This action might not be possible to undo. Are you sure you want to continue?
GROUP TH€ORY F R PHYSICISTS O
This page intentionally left blank
PROBLEMS & SOLUTIONS IN
GROUP THEORY F R PHYSICISTS O
ZhongQi MQ XicroYan Gu
Institute o High Energy Physics f
China
N E W JERSEY

vp World Scientific
L O N D O N * SINGAPORE BElJlNG * SHANGHAI

HONG KONG
TAIPEI
*
CHENNAI
Published by World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA oflce: Suite 202, 1060 Main Street, River Edge, NJ 07661
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
Library of Congress CataloginginPublication Data Ma, Zhongqi, 1940 Problems and solutions in group theory for physicists / by Z.Q. Ma and X.Y. Gu. p. cm. Includes bibliographical references and index. ISBN 98 1238832X (alk. paper)  ISBN 98 12388338 (pbk.: alk. paper) 1. Group theory. 2. Mathematical physics. I. Gu, X.Y. (XiaoYan). 11. Title.
QC20.7.G76 M3 2004 530.15’22d~22
2004041980
British Library CataloguinginPublicationData A catalogue record for this book is available from the British Library.
Copyright 0 2004 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereox may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any informution storage und retrievul system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.
Printed in Singapore by World Scientific Printers (S)Pte Ltd
Preface
Group theory is a powerful tool for studying the symmetry of a physical system, especially the symmetry of a quantum system. Since the exact solution of the dynamic equation in the quantum theory is generally difficult to obtain, one has to find other methods to analyze the property of the system. Group theory provides an effective method by analyzing symmetry of the system to obtain some precise information of the system verifiable with observations. Now, Group Theory is a required course for graduate students major in physics and theoretical chemistry. The course of Group Theory for the students major in physics is very different from the same course for those major in mathematics. A graduate student in physics needs to know the theoretical framework of group theory and more importantly to master the techniques in application of group theory to various fields of physics, which is actually his main objective for taking the class. However, no course or textbook on group theory can be expected to include all explicit solutions to every problem of group theory in his research field of physics. A student of physics has to know the fundamental theory of group theory, otherwise he may not be able to apply the techniques creatively. On the other hand, physics students are not expected to completely grasp all the mathematics behind group theory due to the breadth of the knowledge required. One of the authors (Ma) first taught the group theory course in 1962. Since 1986, he has been teaching Group Theory to graduate students mainly major in physics a t the Graduate School of Chinese Academy of Sciences. In addition, most of his research work has been related to applications of group theory to physics. In 1996 the Chinese Academy of Sciences decided to publish a series of textbooks for graduate students. He was invited to write a textbook on group theory for the series. In his book, based on his
v
vi
Problems and Solutions in Group Theory
experience in teaching and research work, he explained the fundamental concepts and techniques of group theory progressively and systematically using the language familiar to physicists, and also emphasized the ways with which group theory is applied to physics. The textbook (in Chinese) has been widely used for Group Theory classes in China since it was published by Science Press in Beijing six years ago. He is honored and flattered by the tremendous reception the book has received. By the request of the readers, an exercise book on group theory by the same author was published in 2002 by Science Press to form a complete set of textbooks on group theory. In order to make the exercise book selfcontained, a brief review of the main concepts and techniques is given before the problems in each section. The reviews can be used as a concise textbook on group theory. The present book is the new edition of that book. A great deal of new materials drawn from teaching and research experiences since the publication of the previous edition are included. The reviews of each chapter has been extensively revised. Last four chapters are essentially new. This book consists of ten chapters. Chapter 1 is a short review on linear algebra. The reader is required not only to be familiar with its basic concepts but also to master its applications, especially the similarity transformation method. In Chapter 2, the concepts of a group and its subsets are studied through examples of some finite groups, where the importance of the multiplication table of a finite group is emphasized. Readers should pay special attention to Problem 17 of Chapter 2 and Problem 14 of Chapter 3 which demonstrate a systematic method for analyzing a finite group. The theory of representations of a group is studied in Chapter 3. The transformation operator PR for the scalar functions bridges the gap between the representation theory and the physical application. The subduced and induced representations of groups are used to construct the character tables of finite groups in Chapter 3, and to study the outer product of representations of the permutation groups in Chapter 6. The symmetry groups T for a tetrahedron, 0 for a cube and I for an icosahedron are studied in Chapters 3 and 4. The ClebschGordan series and coefficients are introduced in Chapter 3 and are calculated for various situations in the subsequent chapters. The calculated results of the ClebschGordan coefficients and the ClebschGordan series for the group I and for the permutation groups listed in Problem 27 of Chapter 3 and Problem 31 of Chapter 6, due to its complexity, are only for reference. The classification and representations of semisimple Lie algebras are
Preface
vi i
introduced in Chapter 7 and partly in Chapter 4 by the language familiar
to physicists. The methods of block weight diagrams and dominant weight diagrams are recommended for calculating the representation matrices of the generators and the ClebschGordan series in a simple Lie algebra. The readers who are interested in the strict mathematical definitions and proofs in the theory of semisimple Lie algebras are recommended to read the more mathematically oriented books, e.g. [Bourbaki (1989)l. The remaining part of the book is devoted to the properties of some important symmetry groups of physical systems. In Chapter 4 the symmetry group SO(3) of a spherically symmetric system in three dimensions is studied. The unitary representations with infinite dimensions of the noncompact group are discussed with the simplest example SO(2,l) in Problem 25 of Chapter 4. In Chapter 5 we introduce the symmetry of the crystals. More attention should be paid to the analysis method for the symmetry of a crystal from its International Notation (Problem 6). The commonly used matrix groups are studied in the last three chapters, while the Lorentz group is briefly discussed in Chapter 9. The systematic examination of Young operators is an important characteristic of this book. We calculate the characters, the representation matrices, and the outer product of the irreducible representations of the permutation groups using the Young operators. The method of block weight diagrams can only symbolically give the basis states in the representation space of a Lie algebra. However, for the matrix groups SU(N), S O ( N ) and Sp(24, which are related to four classical Lie algebras, the basis states can be explicitly calculated using the Young operators. The relationship between two methods for the irreducible representations of the classical Lie algebras is demonstrated in the last three chapters in detail. The dimensions of the irreducible representations of the permutation groups, the SU(N) groups, the S O ( N ) groups, and the Sp(2l) groups are all calculated with the hook rule, a method based on the Young diagram. In summary, this book is written mainly for physics students and young physicists. Great, care has been taken to make the book as selfcontained as possible. However, this book reflects mainly the experiences of the authors. We sincerely welcome any suggestions and comments from the readers. This book was supported by the National Natural Science Foundation of China.
Institute of High Energy Physics Beijing, China December, 2003
ZhongQi Ma XiaoYan Gu
This page intentionally left blank
Contents
Preface
V
1. REVIEW ON LINEAR ALGEBRAS 1.1 Eigenvalues and Eigenvectors of a Matrix . . . . . . . . . . 1.2 Some Special Matrices . . . . . . . . . . . . . . . . . . . . . 1.3 Similarity Transformation . . . . . . . . . . . . . . . . . . . 2 . GROUP AND ITS SUBSETS 2.1 Definition of a Group . . . . . . . . . . . . . . . . . . . . . . 2.2 Subsets in a Group . . . . . . . . . . . . . . . . . . . . . . . 2.3 Homomorphism of Groups . . . . . . . . . . . . . . . . . . .
1 1 4 7 27 27 29 33 43
3 . THEORY OF REPRESENTATIONS 3.1 Transformation Operators for a Scalar.Function 3.2 Inequivalent and Irreducible Representations . 3.3 Subduced and Induced Representations . . . . 3.4 The ClebschGordan Coefficients . . . . . . . .
. . . . . . . 43 . . . . . . . 47 . . . . . . . 65 . . . . . . . 79
115 115 123 140 146 166
4 . THREEDIMENSIONAL ROTATION GROUP 4.1 SO(3) Group and Its Covering Group SU(2) . . . . . . . . . 4.2 Inequivalent and Irreducible Representations . . . . . . . . 4.3 Lie Groups and Lie Theorems . . . . . . . . . . . . . . . . . 4.4 Irreducible Tensor Operators . . . . . . . . . . . . . . . . . 4.5 Unitary Representations with Infinite Dimensions . . . . . .
5. SYMMETRY OF CRYSTALS 173 5.1 Symmetric Operations and Space Groups . . . . . . . . . . 173
ix
X
Problems and Solutions in Group Theory
5.2 Symmetric Elements . . . . . . . . . . . . . . . . . . . . . . 177 5.3 International Notations for Space Groups . . . . . . . . . . 186 193 6 . PERMUTATION GROUPS 6.1 Multiplication of Permutations . . . . . . . . . . . . . . . . 193 6.2 Young Patterns, Young Tableaux and Young Operators . . 197 6.3 Primitive Idempotents in the Group Algebra . . . . . . . . 205 6.4 Irreducible Representations and Characters . . . . . . . . . 211 6.5 The Inner and Outer Products of Representations . . . . . 237 7. LIE 7.1 7.2 7.3 269 GROUPS AND LIE ALGEBRAS Classification of Semisimple Lie Algebras . . . . . . . . . . 269 Irreducible Representations and the Chevalley Bases . . . . 279 Reduction of the Direct Product of Representations . . . . 299 317 317 321 336 362 369
375
8. UNITARY GROUPS 8.1 The SU(N) Group and Its Lie Algebra . . . . . . . . . . . . 8.2 Irreducible Tensor Representations of SU(N) . . . . . . . . 8.3 Orthonormal Bases for Irreducible Representations . . . . . 8.4 Subduced Representations . . . . . . . . . . . . . . . . . . . 8.5 Casimir Invariants of SU(N) . . . . . . . . . . . . . . . .
.
9 . REAL ORTHOGONAL GROUPS 9.1 Tensor Represent ations of SO ( N ) . . . . . . . . . . . . . . . 9.2 Spinor Representations of S O ( N ) . . . . . . . . . . . . . . . 9.3 SO(4) Group and the Lorentz Group . . . . . . . . . . . . .
375 403 415
433 10. THE SYMPLECTIC GROUPS 10.1 The Groups Sp(2l. R ) and USp(2l) . . . . . . . . . . . . . . 433 10.2 Irreducible Representations of Sp(2t) . . . . . . . . . . . . . 440 Bibliography 457 461
Index
PROBKMS &I SOLUTIONS IN
GROUP THEORY F R PHYSICISTS O
This page intentionally left blank
Chapter 1
REVIEW O N LINEAR ALGEBRAS
1.1
Eigenvalues and Eigenvectors of a Matrix
*
The eigenequation of a matrix R is
Ra = Xa,
(1.1)
where X is the eigenvalue and a is the eigenvector for the eigenvalue. An eigenvector is determined up to a constant factor. A null vector is a trivial eigenvector of any matrix. We only discuss nontrivial eigenvectors. Equation (1.1) is a set of linear homogeneous equations with respect to the components ap. The necessary and sufficient condition for the existence of a nontrivial solution to Eq. (1.1) is the vanishing of the coefficient determinant :
=
+ (X)”’TrR +
,
. . + det R = 0,
where rn is the dimension of the matrix R, TrR is its trace (the sum of the diagonal elements), and detR denotes its determinant. Equation (1.2) is called the secular equation of R. Evidently, this is an algebraic equation of order m with respect to A, and there are rn complex roots including multiple roots. Each root is an eigenvalue of the matrix. For a given eigenvalue A, there is at least one eigenvector a obtained from solving Eq. (1.1). However, for a root X with multiplicity n, it is not certain to obtain n linearly independent eigenvectors from Eq. (1.1).
1
2
Problems and Solutions in Group Theory
The rank of a matrix R is said to be r if only r vectors among rn rowvectors (or columnvectors) of R are linearly independent. If the rank of ( R  X l ) is r , then (rn  r ) linearly independent eigenvectors can be obtained from Eq. (1.1). Any linear combination of eigenvectors for a given eigenvalue is still an eigenvector for this eigenvalue. For a lowerdimensional matrix or for a sparse matrix (with many zero matrix entries), its eigenvalues and eigenvectors can be obtained by guess from experience or from some known results. So long as the eigenvalue and the eigenvector satisfy the eigenequation (l.l), they are the correct results. On the other hand, even if the results are calculated, they should also be checked whether Eq. (1.1) is satisfied.
