua(I ± Y5)i\m/4, and (1 ± Yo)i\mf4'
ua(I ± Yo)i\m/4, respectively. The chiral subgroup contains the
"static" U(6) group of Sakita, Gursey, and Radicati, while the
Casimir subgroup contains, in addition, the collinear group of
ref. 9, constructed from the relativistic spin operator as discussed
following Eq. (AII-22). Further developments and a description
of applications made to date will be found in a paper by
Ne'eman (13).

References for Appendix n

1. M. Gell.Mann, Phys. Rev., 125, 1067 (1962).
2. J. Schwinger, Phys. Rev. Letters, 3, 296 (1959).
3. E. P. Wigner, Phys. Rev., 51, 106 (1957).
4. B. Sakita, Phys. Rev., 136, B1756 (1964).
5. F. Gursey, and L. A. Radicati, Phys. Rev. Letters, 13, 173 (1964).
6. R. P. Feynman, M. Gell·Mann, and G. Zweig, Phys. Rev. Letters, 13, 678 (1964).
7. R. E. Marshak, and S. Okubo, Phys. Rev. Letters, 13, 818 (1964).
8. R. Dashen, and M. Gell·Mann, Phys. Lell., 17,145 (1965); 17,148 (1965).
9. K. J. Barnes, P. Carruthers, and F. von Hippel, Phys. Rev. Lellers, 14, 82
(1965).
10. H. J. Lipkin, and S. Meshkov, Phys. Rev. Lellers, 14, 670 (1965).
11. J. D. Jackson, Phys. Rev. Lellers, IS, 990 (1965).
12. K. J. Barnes, Phys. Rev. Lellers, 14, 798 (1965).
13. Y. Ne'eman, Lectures given at the Pacific Summer School in Physics, Honolulu,
1965, to be published.
 Author Index

Abers, E., 189, 190 Connolly, P. L., 164

Altarelli, G., 140 Cutkosky, R. E., 52, 99, 115, 131, 140,
176, 189, 190

Baird, G., 55, 74, 96, 97, 115

Baldin, A. M., 190 Dalitz, R. R., 22, 54, 141, 157, 158, 164
Ball, J . S., 121, 140, 185, 190 Dashen, R. F., 149, 164, 215, 217
Barbaro-Galtieri, A., 54, 141 Diu, B., 152, 164
Barkas, W. R., 54, 141 Donnachie, A., 141
Barnes, K. J., 215, 217 Dreitlein, J., 54, 96, 115
Barton, G., 171, 190 Dullemond, C., 40, 54, 140
Barut, A. 0.,175,190 Dyson, F . J ., 175, 190
Bastien, P. L., 54, 141
Becchi, C., 140
Behrends, R., 54, 96, 99, 115 Eberle, E., 140
Biedenharn, L. C., 54, 74, 76, 96, 97, Edmonds, A. R., 14, 21,96, 115, 190
99, 115, 183
Breit, G., 183
Bronzan, J. B., 95, 96 Feldman, G., 144, 163
Buccella, D ., 140 Feynman, R. P., 217
Bulos, F ., 141 Frazer, W. R ., 121, 140
Fronsdal, C., 54, 96, 115
Frye, G., 120, 140
Cabibbo, N., 164 Fulco, J. R., 121, 140
Capps, R. R ., 189, 190
Carruthers, P., 21, 140, 141, 190, 215,
217 Gasiorowicz, S., 59, 96, 99, 115
Casimir, R. B. G., 214 Gatto, R., 140, 164
Chew, G. F., 84, 96, 185, 190,212 Gelfand, N., 141
Chi~ton, F., 54 Gell-Mann, M., 19, 25, 39, 54, 77, 84,
Coleman, S., 113, 115, 155, 163 96, 128, 140, 163, 189, 207, 215, 217
Condon, E. U., 167, 190 Ginibre, J., 152, 164
219
220 AUTHOR INDEX

Glashow, S. L ., 84, 96, 128, 131, 134, Li, K. K., 164

140, 141, 155, 163, 189, 190 Lipkin, H. J., 35, 125, 139, 140, 141
Goldanskii, V. I., 190 163, 164, 217
Goldberg, H., 152, 164 Low, F. E., 95, 96, 181, 190
Goldberger, M. L., 190 Lurie, D., 105, 115
Golowich, E., 132, 140, 189
Giirsey, F., 39, 54, 164, 211, 217
Gupta, N., 140 Macfarlane, A. J., 40,55,105, US, 140
164
Mandelstam, S., 165, 175, 190
Hamermesh, M., 55,96, 97, 109, 114 Marshak, R. E., 6, 55, 144, 163, 214
Hamilton, J., 140 217
Hatsukade, S., 120, 140 Martin, A. W., 126, 140, 190-
Hill, D. A., 164 Matthews, P. T., 144, 163
Hippel, F. v., 157, 158, 164, 215, 217 McNamee, P., 54
Meshkov, S., 35, 139, 141, 217
Messiah, A., 109, 115
Iddings, C. K., 140 Michel, L., 21
Ikeda, M., 25, 54 Mills, R., 127, 140
Morpurgo, G., 140
Mukunda, N., 115
Jackson, J. D., 217
Jauch, J. M., 189
Jenkins, E. W., 164 Nauenberg, M., 39, 54, 164
Ne'eman, Y., 25, 54, 217
Neville, D., 176, 190
Kallen, G., 21
Nieto, M., 176, 190
Kirz, J., 54, 141 Nishijima, K., 19
Klein, A., 76, 96
Kleitman, D., 163
Kotanski, A., 175, 190 Oakes, R. J., 164
Krammer, M., 176, 190 Ogawa, S., 25, 54
Krisch, J. P., 21, 190 Ohnuki, Y., 25, 54
Kuo, T. K., 120, 140 Okubo, S., 32, 48, 54, 84, 113, 115, 151
Kycia, T. F., 164 164, 217