A matrix R is said to be hermitian if Rt = R. R is said to be unitary if Rt = Rl. A real and hermitian matrix is a real symmetric matrix. A real and unitary matrix is a real orthogonal matrix. R is said to be positive definite if its eigenvalues are all positive. R is said to be positive semidefinite if its eigenvalues are nonnegative. A negative definite or negative semidefinite matrix can be defined similarly.
1. Prove that the sum of the eigenvalues of a matrix is equal to the trace of the matrix, and the product of eigenvalues is equal to the determinant of the matrix. Solution. The secular equation of an mdimensional matrix R is an algebraic equation of order rn. Its rn complex roots, including multiple roots, are the eigenvalues of the matrix R. So, the secular equation can also be expressed in the following way:
*
det(R X l ) =
n
m
(Xj A)
m
m
j=l
j=1
j=1
= 0.
Comparing it with Eq. (1.2), we obtain that the sum of the eigenvalues is the trace of the matrix and the product of the eigenvalues is equal to the determinant of the matrix.
Review on Linear Algebras
3
2. Calculate the eigenvalues and eigenvectors of the Pauli matrices
u : 2
01
and
a,=(;
;),
0 2 = ( i0
i ) . 0
Solution. The action of 01 on a vector is to interchange two components of the vector. A vector is an eigenvector of 01 for the eigenvalue 1 if its two components are equal. A vector is an eigenvector of 01 for the eigenvalue 1 if its two components are different by sign.
.1(:)=(:>.
Q(!l)=

(1').
( i ) . '
Similarly, if two components of a vector are different by a factor fi,then it is an eigenvector of 0 2 :
.2(])=(]),
",( i'>=

Sometimes a matrix R may contain a submatrix 01 or 0 2 in the form of direct sum or direct product. Thus, some eigenvalues and eigenvectors of R can be obtained in terms of the above results. A matrix R is said to contain a twodimensional submatrix in the form of direct sum if for given a and b, R,, = R,, = Rbc = Rcb = 0, where c is not equal to a and b.
3. Calculate the eigenvalues and eigenvectors of the matrix R
Solution. R can be regarded as the direct sum of two submatrices 01, one lies in the first and fourth rows (columns), the other in the second and third rows (columns). F'rom the result of Problem 2, two eigenvalues of R are 1, the remaining two are 1. The relative eigenvectors are as follows.
1:
(i), (i).
l:
(i),
1
(;l).
4
Problems and Solutions an Group Theory
4. Calculate the eigenvalues and eigenvectors of the matrix R
Solution. This is the generalization of 01. The action of R on a vector is to make a cycle on the three components of the vector in order. R3 = 1, so the eigenvalues of the matrix R is 1, w and w 2 , where w = exp{i2~/3}. The ratio of two adjacent components is the eigenvalue. The eigenvector for each eigenvalue is:
# 0, prove that both R t R and RRt are positive definite hermitian matrices. Solution. It is obvious that both R t R and RRt are hermitian. Letting X and a be the eigenvalue and the eigenvector of RtR, where the eigenvector is written in the form of a column matrix:
5. If det R
(RtR)g = Xa,
we have A
at, > 0,
{ata} = a t ( R t R ) a = ( R G ) (Ra) 2 0, ~
namely, X 2 0. Since detR # 0, det(RtR) # 0 and X # 0. Thus, (RtR) is positive definite. Since detRt # 0, one can similarly prove that RRt is positive definite.
1.2
Some Special Matrices
The inner product of two column matrices
*
a and b is defined as
Two column matrices are called orthogonal to each other if their inner product is vanishing. Q~CJmust be a nonnegative real number, called the
Review on Linear Algebras
5
square of the module of a. at, = 0 only when g = 0.
R is called a unitary matrix if RtR = 1. The column (or row) matrices of R are normalized and orthogonal to each other:
D D
*
P
P
A unitary transformation R does not change the inner product of any
two vectors:
t (Rg) (Rb)= gtRtRb = gtb.
(1.5)
The module of any eigenvalue of a unitary matrix is one, and its two eigenvectors for different eigenvalues are orthogonal to each other. The module of the determinant of a unitary matrix is one, I det R = 1. Letting Ra = A I g and R = d , have b we
The product of two unitary matrices is a unitary matrix. There are rn linearly independent orthonormal eigenvectors for an rndimensional unitary matrix. A real unitary matrix R is called a real orthogonal matrix. The product of two real orthogonal matrices is a real orthogonal matrix. The complex conjugate vector a* of any eigenvector g of a real orthogonal matrix R for the eigenvalue X is an eigenvector of R for the eigenvalue A*. The module of the determinant of an orthogonal matrix R is det R = f l .
R is called a hermitian matrix if Rt = R, or REu = Rup. A hermitian matrix R satisfies
*
Any eigenvalue of a hermitian matrix R is real, and two eigenvectors of R for different eigenvalues are orthogonal to each other. The determinant of a hermitian matrix is real. Letting Rg = X and Rb = rb, we have a
6
Problems and Solutions in Group Theory
The sum of two hermitian matrices is a hermitian matrix. There are rn linearly independent orthonormal eigenvectors for an rndimensional hermitian matrix. A real hermitian matrix R is called the real symmetric matrix. The sum of two real symmetric matrices is a real symmetric matrix. There are rn linearly independent or thonormal real eigenvectors for an rndimensional real symmetric matrix. 6. Prove: (I) if RtR = 1, then RRt = 1; (2) if RlR = 1, then RRl = 1;
(3) if RTR = 1, then RRT = 1.
Solution. The method for proving these three equalities are similar. We prove the first equality as example. Since RtR = 1, Rt are nonsingular. Let S be the inverse matrix of Rt, SRt = 1. So RRt = (SRt)RRt = S (RtR)Rt = SRt = 1. Replacing Rt with R' or RT, we can prove the remaining two equalities.
7. Find the independent real parameters in a 2 x 2 unitary matrix, a real
orthogonal matrix and a hermitian matrix, and give their general expressions.
Solution. A 2 x 2 complex matrix contains four complex parameters, i.e., eight real parameters. For a unitary matrix, the normalization conditions of two columns give two real constraints for the parameters, and the orthogonal condition of two column matrices gives a complex constraint, i.e., two real constraints. Altogether, there are four real constraints. A 2 x 2 unitary matrix contains four independent real parameters. For a hermitian matrix, its diagonal elements are real, that is, their imaginary parts are zero. It leads to two real constraints. Two nondiagonal matrix entries are complex conjugate to each other. It leads to a complex constraint. Thus, a 2 x 2 hermitian matrix also contains four independent real parameters. A 2 x 2 real matrix contains four real parameters. For a real orthogonal matrix, the normalization conditions of two columns give two real constraints for the parameters. The orthogonal condition of two column matrices gives a real constraint. There are three real constraints altogether. Thus, a 2 x 2 real orthogonal matrix contains only one independent real parameter . Letting u be an arbitrary 2 x 2 unitary matrix, we have detu = exp{icp}. Attracting the factor exp{icp/2}, we obtain
Review on Linear Algebras
7
From the definition,
aa* + bb* = cc* + dd* = 1,
we have
ac*
+ bd* = 0 ,
ad  bc = 1,
a=a(cc*+dd*) =d*(bc+ad) =d*, b = b(cc* dd*) = c* (bc  ad) = c*.
+
So, the general expression for a 2 x 2 unitary matrix is
U = eip/2
( )*!;
aa* + bb* = 1,
where two complex parameters a and b with a constraint contain three real parameters. Together with cp, there are four independent real parameters. A real orthogonal matrix is also a unitary matrix. Its determinant may be fl. When its determinant is 1, cp = 0 and both a and b are real numbers, satisfying a2 + b2 = 1. Usually, we take a = cose and b = sine. When the determinant is 1, one may change the sign of the matrix elements in the first row. Thus, we obtain the general expression of a 2 x 2 real orthogonal matrix
R=
(cos8 sin8 sine
C O S ~
, or R' =
cos8 sin8
sin8 cod
The general expression of a 2 x 2 hermitian matrix is
where a, b, c and d all are independent real parameters.
1.3
Similarity Transformation
Ir In an mdimensional space chosen bases e,,
L, any vector a can be expanded in the
a=
C e,aiL.
,=l
m
8
Problems and Solutions an Group Theory
a, is called the component of the vector a in the basis e,. Arrange these components in one column matrix a with rn rows, which is called the column matrix form of a vector a with respect to the bases e,,
An rndimensional space L is said to be invariant for an operator R if any vector in C is transformed by R to a vector belonging to C. Specially, the basis e, is transformed by R to a linear combination of the bases,
(1.10)
*
One may arrange the combination coefficients D,,(R) as an r x rn matrix n D ( R ) which is called the matrix form of an operator R in the bases e,. Choose a new set of the bases e i . Their column matrix forms S., in the original bases e , can be arranged as a matrix S:
ei =
*
C e,s,,.
,=I
m
(1.11)
This matrix S is called the transformation matrix between two sets of bases. S is an rndimensional nonsingular matrix. In the new bases e:, the matrix forms of a vector a and an operator R are
m
,=l
m
(1.12)
p= 1
The relations between two matrix forms in two sets of bases are
 = S'a, a'

D ( R ) = SlD(R)S.
(1.13)
The matrix D ( R ) is transformed into the matrix D(R)through the similarity transformation S. The matrices related by a similarity transformation have the same eigenvalues. They are the matrix forms of one operator in two sets of bases. It is also said in literature that B(R)is equivalent to
D(R).
Write the similarity transformation in the form of matrix entries:
m
m
m
S., is the column matrix form of the new basis ei in the original bases, while B,,(R) in Eq. (1.14) plays the role of the combination coefficients.
Review o n Linear Algebras
9
So Eq. (1.14) is the matrix form of Eq. (1.12) in the original bases e,.
This form is very useful in later calculation.
If the first column of S is an eigenvector for X I of the matrix D ( R ) , then all the components in the first column of D(R) are vanishing except for the first one, which is equal to XI. If all columns of S are the linearly independent eigenvectors of D ( R ) , then D(R) is a diagonal matrix where the diagonal elements are the eigenvalues of the corresponding eigenvectors. This is the essence of diagonalization of a matrix through a similarity transformation. The necessary and sufficient condition for diagonalization of an mdimensional matrix D ( R ) is that there must exist m linearly independent eigenvectors for D ( R ) . Each column of the similarity transformation matrix S which diagonalizes D ( R ) is the eigenvector of D ( R ) . Such a matrix S is not unique because it may be rightmultiplied by a matrix X which commutes with the diagonal matrix D(R). If the m eigenvalues of D ( R ) are different from each other, then X is a diagonal matrix. This means that each eigenvector may include an arbitrary constant factor. If the multiplicity of an eigenvalue X j of D ( R ) is nj, C jnj = m, then X is a block matrix, n; parameters. These parameters reflect the arbitrary linear containing combination between eigenvectors for the same eigenvalue.
*
* A unitary matrix or a hermitian matrix can be diagonalized by a unitary similarity transformation. A real symmetric matrix can be diagonalized by a real orthogonal similarity transformation. A real orthogonal matrix can
be diagonalized by a unitary similarity transformation, but generally not by a real orthogonal similarity transformation, A unitary matrix remains unitary in any unitary similarity transformation. A hermitian matrix remains hermitian in any unitary similarity transformation. A real symmetric matrix remains real symmetric in any real orthogonal similarity transformation.
8 . Find the similarity transformation to diagonalize the following matrices:
Solution. We will denote the given matrix by D ( R ) .
10
Problems and Solutions in Group Theory
(1) The secular equation of D ( R ) is A3
+ 2X2  4X + 8 = (2  X)(X2 + 4) = 0.
We obtain the eigenvalues 2, 2i and 2i. Noticing that the matrix D ( R ) / 2 is a real orthogonal matrix, so its eigenvectors are orthogonal to each other. If the second component of an eigenvector is zero, we see from the second row of D ( R ) that the first component of the eigenvector has to be equal to its third component. Then, from the first or third row of D ( R ) we know that (1,0, l)Tis the eigenvector for the eigenvalue 2. Due to orthogonality, the remaining eigenvectors take the form of (1,a, l)T. Substituting it into the eigenequation
we obtain a = r i a . After normalization, we have
(2) The secular equation of D ( R ) is
The eigenvalues are exp(M ) . Substituting them into the eigenequation,
we obtain a = ri. After normalization, we have
9. Find a similarity transformation matrix M which satisfies
0
Ml(
 cos 8
case
0
sin 8 sin cp sinecosy) 0
M=
 sin 8 sin cp sin 8 cos cp
(!
0
1 0 0 0)
I
Review on Linear Algebras
11
Solution. Denote by D ( R ) and D ( R ) the matrices before and after the similarity transformation M ,respectively. D(R) is the direct sum of ia2 and 0, so it is commutable with matrix X:
If M satisfies M  l D ( R ) M = D ( R ) , M X also satisfies this equation.
Namely, the similarity transformation M is determined up to the three parameters . From a(R) see that the eigenvalues of D ( R ) are 0 and f i . The third we column of M is the eigenvector for the zero eigenvalue, and the first two columns are the linear combinations of eigenvectors for the eigenvalues f i . We first calculate the eigenvector for the zero eigenvalue. Take the third component of the eigenvector to be cos 8, which is a choice for the parameter y. From the eigenequation, the first two components of the eigenvectors are calculated to be sin B cos cp and sin 8 sin cp, respectively. Let
M =
Substituting M into
(
al a2 sin8coscp bl b 2 sinesincp c1 c2
case
)
.
we obtain
= bl cos 8 c1 sin 8 sin cp, = a1 cos8  c1 sinOcoscp, c2 = a1 sin 8 sin cp bl sin 8 cos cp, a1 = 4 2 cos 8 c2 sin 8 sin cp, bl = a2 cos 8  c2 sin 0 cos cp, c1 = a2 sin 8 sin cp b2 sin 8 cos cp.
a2
b2
+
+
+
+
Substituting the first three equations into the last three, or vice versa, we have
a1
sinecoscp
+ bl sin8sincp + c1 cos8 = 0,
+ bz sinosincp + c2 c o d = 0.
a2 sin8coscp
12
Problems and Solutions an Group Theory
These two relations reflect the fact that the first two column vectors of M are orthogonal to the third one. The reason is that D ( R ) is antisymmetric. In fact, denoting by b the eigenvector for the zero eigenvalue and D(R)a = X # 0, we have b T D ( R )= 0, and a
0 = bTD(R)a= XbTa.
Now, choosing c1 =  sine, we obtain a solution from the orthogonality: a1 = cos8coscp and bl = cos8sincp. It is a choice for the parameters (u and p. Substituting the results into the preceding formulas, we have a =  sincp, bz = coscp and c2 = 0. Finally, we obtain the real orthogonal 2 matrix M : cos 8 cos cp cos8sincp  sin8
 sin cp sin 8 cos cp coscp sin8sincp 0 cos e
10. Find a similarity transformation matrix M which satisfies the following three equations simultaneously
1 0 0
0
0
0
0 0 1
M  l ( : 0 0 ;0i ) i M = A y
0 O i
(i y A),
0 1 0
1
0 1
0
i 0 0
Solution. The first equation is the diagonalization of a matrix by a similarity transformation M , whose three column matrices are eigenvectors of the matrix. In order to satisfy the last two equations, we have to keep the multiplied factors in the eigenvectors temporarily, i.e.,
M=
Choose the common factor of A such that b = 1. Substituting J into 4 d the second equation, where M' is moved to the righthand side of the
(:::)
ia 0 ic
.
Review on Linear Algebras
13
equation to avoid the calculation of M  l , we obtain
(",
0 Thus,
0
;i
8) 5 (;
0
=
0
a+c
i(a;c)
;).
0
Ad=+
1
0 i). 1 4 0
fi
0
After checking, it does satisfy the third equation.
11. Let
Find the common similarity transformation matrix X satisfying
Xl(RxR)X=
(:0' !l),
1 0 0 0
0
0
0
0
oofi
1
Solution. According to the definition of the direct product of matrices,
Since the matrices before and after the similarity transformation X are both real orthogonal, we can also choose X to be real orthogonal. R x R is a diagonal matrix. It is easy to see that its eigenvectors for the eigenvalues
14
Problems and Solutions an Group Theory
1 and 1 respectively are
Thus, X is in the form of
Since the first two columns of X have to be the eigenvectors of S x S for the eigenvalue 1, we obtain a = b and c = d. Because X is a real orthogonal matrix, we have a' = b' and c' = d'. Choosing a = c = c' and calculating the application of S x S to the fourth column of X , we have
Thus, a' = a. After normalization we obtain
x=1
Jz
1 0 10 0 1 0 1 0 1 0 1 1 0 1 0
12. Find the similarity transformation matrix X to diagonalize the following three matrices simultaneously,
'000100 000010 000001 100000 010000 \001000 '000100 000001 000010 100000 001000 ,010000 /000111 000111 000111 111000 111000 \ l l l O O O
Solution. Both the first two matrices are the direct sum of three 01, but different in the rows (columns). In the first matrix, the submatrices relate to the rows (columns) (1,4), (2,5) and (3,6), while in the second matrix
Review on Linear Algebras
15
they relate t o (1,4), (2,6) and (3,5). So, three eigenvalues for each of two matrices are 1, and the other three are 1. The corresponding eigenvectors are as follows: d e
d' e'
f
d
f'
d'
,
f
e
f'
el 1
1
1
The eigenequations for the third matrix are
~ 0 0 0 1 1 1 ' P 0 0 0 1 1 1 0 0 0 1 1 1 1 1 1 0 0 0 S t 1 1 1 0 0 0 \I 1 1 0 0 0 U
:;=I
s+t+u s+t+u s+t+u p+q+r
p+q+r
':I*
S
t
U
Except for X = 0, the first three components must be equal to each other, so do the last three components. Namely, the eigenvalues are f3, and the eigenvectors are 1 1 1 1 1 1
+3:

1
4
7
3:
1 
4
These two vectors are also the eigenvectors of the first two matrices for the eigenvalues 1 and 1, respectively. In the remaining subspace, all the eigenvalues of the third matrix are zero, but the eigenvalues of the first two matrices are f1, respectively. Arranging the six eigenvectors, we obtain
16
Problems and Solutions in Group Theory
the similarity transformation matrix X
which diagonalize all three matrices simultaneously: diag(1,  1, 1, 1,  1,  l}, diag{l,  1, 1,  1, 1,  l}, diag (3,  3, 0, 0, 0, 0 ) .
13. Show the general form of an m x m matrix, both unitary and hermitian.
Solution. Both a unitary matrix and a hermitian matrix can be diagonalized by a unitary similarity transformation. After diagonalization, the module of each diagonal element of the unitary matrix is 1, while each diagonal element of the hermitian matrix is real. So, each diagonal element in the diagonalized matrix, both unitary and hermitian, must be 1 or 1. We denoted by rn the diagonalized matrix where the first n diagonal elements are 1, and the last m  n diagonal elements are 1. In general, a matrix both unitary and hermitian is in the form of UI',Ul, where U is a unitary matrix with determinant one.
14. Prove that any unitary matrix R can be diagonalized by a unitary similarity transformation, and any hermitian matrix R can be diagonalized by a unitary similarity transformation.
Solution. We prove this proposition by induction. Obviously, the proposition holds when the dimension of R is one. Assuming that the proposition holds when the dimension of R is ( m  l), we are going to prove that it holds when the dimension of R is rn. Let A1 be an eigenvalue of a unitary matrix R with dimension m and S(t) be the normalized eigenvector of it. Arbitrarily find a complete set of orthonormal column matrices S!;), where S!:) is the first one in the set. Thus, we obtain a unitary matrix S(l).After the similarity transformation S ( l ) ,(S('))' R S ( l )keeps unitary, where all its matrix entries in the first column are vanishing except for the first one which is X I . Since [ A l l 2 = 1 and (S(l))'R S ( l )is a unitary matrix, all its
Review on Linear Algebras
17
matrix entries in the first row vanish except the first one.
where the submatrix R ( l ) is unitary and (m  1)dimensional. Therefore, R ( l )as well as R can be diagonalized by a unitary similarity transformation. It means that there exist m orthonormal eigenvectors for an mdimensional unitary matrix. If R is a hermitian matrix, we introduce the similarity transformation S ( l ) in the same way. After the similarity transformation S(l), 1 ($I))R S ( l ) keeps hermitian, where all its matrix entries in the first column are vanishing except for the first one which is A1. Since ( S ( l ) ) R S ( l ) is hermitian, all its matrix entries in the first row vanish except the first one. The remaining proof is the same as that for a unitary matrix. It means that there exist m orthonormal eigenvectors for an mdimensional hermitian matrix.
'
15. Prove that R and Rt can be diagonalized by a common unitary similarity transformation if Rt is commutable with R. Further prove that the necessary and sufficient condition for a matrix R which can be diagonalized by a unitary similarity transformation is that Rt is commutable with R. Solution. The first part of this problem is similar to the proof given in the preceding problem. We prove the first part of the present problem by induction. Obviously, it holds when the dimension of R is one. Assuming that it holds when the dimension of R is (m  1), we are going to prove that it holds when the dimension of R is m. We introduce the similarity transformation S ( l ) in the same wa . After the similarity transformation 7 S ( l ) ,(S('))' Bt (S(l)) ( S ( l ) ) R S ( l )are still conjugate to each other and and commutable with each other. Because
18
Problems and Solutions in Group Theory
we have
Thus, all rn vanish. The submatrix R(l) is still commutable with ( I t ( ' ) ) + and is ( m  1)dimensional. Therefore, R and Rt can be diagonalized by a unitary similarity transformat ion simultaneously. Conversely, if R can be diagonalized by a unitary similarity transformation S, then SlRtS = ( S  l R S )t is also diagonal. Since two diagonal matrices are commutable, R and Rt are also commutable.
16. Prove that any matrix can be transformed into a direct sum of the standard Jordan forms, each of which is in the form
or 1
when a = b when a + 1 = b the remaining cases.
Solution. As is well known, for each eigenvalue, there exists at least one corresponding eigenvector. But when the eigenvalue is multiple, it is not certain if the same number of linearly independent eigenvectors as the multiplicity of the eigenvalue can be found. If there are m linear independent eigenvectors of an rndimensional matrix R, the similarity transformation matrix X by arranging the eigenvectors as its column matrices can diagonalize R. Now, we only need to deal with the case where the number of linearly independent eigenvectors is less than, at least for one eigenvalue, the multiplicity of the eigenvalue. We will divide this problem into three propositions. First, any matrix R can be transformed to an upper triangular matrix, where the diagonal elements are the eigenvalues of R and the same eigenvalues are arranged together. Second, by a further similarity transformation, R can be transformed into a block matrix, where each submatrix relates to only one eigenvalue. In other words, any nondiagonal matrix entry related to two different eigenvalues is vanishing. Third, each submatrix related to a given eigenvalue can be transformed into the standard Jordan form by a similarity transformation. We will prove the first proposition by induction. The proposition obviously holds for a onedimensional matrix. If the proposition holds for any ndimensional matrix where n < m. We are going to show that the proposition also holds for any rndimensional matrix R.
Review on Linear Algebras
19
Let R have d different eigenvalues X j with multiplicity n j , respectively. For X I , say, we can find one eigenvector S.1. Arbitrary choose a complete set of vector bases S., where S.l is the first one in the set. Arrange them into an rndimensional nonsingular matrix S. After the similarity transformation S, R becomes R' = SlRS, where all matrix entries in the first column are vanishing except for the first one which is X I . Removing the first row and the first column in the matrix R', we obtain an (rn  1)dimensional submatrix A of R'. In comparison with the set of eigenvalues of R, the multiplicity of A1 in the set of eigenvalues of A decreases by one. Being an ( r n  1)dimensional matrix, as we supposed, A can be transformed into an upper triangular matrix A' = X  l A X by an ( r n  1)dimensional similarity transformation X . On the diagonal line the matrix entries of A' are arranged in the order: X I , . .., X I , Xz, ..., Xa, . .., Ad. Let an rndimensional matrix T be the direct sum of the digit one and the (rn  1)dimensional matrix X. Then, R" = TlR'T = ( S T )  l R ( S T ) satisfies the first proposition: R" is an upper triangular matrix, where the diagonal elements are the eigenvalues of R and the same eigenvalues are arranged toget her. For convenience, we remove the double primes on R". Its matrix entry is denoted by Rja,kb, where the row (column) subscript is designated by two indices ( j a ) , 1 _< j _< d, 1 5 a 5 nj: when j > k when j = k and a > b when j = k and a = b the remaining cases.