Lea, A. T., 140 Pandit, L. K., 115

Lee, B. W., 54, 96, 115 Paton, J. E., 190
Lee, T. D., 21, 39, 164 Peierls, R. F., 190
Lehmann, H., 169, 190
Lehrer-Ilamed, Y., 152, 164
Levinger, J. S., 164 Racah, G., 99, 115
Levinson, C. A., 35 Radicati, L. A., 211, 217
AUTHOR INDEX 221

Rashid, M. A., 96 Swart, J. J. de, 40, 54,62, 96, 99, 106,

Riazuddin, 163 107, 113, 115, 126, 140, 176
Rohrlich, F., 189 Symanzik, K., 169, 190
Roman, P., 6, 21, 54
Roos, M., 141
Tarjanne, P., 54, 96
Rose, M. E., 21, 100, 115, 167, 190
Tinkham, M., 104, 115
Rosen, S. P., 152, 164
Trueman, T. L., 190
Rosenfeld, A. H., 22, 54, 84, 96, 131,
134, 141, 189, 190
Rosental', I. L., 190 Unal, B. C., 175, 190
Ross, M., 54
Ruderman, H., 164 Wali, K. C., 126, 132, 140, 141, 190
Warnock, R L., 120, 132, 140, 141
Weyl, H ., 97, 114
Sakata, S., 25, 54 Wick, G. C., 190
Sakita, B., 211, 217 Wigner, E. P., 21, 183, 211, 217
Sakurai, J. J., 96, 127, 130, 140, 147, Wilson, R R, 164
164 Wojtaszek, J. H., 163
Sarker, A. Q., 190 Wong, D. Y., 121, 140, 185, 190
Sarkissian, M. der, 141
Schnitzer, H. J., 120, 140
Yang, C. N., 21, 127, 140
Schwinger, J., 217
Yodh, G. B., 139, 141
Sharp, D. H., 149, 164
Shortley, G. H., 167, 190
Singh, V., 140 Zachariasen, F., 190
Snow, G. A., 139, 141 Zalewski, K ., 175, 190
Speiser, D., 113, 115 Zemach, C., 185, 190
Sudarshan, E. C. G., 6, 40, 54, 140, Zimmerman, W., 169, 190
164 Zweig, G., 39, 54, 96, 217
 Subject Index

Angular momentum and isospin, in Compton scattering relations, 163

5U(2),1-21 Coupling constants, 116 if.
experimental evidence for, 118-120
isospin couplings of baryons and
Baryon, assignment to regular repre- pseudo scalar mesons, 12
sentation, 79 between octet baryons and pseudo-
interaction with pseudo scalar mesons, scalar mesons, 118
9,116 between octet baryons and vector
interaction with vector mesons, 120- mesons, 120-122
122 between octet mesons, 125
magnetic moments, 159-163 in Sakata model, 123
Bootstrap, 180 if. Crossing matrices, bounds on, 172
and Chew-Low theory, 183 properties of, 173
relation between baryon octet and for 5 U(2) (isospin), 168
decuplet, 186 for 5U(3), 176-180
Branching ratios, 128-136 Crossing phases, for 5 U(2), 172
for members of baryon decuplet, 130 for 5 U(3), 177
for octet into two octets, 132-133 Crossing symmetry, 165 if.
for vector meson decays, 121, 130 Current, algebra, 206
Broken 5U(3), decay of P3 /2 decuplet, electromagnetic, 159
130 isospin, 121, 127, 206-207
phase space correction, 129 strangeness changing, 127, 206- 207
structure of Lagrangian, 144

dlJk' definition of, 30

Charge conjugation, 16
Decay widths, see Branching ratios
Chew-Low theory of 3-3 resonance,
D I F ratio, see Mixing parameter
188 if.
Dimension formula, 99, 112
Collinear group, 215
D matrices, 52 if.
Complex conjugate representations,
for 5U(2), 7, 15
for 5U(3), 42 if.
inequivalence of 3" to 3, 42 Eigenvalue lattices, 97-98
223
224 SUBJECT INDEX