Rja,kb = Rja,kb
The second proposition says that there exists a similarity transformation X such that after the transformation all Ria,$,with j # k become vanishing. Let S(ab) a matrix, where all the diagonal matrix entries are one, and be the nondiagonal matrix entries are vanishing except for sj::ib = w , where j < k. Denote by T ( a b the inverse of S(ab) ) for convenience. In fact, the difference between S(ab) and T ( a bis that w is replaced with w. By ) this similarity transformation, R' = T ( a b ) RS("'),only the following matrix entries are changed:
20
Problems and Solutions in Group Theory
when k
Rjlal,kb
< k or when k = k and ' '
b
< b',
 Rj'aJ,kb RjJa',jaSja,kb R j ' a J , k b k wRjlaJ,ja, k (ab) =
when j'
<j
or when j ' = j and a' < a .
Choosing
we have R i a , k b = 0. Now, we annihilate the nondiagonal matrix entries R j a , k b with j < k by a series of the above similarity transformations S(ab) one by one in the following order. First, beginning with j = 1, k = 2 and b = 1, we annihilate Rla,21 one by one in the decreasing order of a from n1 to 1. Second, increasing 13 by one from 1 to nk, we repeat the above process to annihilate R l a , 2 b in the decreasing order of a. Third, increasing k by one from 2 to d, for a given k we decrease j by one from k  1 to 1, and for the given j and k we repeat the above two processes to annihilate R j a , k b . Thus, after a series of similarity transformations, R is transformed into a block matrix, where each submatrix is an upper triangular matrix, namely, when when when when
j#k j = k and a > b j = k and a = b j = k and a < b.
Rja,kb
=
Rja,jb
The third proposition says that each submatrix related to a given eigenvalue can be transformed into the standard Jordan form by a similarity transformation. Let a submatrix, denoted still by R for convenience, be ndimensional and have the same eigenvalue A. Assume that the rank of the matrix ( R  X l ) is ( n  t ) , namely, there are t linearly independent eigenvectors of R for the eigenvalue A. According to the first proposition, we can find a similarity transformation matrix X , where the first t columns are the eigenvectors of R, such that after the similarity transformation X , R becomes an upper triangle matrix, where the diagonal matrix entries are X and the nondiagonal matrix entries in the first t columns are vanishing. Thus, the problem becomes how to transform an ndimensional echelon matrix R into the standard Jordan form:
Review o n Linear Algebras
21
where A is a t x t constant matrix, A = X1, S is an ( n  t ) x ( n  t ) upper triangular matrix with the diagonal elements A, and T is a t x ( n  t ) matrix. We denote the indices of R by a = 1, 2, ..., n. When c11 takes the first t values, we denote a by the first lowercase Latin letter, say, a = 1 , 2 , . . . , t. When a takes the last (n  t ) values, we denote a by the middle lowercase Latin letter, say, j = (t I), (t 2)) . . ., n.
+
+
Aa,b = XS,b,
Tu,j no limit,
Sj,k
=
X 0
Sj,k
when j = k when j > k when j < k .
(1.15)
In the following we will introduce a series of similarity transformations, each of which keeps the form (1.15) invariant, but simplifies the nondiagonal matrix entries of R. Our aim is to transform the nondiagonal matrix entries of R such that in each row and in each column of R there is at most only one nonzero nondiagonal matrix entry, which is equal to 1. Then, by a simple similarity transformation, which only changes the order of the rows and the columns, R becomes the standard Jordan form. Define a similarity transformation X, where the similarity transformation matrix X ( M ,b) is the direct sum of a bdimensional unit matrix, a (t  b)dimensional nonsingular matrix M' and an ( n  t)dimensional unit matrix, whose action is to transform each column from the (b 1)th row to the tth row of T by M , but to keep A and S invariant. We can choose M t o annihilate all the matrix entries in one column from the (b+ 1 ) t h row to the tth row of T except for one, which is equal to 1. We study the nondiagonal matrix entries in the (t 1) column of T . They are not all vanishing, otherwise R have (t 1) linearly independent eigenvectors. We choose M in the similarity transformation X ( M , 0) to annihilate all matrix entries in the ( t + l ) t h column of T except for Tl,(t+l) = 1. Define a similarity transformation Y , where the similarity transformation matrix Y, ( P I , a1 ; p2, a2; . . .) is a unit matrix plus a few nonzero nondiagonal matrix entries. Those matrix entries are all equal to w and are located in the a1 column of the row, the a2 column of the ,!?2 row, and so on. All ap and Py are different from each other. Its inverse matrix can be obtained from it just by replacing w with w. We can make a similarity transformation Y, (t 1,k) with w = Tl,k to annihilate T 1 , k where k > t + l . From the viewpoint of basis transformation, the original basis is the unit vector e , in the rectangular coordinate system
+
+
+
+
22
Problems and Solutions in Group Theory
and the application of R on the basis is
The similarity transformation Yw(t 1,k ) will transform the kth basis e k + into a new basis e l = e k  Tl,ket+l, which just cancels the term related to el in Rek:
After a series of similarity transformations Y, (t + 1,k ) , w = Tl,k, where k runs from t 2 to n one by one, the matrix entries in the first row of T will all be annihilated except for T1,(t+l)= 1. This method for simplifying the matrix entries in the (t 1)th column and in the first row of T is called the method of the first kind. Now, we investigate the ( t 2)th column of R. There are two cases. The first case is S(t+1),(t+2) 0. In this case, the method of the first = kind can be used to simplify the matrix entries in the (t + 2)th column and in the second row of T . We choose M in the similarity transformation X ( M , 1) to annihilate all matrix entries in the ( t + 2)th column of T except = for T2,(t+2) 1. In the transformation the matrix entries in the ( t + 1)th column and in the first row of T are kept invariant. Then, by a series of similarity transformations Yw(t 2, k ) , where w =  T 2 , k , t + 3 5 k 5 n, we + can annihilate all matrix entries T2,k except for T2,(t+2)= 1. The second case is S(t+l),(t+2) Define a similarity transformation # 0. 2 , where the similarity transformation matrix Zc(cy) is a diagonal matrix with all diagonal matrix entries to be 1 except for the a t h entry to be <. This similarity transformation enlarges the matrix entries in the a t h column of R by C times and reduces the matrix entries in the a t h row by C times, but keeps the diagonal matrix entries invariant. We1first choose the similarity transformation &(t 2) with C = { S(t+l),(t+a)) to change S(t+l),(t+2) 1. Second, we make a series of similarity transformations to be Y w ( a t + 1) with w = Tu,(t+2),5 a 5 t , to annihilate the matrix entries in , 2 the (t 2)th column of T , where the matrix entries in the ( t + 1)th column of T are kept invariant. From the viewpoint of base transformation, the application of R on the original basis is
+
+
+
+
+
t
a=2
Review on Linear Algebras
23
A series of similarity transformations Y w ( a t 1) with LJ = Ta,(t+2), 5 , 2 a 5 t , define the vector in the bracket of the above formula as a new basis et$, to annihilate all Ta,(t+2). Third, we annihilate the matrix entries S(t+l),k, > t+2, in the ( t + l ) t h k row of S by a series of similarity transformations Yw (t 2, k ) with w = S(t+l),k,where k runs from t + 3 to n one by one. The action of this series of similarity transformations is the same as that to annihilate T1,k in the method of the first kind. This method for simplifying the matrix entries in the ( t + 2)th column of T and in the (t 1)th row of S in the second case is called the method of the second kind. In the later application some revision of this method will be made. Based on the above simplification, we are going to prove by induction that R can be transformed into the standard Jordan form by a series of similarity transformations. We define the dsimplified Rmatrix as the R matrix in the form (1.15) satisfying that in each column of R from the ( t + 1)th column to the (t + d)th column there is only one nonzero nondiagonal matrix entry which is 1 and those nonzero nondiagonal matrix entries are all located in different rows. We will prove that the dsimplified Rmatrix can be transformed into the (d + 1)simplified Rmatrix by a series of similarity transformations. We have proved that the lsimplified Rmatrix can be transformed into the 2simplified Rmatrix by a series of similarity transformations. For definiteness, we assume that from the ( t + l ) t h column to the (t+d)th column of a dsimplified Rmatrix, there are g nonzero nondiagonal matrix entries located in the first g rows of T and (d  g) nonzero nondiagonal matrix entries located in S. There are two cases for the positions of the nonzero nondiagonal matrix entries in the (t d + 1)th column of R. In the first case, all nonzero nondiagonal matrix entries in the (t d + 1)th column of R appear in the submatrix T . In the second case, a t least one nonzero nondiagonal matrix entry in the (t d 1)th column of R appears in the submatrix S. We will use the method of the first kind to simplify R in the first case, but the method of the second kind in the second case. In the first case, we choose M in the similarity transformation X ( M ,d ) to annihilate all matrix entries in the (t d 1)th column of T except for T(g+l),(t+d+l)1, where the matrix entries in the first g rows of T are = kept invariant. Then, we annihilate all T(g+llk, where k runs from t + d + 2 to n one by one, by the similarity transformations Yw(t d + 1,k) with + w =  T ( g + l ) k . The result is nothing but the (d + 1)simplified Rmatrix. In the second case, there is a t least one nonzero nondiagonal matrix
+
+
+
+
+
+ +
+ +
24
Problems and Solutions in Group Theory
entry in the ( t d 1)th column of S . We have known that there is only one nonzero nondiagonal matrix entry (which is 1) in each column from the ( t 1)th column to the ( t d)th column of the dsimplified Rmatrix. For a nonzero nondiagonal matrix entry S j , , ( t + d + l ) in the (t d 1)th column of S , if the nonzero nondiagonal matrix entry in the j1 th column of R lies in the a2th row, the nonzero nondiagonal matrix entry in the a2th column lies in the a3th row, and so on; till a, 5 t , then we define the degree of Sjl,(t+d+l) to be u and the degree indices of Sjl, ( t + d + l ) to be (jl,a 2 , a3, . . . , a,). Recall that in the a,th column of R, all nondiagonal matrix entries are vanishing. Among nonzero nondiagonal matrix entries in the ( t + d + 1)th column of S , we choose the one with the highest degree, say Sj, , ( t + d + l ) with the degree u and the degree indices (jl, a 2 , a 3 , . . . , a,). We
+ +
+
+
+ +
first choose a similarity transformation Zc(t+d+l) with C = ( S j l , ( t + d + l ) ) to make Sjl,(t+d+l) 1. This similarity transformation does not change = the positions of the nonzero nondiagonal matrix entries in the ( t d 1)th column of S . The degree v of any other nonzero nondiagonal matrix entry in the ( t d 1)th column of S is not larger than u,say S j , , ( t + d + l ) = a with the degree v and the degree indices ( j 2 , p2, . . . ,p,), v 5 u.In the following we will show that the similarity transformation Y, ( j 2 , j 1 ; p2, a 2 ; . . . ; p,, a,) with LJ= CT can annihilate the matrix entry Sjz,(t+d+l). The reason for this can be understood by the viewpoint of the basis transformation. The new basis induced by this similarity transformation is
1
+ +
+ +
ej,
e,,
+ oej, + aeg,
when r = j1, when r = a 2 ,
Ier
Before this transformation we have
the remaining cases.
Review on Linear Algebras
25
After the similarity transformation we have
When v = u,the terms of e,,,+, and e,lu+lin the last formula disappear. Thus, the action of R on the new basis edp is the same as that on the original basis eap except for that on the basis et'+d+l, where the term of aej, disappears. The action of R on the bases e i , k > t d 1, may also be changed, namely, the nondiagonal matrix entries in the kth column of R may be changed. But we do not care about them. Through a series of similarity transformations, the matrix entries in the (t d 1)th column of S become vanishing except for Sj,,(t+d+l) 1. = If a matrix entry in the ( t d 1)th column of T is nonzero, say Ta(t+d+l) # 0, we can use the similarity transformation Y w ( a , j l )with w = Ta(t+d+l) annihilate Ta(,+dfl). Finally, we choose a series of simito larity transformations Yw(t d + 1,k) with w =  S j l , k , t d + 2 5 k 5 n, + to annihilate all S j l , k in the j l t h row of S . Thus, we have found a series of similarity transformations to change a dsimplified Rmatrix into a (d 1)simplified Rmatrix, namely, we have proved the third proposition. Therefore, we have proved that any matrix R can be transformed into a direct sum of the standard Jordan forms. We will give a typical and instructive example to show how to transform a dsimplified Rmatrix into a (d 1)simplified Rmatrix by a series of similarity transformations. Assume a 7simplified Rmatrix with t = 6, n = 15, g = 4, and d = 7,
+ +
+ +
+ +
+
+
+
'1000000 0 0 0100000 0 0 0000100 0 0 T= 000~00100 0 0 0 0 0 0 0 0 T,4 51 51 T,5 , O 0 0 0 0 0 0 T , 4T , 5 61 61
26
Problems and Solutions in Group Theory
/x 0 0 0 S= 0 0 0
01 0 0x 00 00 00 00
0000 0000 1000 000 0 00 00x1 000
0
s8,14
0
s8,15
'
0
0
s10,14 s10,15 s11,14 s11,15
0
A 0
0
S14,15
s13,14 s13,15
0000000
\ooooooo
x
According to the matrix entries in the 14th column (in general, the (t + d 1)th column) of S, there are two cases. One is the case where the nondiagonal matrix entries in the 14th column of S are all zero. In this case the method of first kind can be used to simplify R. We choose a twodimensional matrix M in the similarity transformation X(M,4) to make T 5 , 1 4 = 1 and T 6 , 1 4 = 0, but to keep the first four rows of T invariant. Then, we use the similarity transformation Yw(14,15) with w =  T 5 , 1 5 t o make T 5 , 1 5 = 0. This has been the 8simplified Rmatrix. The other is the case where the nondiagonal matrix entries in the (t d 1)th column of S are not all zero. In the example we assume that the nondiagonal matrix entries given in S are all nonzero. First to determine the degrees and the degree indices for these matrix entries. s 8 , 1 4 has the degree two and the degree indices (8,2). S 1 0 , 1 4 has the highest degree four and the degree indices (10,9,7,1). S 1 1 , 1 4 has the degree two and the degree indices (11,3). s 1 3 , 1 4 has the degree three and the degree indices (13,12,4). We use the similarity transformation Zc(14), with C = ( s l O , 1 4 )  ' to make s 1 0 , 1 4 = 1. The remaining nondiagonal matrix entries in the 14th column of S have changed their values, but are still denoted by the original symbols. Now, we use the following similarity transformations one by one to obtain the 8simplified Rmatrix:
+
+ +
y, (8,10; 219)
Y w ( l l ,10; 3,9) Yw(13,10; 12,9; 4,7) Yw(5,W
with w = s 8 , 1 4 with w = S 1 1 , 1 4 with w = with w = T 5 , 1 4 with w = T 6 , ~ 4 with w =  S 1 0 , 1 5
transforms transforms transforms transforms transforms transforms
S8,14 S11,14 S13,14 T5,14 T6,14 S10,15
to be zero, t o be zero, to be zero, to be zero, to be zero, to be zero.
yu ( 6 , W
y (14,151 ,
Chapter 2
GROUP AND ITS SUBSETS
2.1
Definition of a Group
A group is a set G of elements R satisfying four axioms with respect to the given multiplication rule of elements. The axioms are: a) The set is closed to this multiplication; b) The multiplication between elements satisfy the associative law; c) There is an identity E E G satisfying E R = R ; d) The set contains the inverse R' of any element R E G satisfying R  l R = E . The multiplication rule of elements completely describes the structure and property of a group. A group G is called a finite group if it contains finite number g of elements, and g is called the order of G. Otherwise, the group is called an infinite group. For a finite group, the multiplication rule can be given by the multiplication table, or called the group table. A group is called the Abelian group if the product of its elements is commutable. A few elements in G are called the generators of G if any element in G can be expressed as their product. The rearrangement theorem says RG = G R = G , namely, there are no duplicate elements in each row and in each column of the multiplication table. Two groups are called isomorphic, G M G', if there is a onetoone correspondence between elements of two groups in such a way products correspond to products. From the viewpoint of group theory, two isomorphic groups are the same as each other. 1. Let E be the identity of a group G , R and S be any two elements in the group G, R' and S' be the inverses of R and S, respectively. Try to show from the definition of a group: (a) RRl = E ; (b) R E = R; (c) if T R = R, then T = E ; (d) if T R = E , then T = R'; (e) The inverse of ( R S ) is Sl R'.
27
*
*
28
Problems and Solutions an Group Theory
Solution. The key to the proof is that each element in a group has its inverse. Recall that only the definition of a group and the proved conclusion can be used in the later proof. (a) We prove the conclusion by two methods. In the first method, since R' is an element in the group, there exists its inverse in the group, denoted by S , SRl = E . So, from the definition of a group we have
Another method is to denote by W' the inverse of RRl = W , WlW = E . Then, we have:
W W = R (RlR) R' = RERl = W, R R  l = W = EW = ( W  l W ) W = W  ' ( W W ) = WlW = E.
(b)
RE = R (RIR) = (RRl) R = ER = R.
(c) If T R = R, then
T = T (RRl) = ( T R )R' = RRl = E.
The conclusion says that the identity in a group is unique. (d) If T R = E , then
T = T E = T ( R R  ~= ( T R )R  ~ E R  ~ R  ~ ) = =
The conclusion says that the inverse of any element in a group is unique. (e) Since (SlR') ( R S )= S' (RlR) S = SlES = SlS = E and the conclusion (d), SlI2l is the inverse of RS. 2. Show that a group composed of all positive real numbers where the multiplication rule of elements is defined with the product of digits, is isomorphic onto a group composed of all real numbers where the multiplication rule of elements is defined with the addition of digits.
Solution. Denote by H the set of all positive real numbers R. The multiplication of elements is defined to be the product of digits. The set is closed for the multiplication. The product of digits satisfies the associative law. The positive real number 1 is the identity in the set. The reciprocal digit 1/R is still a positive real number, which is the inverse of R. Therefore, the set of H is a group, called the multiplication group of positive real numbers.
Group and Its Subsets
29
Denote by G the set of all real numbers S. The multiplication of elements is defined to be the addition of digits. The set is closed for the multiplication. The addition of digits satisfies the associative law. The real number 0 is the identity in the set. S is still a real number, which is the inverse of S. Therefore, the set of G is a group, called the addition group of real numbers. The exponential relation R = e s gives a onetoone correspondence between the positive real number R in H and the real number S in G, and it is invariant for the multiplication of elements:
Therefore, the group H is isomorphic onto the group G.
2.2
Subsets in a Group
subset of a group is called a subgroup of the group if in the multiplication rule of the group elements the subset satisfies four axioms in the group definition. The set composed of the powers of any element in a group forms a cyclic subgroup, also called the period of the element. For a finite group, a subset of a group is its subgroup if and only if the subset is closed with respect to the multiplication rule because the subset contains the period of any element in it. In the same reasoning, a finite set of elements is a group if and only if the set is closed with respect to an associative multiplication rule. It is easy to judge by the multiplication table of the group whether or not a subset is a subgroup. The order of the cyclic subgroup generated by the element is called the order of the element. Only the order of the identity is 1. The identity and the whole group are two trivial subgroups. We will only consider the nontrivial subgroups. subset in a group G obtained by leftmultiplying (or rightmultiplying) an element R E G to a subgroup H of G, where R does not belong to H , is called a left coset RH (or a right coset H R ) of the subgroup H . For a finite group, the number of the elements in a coset is equal to the order of the subgroup. Since a coset does not contain any element in the subgroup, any coset must not be a subgroup due to lack of the identity. Different left cosets (or right cosets) do not contain any common element. The necessary and sufficient condition for two elements R and S belonging to the same left coset is RlS E H , and that for them belonging to the same right coset
*A
*A
30
Problems and Solutions i n Group Theory
is RSl E H . Hence, the order h of a subgroup H must be a whole number divisor of the order g of the group G. The integer n = g/h is called the index of the subgroup H in G. The number of the different left cosets (or right cosets) of a subgroup is ( n  1). In the columns of the multiplication table, related to the elements of a subgroup, the set of elements in each row is the subgroup itself or the left coset of the subgroup. In the rows of the multiplication table, related to the elements of a subgroup, the set of elements in each column is the subgroup itself or the right coset of the subgroup. The element R' = SRS' is said to be an element conjugate to R in a group G, where R, R' and S all belong to G. The conjugate relation of two elements is mutual. If two elements are both conjugate to a third element, they are also conjugate to each other. The subset of all mutually conjugate elements in a group G is called a class C, in G. Any element in G belongs and only belongs to one class. No class forms a subgroup of G except for the class C1 composed of the identity. The set of all the inverse elements R' of R E C, is also a class, denoted by C;'. These two classes are called the reciprocal classes mutually. It is called the selfreciprocal class if C, = C;'. RS is conjugate to SR. Conversely, two conjugate elements can be expressed as two products RS and S R where I? and S are two elements in G. For a finite group, the elements in one class have the same order. However, two elements with the same order are not necessary conjugate to each other. For a multiplication table of a finite group where the arrangement for the rows is the same as that for the columns, two conjugate elements appear and must appear at least once in two symmetrical positions with respect to the diagonal line of the multiplication table. This is the main method to check by the multiplication table whether two elements with the same order are conjugate to each other. * A subgroup H is called an invariant subgroup (or normal subgroup) of the group G if for any element R E G the left coset is equal to the right coset ,
*
RH = HR,
RHRI = H .
A subgroup with index 2 must be an invariant subgroup. An invariant subgroup is composed of a few whole classes and satisfies the criterion for a subgroup. It is the main method to find an invariant subgroup of a finite group G that one first gathers a few classes including the identity to see whether the number of elements in the set is a whole number divisor of the
Group and Its Subsets
31
order of G and whether the set contains the whole period of any element in the set, then, to check whether the set is closed for the multiplication rule of the group elements. The multiplication of two subsets is defined to be a set composed of all the products of any two elements respectively belonging to two subsets. Note that in a set only one element is taken for the duplicated element. Denote the invariant subgroup H of G and its cosets uniformly by R j H , where R1 = E . The aggregate of the subsets R j H satisfies the four axioms with respect to the multiplication rule of subsets. In fact,
*
( R j H )( R k H ) = R j R k H H = ( R j R k )H a
The multiplication rule satisfies the associate law. R1 H = H is the identity in the aggregate. Ryl H is a coset of H and is the inverse of Rj H in the aggregate. Therefore, the aggregate of the subsets R j H forms a group, called the quotient group G / H of the invariant subgroup H .
3. If H1 and HZ are two subgroups of a group G , prove that the common elements in HI and H2 also form a subgroup of G.
Solution. Denote the set of the common elements of the subgroups H1 and H2 by H3 = (R1,R2,  .}, where Rj E H1 and Rj E H2. The set of H3 is also a subset of G , and the product of elements in H3 satisfies the multiplication rule of G . So the associative law is satisfied for H3. Since H1 and H2 are two subgroups of G , the product RiRj of any two elements in H3 must belong to both subgroups H1 and H2 such that it belongs to the set H3. For the same reason, the identity E in G and the inverse Rj' of any element Rj in H3 belong to both subgroups H1 and H2, so that they also belong to the set H3. Therefore, the set H3 is a subgroup of G .
4. Prove that a group whose order g is a prime number must be a cyclic
+
group C,.
Solution. The order of an element is nothing but the order of the cyclic subgroup generated by the element. As we have known, the order h of the subgroup H c G is a whole number divisor of the order g of the group G. When g is a prime number, except for the identity, the order of any element in G must be the order g of the group, so the cyclic subgroup is the group G itself. From it we come to the conclusion that any two groups whose orders are the same prime number must be isomorphic onto each other.
32
Problems and Solutions in Group Theory
5 . Show that up to isomorphism, there are only two different fourthorder
groups: The cyclic group
Cq
and the fourthorder inversion group
V4.
Solution. Since the order of any element in a fourthorder group, except for the identity, has to be 2 or 4. If there is at least one fourthorder element R , the group is the cyclic group Cq = { E , R , R2, R 3 } . If the order of any element in the group, except for the identity, is 2, denote the group by V4 = {e, o, r, p } , where e is the identity and o2 = r2 = p2 = e. From Problem 1, the product of any two elements, say or,cannot be equal to e, o and r , so or = p and the group is the fourthorder inversion group. This completes the proof.
6 . Show that up to isomorphism, there are only two different sixthorder
groups: The cyclic group C6 and the symmetry group D3 for the regular triangle.