Electromagnetism, correlation of form Magnetic moment relations, hyperon,

factor data with pNN couplings, 20
120-121 octet baryons, 159
induced AA7T coupling, 158 octet-decuplet transition moments,
sigma-lambda mixing, 157 161
transformation of e-m interaction, in Mass formula, for baryon octet, 146
SU(2), 13 equal spacing law in U-spin multi-
in SU(3), 139, 159 plet, 146
trimsition masses, 156 Gell-Mann-Okubo, general form, 151-
violation of isospin conservation by, 152
180 for meson octet, 146
neglecting electromagnetism, 145-
/;Jk' definition of, 30 152
Field operators, 102 for Pal2 decuplet, 146
F matrices, 50 If. Mixing angle, for octet states, 133
for phi-omega problem, 148
for sigma-lambda problem, 157
G-conjugation, 18, 215 Mixing parameter, bootstrap deter-
Gell-Mann-Nishijima formula, 19 mination of, 120
Gell-Mann-Okubo mass formula, 152 P B B coupling, 116
GL(n), 108-109 universality of, 136

Hypercharge, in eightfold way, 37

in Sakata model, 33, 37 O(n),109

Isospin, 5 If.
Parallelogram law, 153
conservation of isospin current, 121,
Phase convention, Condon-Shortley;
127
SU(2), 13
crossing matrices, 168
for SU(3); complex conjugate repre-
field operators for isomultiplets, 13
sentation, 43
meson couplings, 125, 126
Phi-omega mixing, 147 If.
pair-conjugate multiplets, 156, 169-
Photoproduction, implication of iso-
170
spin, 20
rho-nucleon couplings, 121
self-conjugate multiplets, 13, 169-
170 Quarks, 77
violation of isospin conservation by model for broken symmetry, 143-144
electromagnetism, 18
Yukawa couplings of pseudoscalar
mesons and baryons, 9, 116 If. Racah coefficients, connection with
crossing matrices, 168
Lagrangian, form for strong and elec- R-conjugation, 95
tromagnetic interactions, 144 elfect on 8 X 8 states, 95
SUBJECT INDEX 225

Reaction amplitudes, implications of Hermitian gencrulurli 1'0" tlH~ rtiKul ur

5 U(2)-invariance, 167 representatioll, 711
implications of 5U(3)-invariances, infinitesimal g CJU, r'HII' '' ''. Z( I, :ll) :\0,
137 32,99
Reduced matrix elements, for 5 U(2), 20 irreducible rcprc "w lIllIl HII" II I'. r,f, fr..
for 5U(3), 104 97 ff,
U Reflection" operators, 39 irreducible ten!!.,.'" , 1111\
Regular representation, general defini- isoscalar factors, I () I
tion,49 physical interpl'ot III iOIl ." H~ I '1t ~\t1 111 n
for 5U(2), 9 tions, 76- 84
for 5U(3), 49 (p, q)notation for r" I"I".. \lIl li l lttlUl "fl j

Root vectors, 28 raising and loworillJ( 01'f~ I I II IH ',I 'tI' "I

Rotations, 1 ff_ reduction of the eli ,."", I" ." 1,,. ,II II ,
and angular momentum, 5 93-94
regular represent 1111011 , I" , "I)
representation lIIul. ( I -' (OJlilulllI
Sakata model, 25
nality of), 10:l I iI,1
hypercharge in, 33, 37
scattering amplilllll "., II ,
Substitution rule, see Crossing sym-
state vectors: 10" .1. I". ,. I "I",
metry
representutiou, II ~
5U(2), Clebsch-Gordan coefficients for,
symmetry properll,," "f ,I" t It 10 , II
6,20
Gordan coefli ci""' M, III '. 11111
complex conjugate representations
symmetry in strong illl N Ut I "U ll!
of, 7, 15
55
Condon-Shortley phase convention,
tensor operator, 101\
13
vector operator, 50
connection with rotations, 3
Wigner-Eckart theu,'"",, Iii L
crossing matrices, 168
definition of, 3
5U(n), Casimir opera!.,,,,. f,,, , 'II
Structure constants, 4<)
phase convention, 13
5U(2), 10
reduced matrix elements, 20
5 U(3), 30-31, 50
regular representation, 9
table of, 31
structure constants, 10
subgroups of 5 U(3), 33
5U(3), algebra of, 35 Tensor analysis, 108- 1111
Casimir operators, 72-76 Tensor operator, 104
Clebsch-Gordan coefficients, 99-102 effective form, 114
Clebsch-Gordan series, 103-104 Time reversal symmctry Hf .""1' II
commutation rules for infinitesimal amplitudes, 169
generators, 29, 30, 32 Trace form of 5U(3) inv",.i,,", III , .
crossing matrices, 176-180 actions, 123-125
decuplet and the P 312 resonances, 83, Transition mass, 7TT}, 156
84 EA, 155
defining representation 3, 25-26 .p-w, 147
226 SUBJECT INDEX

Triality,47 coupling to pseudoscalar mesons,

T-spin,35 124-125
decay widths, 121, 129
Vector operators, in 5 U(3), 50
Universality of D/ F ratio, 136
V-spin, 35
U-spin,35
conservation by electromagnetism,
139
use in relating reaction amplitudes,
Weight diagrams, 28-29
139-140
reflection symmetry of, 39-42
Weight vectors, 28
Vector meson, coupling to baryons, Wigner-Eckart theorem, for 5U(2), 19
120-122 for 5U(3), 104