Solution. Since the period of any element is a cyclic subgroup, in a sixthorder group, except for the identity, the order of any element has to be 2, 3 or 6. There are three cases. In the first case where there is at least one sixthorder element, the group is a cyclic group CS. In the second case the group contains no sixthorder element, but contains a t least one thirdorder element, denoted by R. The cyclic subgroup H = { E ,R , R2}generated by R has the index two so that it is an invariant subgroup of the group. Denote its coset by HSo = { S O , S ~ , S ; where RmSo = Sm, R"Sj = Sj+m, and ?}, Sj+3 = Sj. S; cannot be equal to R or R 2 , otherwise the order of Sj is 6. From the rearrangement theorem, S; cannot be equal to Sk. Thus, S; = E . It can be deduced that Rm = Sj+mSj and SjRm = Sjm. This group is nothing but D3. In the third case, the order of any element in the group is 2 except for the identity. Arbitrarily take two elements R and S in the group, and let RS = T . From Problem 1 we conclude that T is not equal to E , R or S. The subset composed of E , R, S and T forms a subgroup. It is isomorphic onto the inversion group V4 of order 4. Since 6/4 is not an integer, the third case cannot exist. This completes the proof.
7. Show that a group must be an Abelian group if the order of any element in the group, except for the identity, is 2. Solution. What we need to show is whether RS = S R for any two different elements R and S in the group G. Letting RS = T , we have R2 = S2 = T 2 = E . From Problem 1, T S = RS2 = R , T R = T 2 S = S, and S R = T R 2 = T = RS. Therefore, the group is an Abelian group.
Group and Its Subsets
33
2.3
Homomorphism of Groups
group G is said to be homomorphic onto another group G', G' G, if one and only one element of G' corresponds to any element of G, if at least one element of G corresponds to any element of G', and if the correspondence is such that the product of two elements of G is in the same way to map onto the product of two corresponding elements of G'. In other words, if G' G, there is a onetomany correspondence between elements of G' and G and the correspondence is invariant in the multiplication of elements. The set H of elements in G which corresponds to the identity in G' forms an invariant subgroup of G, which is called the kernel of homomorphism. The quotient group G / H is isomorphic onto G'. G' describes only part of properties of G, namely, G' describes the property of the quotient group G / H , but does not describe the difference among elements in the kernel of homomorphism.
N
*A

8 . The Pauli matrices aa are defined as follows:
;),
u 2 = ( 5 )
,
u3=(;"1)
7
where Eabd is the totally antisymmetrical tensor of rank 3. Show that all possible products generated by 01 and 02 form a group. List the multiplication table of this group. Point out the order of this group, the order of each element, the classes, the invariant subgroups and their quotient groups. Prove that this group is isomorphic onto the symmetry group Dq for the square.
Solution. According to the multiplication rule of the Pauli matrices, there are 8 matrices generated by 01 and 02. The multiplication table is given in the Table. From the multiplication table, the set of these 8 elements is closed for the multiplication of elements. The multiplication of matrices satisfies the associative law. 1 is the identity E in the set. Any element in the set is selfinverse except for fia3. ia3 is the inverse of ia3. Therefore, this set forms a group G with order 8. The order of the identity E is 1. The orders of 1, fa1 and f a 2 all are 2. The orders of f i g 3 are both 4. 1 and 1 form two classes. {fal}, ( f a 2 ) and { f i a 3 } form
34
Problems and Solutions in Group Theory
the classes, respectively. There are five classes altogether. The invariant subgroups in the group are (1,l},(1, 1,01, 01}, (1, 1 , 0 2 ,  0 2 } , and { 1,1, i03,i03}. The cosets of (1, 1) are {kal}, {&02} and {fias}. The square of each coset is the invariant subgroup {fl}, the quotient so group is isomorphic onto the fourthorder inversion group Vq. The quotient groups of the next three invariant subgroups are all isomorphic onto the secondorder inversion group V2. G is homomorphic onto V4 with a 2 : 1 correspondence and onto V2 with the 4 : 1 correspondence where there are three different kernels of homomorphism.
I
1
cJ1
a 2
1 1
cJl
(71
a2
02
i3 a i3 a
02
1 1
01
02
a1
01
a2
02
ia3 ia3
a1 1
ia3
02
i u3
1
a 1
cJ2 i 3 a
1
i3 a
u2
ia3
02
01
1
a1
(72
a 2 a1
1
i3 a
u2
i3 a
i3 a
1
ia3 a2
cJ1
iu3
1
a1
a 2
1
01
81
1
01
a1
1
icJ3
cJ2
1
ia3
a2
i3 a
a2
1
1 0
02
1
01
01
ia3
ia3
1
i3 a
1
Let the fourfold rotation C around the 2 axis in the group D4 corre4 spond to io3,and the twofold rotation C; around the X axis correspond to 01. The correspondence between the remaining elements can be obtained by the products of elements:
Thus, the multiplications of elements also satisfy the onetoone correspondence, so two groups are isomorphic.
9. Show that the set of all possible products generated by i l and i 0 2 a forms a group. List the multiplication table of this group. Point out the order of this group, the order of each element, the classes, the invariant subgroups and their quotient groups in the group, respectively. Show that this group is not isomorphic onto the group D4.
Solution. There are 8 matrices generated by the products of iol and ia2. The multiplication table is as follows.
Group and I t s Subsets
35
1
ial
1
ial
ia2
il a 1
a02
iu3
i3 a iaz
a01
1
ia1
2u1
aa2
ia3
iff2
1
ia3
i3 a
iu2 i3 a
1
a01
iu3
a02
i3 a
1 ial iu2
a03
1 il a
iaz
a03
1
ia3 auz
iaz au3 1 il a ia2 i3 a
iz a
iu1 1 ia3
aaz
1 ial
ia2
il a 1
i3 a
ia2 ia3
ial 1 ia3
iaz
1 ia1
il a 1
ia3 1
ia1
iz a
a01
1
From the multiplication table, the set of these 8 elements is closed for the multiplication of elements. The multiplication of matrices satisfies the associative law. Therefore, this set forms a group G with order 8. The order of the identity 1 is 1. The order of 1 is 2. The orders of fia, all are 4. 1 and 1 form two classes. (fial}, (fiaz} and (&ia3} form the classes, respectively. There are five classes altogether. The invariant subgroups in the group are (1, 1}, (1, l,iol, ia1}, (1, 1,ia2, ia2}, and (1, 1,203, ia3}. Thecosetsof (&1}are (fial}, (fiaz} and {fia3}. The square of each coset is the invariant subgroup ( k l } , so the quotient group is isomorphic onto the fourthorder inversion group V4. The quotient groups of the next three invariant subgroups all are isomorphic onto the secondorder inversion group V2. G is homomorphic onto V4 with a 2 : 1 correspondence and onto V2 with the 4 : 1 correspondences where there are three different kernels of homomorphism. Because this group contains six fourthorder elements, this group is isomorphic neither onto the group D4, nor onto the group C4h. In literature, this group is usually called the quaternion group Q8. 10. Up to isomorphism, prove that there are only five different eighthorder group: The cyclic group c8, C4h =C4 xV2, the symmetry group D4 of a square, the quaternion group Q8 and D2h =D2 x V ~ .
Solution. The order of an element in an eighthorder group G, except for the identity, has to be 2, 4 or 8. If there is at least one eighthorder element, G is the cyclic group c8. If the orders of all elements in G , except for the identity E , are 2, we obtain from Problem 7 that G is an Abelian group, which is isomorphic onto D2h =D2 xV2. If there is no eighthorder element G, but there is at least one fourthorder element, denoted by R , the cyclic subgroup H = { E ,R, R 2 ,R3} generated by R is an invariant subgroup of G because its index is 2. Denote its coset by HSo = {So,SlrS2,S3), where RmSo = Sm, RmSj = Sj+m, and
36
Problems and Solutions in Group Theory
= Sj. Sj cannot be equal to R or R3, otherwise the order of Sj is 8. From the rearrangement theorem, S; cannot be equal to Sk. If S" R2, j= then Sj is a fourthorder element. Thus, SJrl = Sj" = R2Sj = Sj+2, and Sj"+2 SJT2 = R2. Now, there are three cases. The first case is that all four = Sj satisfy 5'; = R2. Then, 5'15'0 = RS; = R3, SoS1 = R3S? = R, and this group is isomorphic onto the quaternion group Qs. The correspondence is
Sj+4
The second case is that only a pair of Sj satisfy Sj" = R2,say S: = S$ = R 2 . Namely, S and S3 are the fourthorder elements, and SO and S2 are the 1 secondorder elements, S2 = Sz = E . Thus, S1So = RS; = R, SoS1 = o R3S: = R, SoR = SiS1 = S = RSo, SIR = SfSo = R2So = S2 = RS1, 1 and this group is isomorphic onto the Abelian group C4h =Cq xV2. The correspondence is
where a is the space inversion. The third case is that the orders of all Sj are 2, S"= E , then from R"Sj = Sj+" we have R = Sj+"Sj and j " Sj R = Sjm. Thus, the group is isomorphic onto D4. The correspondence " is
The orders of elements and the classes in these five groups are listed as follows.
Group
c 8
Q8
C4h
1storder 1 1
1
D4
D2h
1 1
No. of elements 2ndorder 4thorder 2 1 6 1 3 4 2 5 0 7
8thorder 4 0
0
No. of classes
0 0
11. Investigate all ninthorder group which are not isomorphic onto each
other. Solution. The order of any element in a ninthorder group G, except for the identity, has to be 3 or 9. If there is at least one ninthorder element in the group, then this group is a cyclic group Cg. If there is no ninthorder element, all elements in G, except for the identity, are the thirdorder elements. Arbitrary take a thirdorder element,
Group and Its Subsets
37
denoted by A. The cyclic subgroup generated by A is H = { E , A , A 2 } . Denote one right coset by H B = { B ,C, D } where A B = C , A C = D , and A D = B. Since B , C and D are all thirdorder elements, their square cannot be equal to E , A or A 2 . From the rearrangement theorem, their square cannot be equal to B , C or D. They have to be different from each other due to the uniqueness of the inverse element. Therefore, the remaining three elements in the group are B 2 , C2 and D2, which form another right coset of H . From the rearrangement theorem, AB2 = C B cannot be equal to C2 and B 2 ,so it has to be equal to D2. Similarly, we can calculate the remaining products and give the following multiplication table of the group. From the multiplication table, we know that this group is an Abelian group with nine classes. There are only two different ninthorder groups.
12. Investigate all tenthorder groups which are not isomorphic.
1
D B2
E E A A2 B C D B2 D2
A A A2 E C D B D2
A2
A2 E A D B C
B B C D B2
0 2
C
C D
D D B
B
0 2
c
B2 B2 D2
C2
C2
D2
0 2
c2
A2 E A D C B
B2 C2 D 2 B2
A A2 E
C2
B2
C2
B2
c2
D2
B2
C2 E
C2
C2
B2
C2
D2
A2 A
A E A2
D2 A2 A E
E A A2 B D C
c
B D
Solution. The order of any element in a tenthorder group G, except for the identity, has to be 2, 5 and 10. If there is at least one tenthorder element in the group, then this group is a cyclic group Clo. If all elements in a tenthorder group, except for the identity, are secondorder elements, arbitrary taking two different elements R and s, we have R2 = S2 = E , RS = T . T has to be a new element as we have known in Problem 1. Thus, we obtain a fourthorder subgroup { E ,R, S , T } isomorphic onto the fourthorder inversion group V4. Since the order of the subgroup is not a whole number divisor of ten, a contradiction. If there is no tenthorder element in G, but there is at least one fifthorder element, denoted as R, then the cyclic subgroup { E ,R, R2,R3, R4} generated by R is an invariant subgroup because its index is 2. Denote its coset by {SO,S~,S~,S~,S~}, = S,, R"Sj = Sj+", and where R"S0 Sj+5 = S . From the rearrangement theorem, S; cannot be equal to Sk. j
38
Problems and Solutions in Group Theory
If it is equal to Rj, where j is not a multiple of 5, then Sj is a tenthorder element, thus a contradiction. So Sj" = E , S all are secondorder elements. j Thus, we have R = Sj+"Sj and SjR" = Sjm. This group is the group " D5. Up to isomorphism, there are only two tenthorder group, C ~ and O D5. From the investigation on the sixthorder group (Problem 6) and the tenthorder group, we conclude that when the order g of a group G is the double of a prime number n > 1, the group has to be a cyclic group C2n or the symmetry group D, for a regular nsided polygon.
13. Give an example to show that the invariant subgroup of an invariant subgroup of a group G is not necessary an invariant subgroup of the group G. Conversely, show that if an invariant subgroup of a group G completely belongs to a subgroup H of G, then it is also an invariant subgroup of H .
Solution. Let H be a subgroup of G, and let H1 be a subgroup of H . For an arbitrary element R in H I , an element S R S  l conjugate to R for G may not be an element conjugate to R for H , because S E G may not belong to H . Conversely, an element SRSl conjugate to R for H must be an element conjugate to R for G, because S E H must belong to G. Therefore, the invariant subgroup of an invariant subgroup H of a group G is not necessary an invariant subgroup of G, but the invariant subgroup HI of a group G must be an invariant subgroup of the subgroup H c G if Hi C H . An example is as follows. The group T contains three twofold axes and four threefold axes. T contains an invariant subgroup D2 composed of the identity E and three twofold rotations. However, three secondorder subgroups in D2 are all the invariant subgroups of Dz, but not the invariant subgroup of T.
} be a class 14. In a finite group G of order g, let C, = { S , . . . , 1 S2, of G containing n(a) elements. For any two elements Si and Sj in the class C, (may be different or the same), show that the number m(a) of elements P E G satisfying Si = PSjPl is g/n(a).
Solution. For a given element Sj E C,, let the number of elements R E G, commutable with S j , be rn(a).It is easy to show that m(a)does not depend on Sj. In fact, if Si = PSjPl, PRPl commutes with Si. Under the multiplication rule of G, the set H of R E G, commutable with Sj, forms a subgroup of the finite group G, because if both R and
Group and Its Subsets
39
R’ can commute with S j , RR’ also commutes with Sj. The order of H is m(a). Note that H is not necessary an invariant subgroup of G. Any element T E G, which does not belong to the subgroup H , cannot commute with Sj. Let TSjTl = Si E Ca and Sj # Si. Any element T R in the left coset T H of the subgroup H satisfies TRSjRlTl = Si. The number of elements in the left coset TH is still rn(.>. On the other hand, for any element P which satisfies PSjPl = Si, we have
Thus, PlT E H , and P belongs to the left coset T H . Therefore, we obtain a onetoone correspondence between the element Si E C, and the left coset T H by the relation
TRSjRlTl = Si.
When Si = S j , T E H and T H = H . The index g/m(a) of the subgroup H is equal to the number n(a) of the elements Sj in the class C,. Namely, g = m(cr)n(a). Both the order m(a) of a subgroup H and the number n(a) of elements in a class are the divisors of the order g of the group G.
15. Prove that being the product of two subsets, the product of two classes in a group G must be a sum aggregate of a few whole classes. Namely, the sum aggregate contains all elements conjugate to any product of two elements belonging to the two classes, respectively. Solution. Let C, and C p be two classes in a group G. For given R E C, and S E C p , we have TRTl E C, and TSTl E C p , where T is an arbitrary element in G. Therefore,
T(I2S)Tl = TRTlTSTl E
c,co.
The conclusion is proved. It is easy to prove that the multiplicity of RS in the product space of C,Cp is the same as that of the element conjugate to RS. Therefore, considering the multiplicity f ( a ,p, r),we have
16. Calculate the multiplication table of the group T by extending the multiplication table of the subgroup C3 = { E , R1, R:} of T.
40
Problems and Solutions in Group Theory
Solution. The symmetry group T for a tetrahedron contains three twofold axes orthogonal to each other and four threefold axes. Denote the twofold rotations by TZ, Ty”and TZ, respectively. T = T:Ti. The threefold axes Z in T are respectively along the directions e , f ey f e,, and ex ey =t e,. Denoting by R1 the rotation through 2 ~ / around the direction ez+ey+e,, 3 we have
R;T:RI = T,”, R;T:R1 = T:, R1T:R: = Ty”, R1Ty”Rq= T:,
R;T;R1 = Ty”, RlTzR? = T,”.
The remaining threefold rotations can be defined as
Substituting them into the above formulas, we obtain
R1Ti = R3,
R1Ty”= R2,
R1T: = R4.
Taking the inverse of the formulas, we have
In terms of TzTy”= T: we obtain
T: Rq = R1, Ty”R4= R2, T;R4 = R3, T,2Ri = Ri, Ty“Ri= Rz, T,”R,” R;. =
The above equations give the formulas for the right cosets of the subgroup C3 and the formulas for leftmultiplying T:, T; and T,” on the elements in T. From them we can obtain the multiplication table of T. Arrange the rows in the multiplication table (the first column of the table) in the order of the subgroup C3 and its left cosets T;C3, Tic3 and T;C3, and the columns (the first row of the table) in the order of C3 and its right cosets C3T:, C3Ti and C3TZ. Thus, the multiplication table is divided into 16 subtables. Each subtable contains 3 x 3 multiplication elements. The first subtable in the first row is just the multiplication table
Group and Its Subsets
41
of the subgroup Cs. The remaining three subtables are obtained by rightmultiplying the first subtable with T z , T i and T j , respectively. The next three rows are obtained by leftmultiplying the first row with T:, T i and T’:, respectively. The relative arrangement of elements in each subtable is the same as the multiplication table of C3, where the only difference is the contained elements, which are given in the above formulas. Finally, we obtain multiplication table of the symmetry group T for the regular tet r ahedr a1.
17. The multiplication table of the finite group G is as follows.
IE E A B C D
F I J K L M N
A
A E F I J B C D M N K L
B B F A L K E N M J I D C
C C I K A L M E N F J B D
D D J L K A N M E I F C B
F F B E N M A L K D C J I
I I C M E N K A L B D F J
J J D N M E L K A C B I F
K K M I J F C D B N E L A
L L N J F I D B C E M A K
M M K C D B I J F L A N E
N N L D B C J F I A K E M
E A B C D F I J K L M N
a) Find the inverse of each element in G; b) Point out which elements can commute with any element in G; c) List the period and order of each element; d) Find the elements in each class of G; e) Find all invariant subgroups in G. For each invariant subgroup, list its cosets and point out onto which group its quotient group is iso
42
Problems and Solutions in Group Theory
morp hic; f ) Make a judgment whether G is isomorphic onto the tetrahedral symmetric group T, or isomorphic onto the regular sixsided polygon symmetric group DG.
Solution. a) The pairs of elements which are inverse to each other are given as follows: E' = E , D' = J ,
A' = A , K' = L ,
B' =F, M' = N .
C' = I
b) E and A can commute with any element in the group. c) The element whose order is 1 is the identity E . The element of order 2 is A . Its period is { E ,A } . The elements of order 3 are M and N . They belong to a common period { E ,M , } . N The elements of order 4 are B , C , D , F , I and J . They belong to the periods { E , B , A , F } , { E , C, A , I } , and { E , D , A , J } , respectively. The sixthorder elements are K and L. They belong to a common period { E ,K , N , A , M, } . L d) E and A form two classes, { E } and { A } . The elements of order 3 and 6 respectively form two classes, { M , N } and { K , L } . The elements of order 4 are divided into two classes, { B ,C, D} and { F ,I , J } . e) The invariant subgroup consists of several whole classes, and contains the period of each element. Its order has to be a divisor of 12, the order of G, namely, it has to be 2, 3, 4 or 6 . { E ,A } is an invariant subgroup of G. Its cosets are { B ,F } , {C,I } , { D ,J } , { K ,M}, and { L ,N } . Because B2 = C2 = D2 = A, the squares of the first three cosets all are the invariant subgroup. Thus, its quotient group is isomorphic onto the regular triangle symmetric group D3. { E ,M , N } is an invariant subgroup of G. Its cosets are { A ,K , L } , { B , C ,D } and { F , I ,J } . Due to B2 = F 2 = A , the squares of the last two cosets are not the invariant subgroup. Thus, its quotient group is isomorphic onto fourthorder cyclic group Cq. { E ,K , N , A , M, L } is an invariant subgroup with index 2. Its coset is { B ,I , D , F, C, J } . Its quotient group is isomorphic onto two order inversion group V2. f ) Since the group T does not contain any sixthorder element, the group G is not isomorphic onto T. Since D6 does not contain any fourthorder element, the group G is not isomorphic onto Dg.
Chapter 3
THEORY OF REPRESENTATIONS
3.1
Transformation Operators for a Scalar Function
An rndimensional matrix group D(G) is called a representation of the given group G, if G is isomorphic or homomorphic onto D ( G ) . The element D ( R ) in D (G), which is nonsingular , is called the representation matrix of the group element R E G in the representation D ( G ) . The trace T r D ( R ) x ( R ) is called the character of R in the representation D ( G ) . The representation matrix D ( E ) of the identity E is a unit matrix, and the representation matrices of two elements R' and R are mutually inverse matrices, D ( R  l ) = D(R)'. The representation D ( G ) is said to be faithful if G is isomorphic onto D ( G ) . If all matrices D ( R ) are unitary, D ( G ) is called a unitary representation. If all matrices D ( R ) are real orthogonal, D ( G ) is called a real orthogonal representation. D(G)* composed of the complex conjugate D ( R ) * of the representation matrices D ( R ) is the conjugate representation of G with respect to D ( G ) . The representation is called a selfconjugate representation if all characters in the representation are real. The representation D ( G ) is called the identical representation if D ( R ) = 1 for any element R in G. The dimension m of the representation is assumed to be finite in this book if without special notification. Denote simply by x all coordinates of degrees of freedom in a quantum system, and by $(x) the scalar wave function. Under a transformation R , 12: is transformed into x = Rx, ' the wave function $(x) is transformed into $'(x) = PR$(z). Being a scalar wave function, the value of the transformed wave function PR$ at the point Rx must be equal to that of the wave
*
*
43
44
Problems and Solutions an Group Theory
function $ before transformation a t the point z, namely
The transformed wave function PR$(~) be obtained from the original can wave function $(z) according to Eq. (3.1), namely, first to replace x with R  l x in the wave function $(x), then to regard it as the function of x, which is nothing but the transformed function PR$(~). is a linear PR operator and has a onetoone correspondence with the transformation R. This correspondence is invariant in the product of transformations, namely, PRPScorresponds to RS in the same rule. If the set of transformations R forms a group G, then the set PG of PR also forms a group and is isomorphic onto G. A linear operator L ( z ) for the scalar wave functions is transformed into L ' ( z ) in the transformation R:
Assume that a quantum system with the hamiltonian H ( z ) is described by a scalar wave function. R is said to be a symmetry transformation if PR can commute with H ( z ) . The set of symmetry transformations R forms the symmetry group G of the system. If the energy level E is m degeneracy, we denote by Q p ( z )the basis in the eigenfunction space. Then, the mdimensional eigenfunction space for the energy E keeps invariant under the application of PR. The matrix form of PR in this space with respect to the basis qp(z)is D ( R ) :
*
The set of D ( R ) forms a representation of the symmetry group.
1. Let G be a nonAbelian group, D ( G ) be a faithful representation of the group G, and D ( R ) be the representation matrix of element R. If the element R in G corresponds to the following matrix in the set, please decide whether the following set forms a representation of the group G. For example, in a), if R H D ( R ) t , please decide whether the set D ( G ) t composed of D ( R ) t forms a representation of G. a) D ( R ) t ; b) D ( R ) T ; c) D(I2l); d) D ( R ) * ; e) D ( R  l ) t ; f ) detD(R); g) Tr D ( R ) .
Solution. Each subproblem gives a onetoone correspondence between the element in G and the matrix in the given set of matrices. We should
Theory of Representations
45
judge whether their products satisfy the same onetoone correspondence in the given rule. If yes, it is a representation of G. Otherwise, it is not. Notice that D ( R S ) = D ( R ) D ( S ) . a) Since D ( R ) t D ( S ) t # D ( R S ) t , the set of D ( R ) t is not a representation of G. b) Since D ( R ) T D ( S ) T# D ( R S ) T ,the set of D(R)T is not a representation of G. c) Since D(R'>D(S') # D [ ( R S )  ' ] ,the set of D ( R  l ) is not a representation of G. d) Since D ( R ) * D ( S ) * D(RS)*,the set of D ( R ) * is a representation = of G, which is called the conjugate representation of D(G). e) Since D ( R  ' ) t D ( S  ' ) t = D [(RS)'It, the set of D ( R  l ) t is a representation of G. f ) Since det D ( R )det D ( S ) = det D ( R S ) ,the set of det D ( R ) is a onedimensional representation of G. g) Since " r D ( R ) TrD(S) # TrD(RS),the set of T r D ( R ) is not a representation of G.
2. The homogeneous function space of degree 2 spanned by the basis functions $ l ( z , y ) = x2, $2(z,y) = zy, and +3(z,y) = y2 is invariant in the
following rotations R in the twodimensional coordinate space. Calculate the matrix form D ( R ) of the corresponding transformation operator PR for the scalar function in the threedimensional function space:
a) R = ( I 01
" )'
13)
.=A(
2
1 fi a 1
),
c) R =
cos a  s i n a s i n a cosa
If replacing the basis function $2(z,y) = zy with fizy, how will the matrix form of the operator PR change? Solution. Three transformation matrices in the problem are all real orthogonal. The inverse matrix of R is equal to its transpose.
a)
46
Problems and Solutions in Group Theory
b)
c)
P R $ ~ y) = (z cos a + y sin a )2 (z,
= cos2a$1(z,y) sin(2a)$a(z,y) sin2a$3(x,y), PR$2(z,y) = ( z c o s a y s i n a ) (  z s i n a ycosa)
+
+
+
+
=  sin a cos a$1 (z, y)
+ cos(2a)$2(x, y) + sin a cos a$3 (z,y),
P ~ $ 3 ( z , y )= (zsina!
+ y c o s a )2
= sin2 a $ l ( z , y ) sin(2a)$+,y)
+ cos2 a!$3(z,y),
2 sin2 a 2 cos2 a  sin(2a) 2sin(2a) 2cos(2a) 2sin(2a) 2 sin2 a sin(2a) 2 cos2 a Multiplying the basis function $2(z,y) by we obtain the matrix form of PR by a diagonal similarity transformation M .
a,
M = diag (1,
h,I},

D(R)= M'D(R)M.
Each D ( R ) becomes a real orthogonal matrix.
Theory of Representations
47
3.2
Inequivalent and Irreducible Representations
Ir Two representations D ( G ) and D ( G ) with the same dimension are called equivalent to each other if there is a similarity transformation X relating two representation matrices for each element R in the group G, D ( R ) = X  ' D ( R ) X . The characters of each element R in two equivalent representations must be the same. For a finite group and a compact Lie group, any representation is equivalent to a unitary representation, two equivalent unitary representations can be related by a unitary similarity transformation, and two representations are equivalent if and only if the corresponding characters of each element in two representations are equal to each other.
j A representation D ( G ) of the group G is said to be reducible if the , representation matrix D ( R ) of any element R of G in D ( G ) can be transformed into the same form of the echelon matrix by a common similarity transformation X ,
Otherwise, it is called an irreducible representation. The necessary and sufficient condition for a reducible representation is that there is a nontrivial invariant subspace with respect to D ( G ) in its representation space. If both complementary subspaces are invariant with respect to D ( G ) ,the reducible representation is called the completely reducible one where there exists a common similarity transformation X such that the T ( R )in Eq. (3.3) for each element R of G is vanishing:
This form of reducible representation, where the representation matrix D ( R ) of each element R of G is a direct sum of two submatrices D ( l ) ( R ) and ( R ) ,is called the reduced representation. The representation D ( G ) is said to be the direct sum of two representations D ( l ) ( G )and D ( 2 ) ( G ) . For a finite group and a compact Lie group, any reducible representation is complete reducible, and can be transformed into the direct sum of two representations by a similarity transformation.
j , The Schur theorem says that X must be a null matrix if D ( l ) ( R ) X= XD(2)(R) holds for each element R of G where D ( l ) ( G )and D ( 2 ) ( G are )
48
Problems and Solutions i n Group Theory
two inequivalent and irreducible representations of G, and X must be a constant matrix if D ( R ) X = X D ( R ) holds for each element R of G where D ( G ) is an irreducible representations of G. The necessary and sufficient condition for two inequivalent and irreducible representations Di(G) and D j ( G ) of a finite group G is
*
where n(a) denotes the number of elements in the class Ca, and g is the order of G. Equation (3.5) says that the characters of two inequivalent and irreducible representations are orthogonal to each other in the class space with the weight n(a). When i = j , Eq. (3.5) is the necessary and sufficient condition for an irreducible representation of G. The characters in all inequivalent and irreducible representations of G form a complete bases in the class space, namely the characters satisfy
The number of all inequivalent and irreducible representations of a finite group G is equal to the number gc of the classes in G. The sum of dimension squares m; of all inequivalent and irreducible representations is equal to the order g of G,
Equations (3.57) are the necessary conditions satisfied by the characters in any finite group G. However, they are not enough to determine all characters of G. Considering some other methods, such as the representations of the quotient group of the invariant subgroups of G, the subduced and induced representations, the representations of the direct product of two subgroups etc., we can determine the characters in all inequivalent and irreducible representations of G, and obtain the character table of G. Based on the character table, we can find out the convenient forms of the unitary representation matrices of G. Usually, we choose as many diagonal representation matrices of generators as possible. For the complicated groups, the special method are needed. For a given quantum system with the Hamiltonian H ( z ) ,we first find its symmetry group G, where the transformation operator PR for each element
*
Theory of Representations
49
R in G can commute with the Hamiltonian H ( x ) ,
PR is called the symmetry operator of the system. Second, we study the symmetry group G of the system. We want to find out its character table and the convenient forms of the representation matrices of all generators R in G. The representation matrices of other elements can be calculated from those of the generators. Third, if the energy level E is rn degeneracy, we can find arbitrarily rn linearly independent eigenfunctions $ p (x) for the energy E :
Due to Eq. (3.8),PR$~(Z) be an eigenfunction of H ( z ) with the same must energy E , namely, the rndimensional space spanned by $ p ( 2 ) is invariant in the action of the symmetry operator PR. We can calculate the matrix form D ( R ) of PR in the basis function $,(x):
The set of D ( R ) forms a representation D ( G ) of the symmetry group G of the system, called the representation corresponding to the energy level E . The character x(R) =TrD(R) of R is easy to calculate. Generally speaking, the representation D ( G ) is reducible and not in the convenient form. In terms of the following method, the representation D ( G ) can be reduced into the direct sum of the irreducible representations and the eigenfunctions +,(x) can be combined into the basis functions transformed according to the irreducible representation. Make a similarity transformation X,
X'D(R)X = @
UjDj(R),
xff=
c
ajx:.
(3.10)
The rnultiplicity a j of the irreducible representation D j (G) in the reducible representation D ( G ) can be calculated from the orthogonal relation (3.5):
'
(3.11)
REG
Substituting it into Eq. (3.10), one is able to calculate the similarity transformation matrix X . Let R be the generator A whose representation matrix
50
Problems and Solutions in Group Theory
D j ( A ) is diagonal, where the problem becomes that of finding a similarity transformation X to diagonalize a matrix D ( A ) . An important point is to keep all the undetermined parameters in X waiting for the subsequent calculation. Substituting the matrix X into Eq. (3.10) for the remaining generator B whose representation matrix D J ( B )may not be diagonal. In the calculation, X' in Eq. (3.10) should be moved to the righthand side of the equation for avoiding the calculation for X  l . After the calculation of Eq. (3.10) for all generators of G, if there are still some parameters undetermined, they should be chosen by any feasible way. Do not leave them as the undetermined parameters. Make sure that detX should not be vanishing. The row index p of X is the same as that of D ( R ) ,and the column index of X is a combined set of the index j for the irreducible representation and its row index p. When a j > 1, an additional index r is needed to distinguish different Dj in the reduction (3.10). The new eigenfunctions are the combinations of $CL X : by
(3.12)
(ai,(x) is called a function belonging to the pth row of the irreducible representation Dj of G. When a j > 1, any linear combination of the functions (a{,(.) with the same j and p, where the combination coefficients are independent of j and p, is the eigenfunction belonging to the pth row of D j . This combination reflects the arbitrary choice of the undetermined parameters. The number of the undetermined parameters is Cjus .
For the inner product of two wave functions used in quantum mechanics, the symmetry operator PR is usually unitary (there is exceptional case). Under this condition, the functions belonging to two inequivalent and irreducible representations of the symmetry group G are orthogonal to each other.
*
(3.13)
where ($i(I!Pj) is called the reduced matrix elements, which is a parameter independent of the subscripts u and p. Equation (3.13) is called the Wigner
Theory of Representations
51
Eckart theorem. A group G is called the direct product of two subgroups, G = H I 8 H2, if each element in G can be expressed as a product of R E H I and S E H2, RS = S R , and there is no common element in two subgroups HI and H2 except for the identity E. A pure rotation in the threedimensional space is called a proper rotation. A rotation together with a space inversion is called an improper rotation. A proper point group consists of the proper rotations, and improper point group consists of the proper rotations and the improper rotations. The subset of the proper rotations in an improper point group G forms an invariant subgroup H in G with index 2. An improper point group G is said to be Itype if G = H 8 V 2 , where V2 = {e, a } is the twoorder inversion group. An improper point group G is said t o be Ptype if G does not contain the space inversion a. Multiplying a on each improper rotation in a Ptype improper point group G, we obtain a proper point group G'. G' is isomorphic onto G and contains an invariant subgroup with index 2.
*
3. Prove that the module of any representation matrix in a onedimensional representation of a finite group is equal to 1.
Solution. In fact, the representation matrix in a onedimensional representation is a complex number. For a finite group, any representation is equivalent to the unitary one, while a onedimensional representation is invariant under any similarity transformation. Therefore, it is a unitary one, where the module of any representation matrix is 1. There are many other methods to show this conclusion. In a finite group G, any element R in G with the order n satisfies R" = E , and the representation matrix of the identity E in any representation is the unit matrix. Therefore, in a onedimensional representation D ( G ) of G, D ( R ) n = D ( E ) = 1, namely, its module is 1.
4. Prove that any irreducible representation of an infinite Abelian group is
onedimensional. Solution. In an Abelian group G, the product of elements are commutable, so the representation matrix D ( R ) of any element R in an irreducible representation is commutable with the representation matrix of any other element in G. From the Schur theorem, D ( R ) has to be a constant matrix. Since R is an arbitrary element in G, and the representation is irreducible, the dimension of the representation has to be one.
52
Problems and Solutions in Group Theory
5 . Prove that the similarity transformation matrix between two equivalent
irreducible unitary representations of a finite group, if restricting its determinant t o be 1, has to be unitary.
Solution. Let D ( G ) and D(G) be two equivalent irreducible unitary representations of a finite group. There exists a unitary similarity transformation M , M t M = 1 and D(R) = M  l D ( R ) M . If they can be related by another similarity transformation X , D ( R )= X  l D ( R ) X , then
D ( R ) = ( X M y D ( R )( X M  1 ) .
From the Schur theorem, X M  ' = c l and X = cM, where c is a constant. Since d e t X = 1, IcI = 1. Thus, X is a unitary matrix.
6. Show that for a finite group G, the sum of the characters of all elements in any irreducible representation of G, except for the identical representation, is equal to zero.
Solution. For a finite group G, the characters of two inequivalent and irreducible representation satisfy Eq. (3.5). If the representation D i R ) ( is the identical representation, D i ( R ) = x i ( R ) = 1, Eq. (3.5) shows that the sum of characters xj(R)of any irreducible representation @(G) of G, except for the identical representation, is equal to zero.
7.The linear space spanned by the elements of a finite group G is called
the group space, where the group element is the basis and the vector is a linear combination of the group elements. The group space is invariant for left or rightmultiplication by any element of G. The set of the matrix forms D ( S ) of leftmultiplying an element S in the group space with respect to the basis R forms the regular representation D(G) of G. The set of the matrix forms D ( S ) of rightmultiplication forms an equivalent representation D(G) of G.
DTR(s)
DRT(s)
=
{0 { 01
1 when S R = T when S R # T ,
when R S = T when RS # T , when S = E 0 when S # E .
g
Theory of Representations
53
Calculate the similarity transformation matrix X between two equivalent regular representations in the group space of the D3 group. How to generalize the result to any other finite group?
Solution. Let X be the similarity transformation relating two equivalent regular representations:
PEG
PEG
Substituting the values of the representation matrix entries into the equation, we have X ( T S ) R= X T ( S R ) .Noticing that the rows and the columns are designated by the group elements. From the associative law which is satisfied by the multiplication of elements, the matrix entries of X are equal if the products of their row index and the column index as the group elements are equal. Namely, we may choose the matrix entries of X such that the entries are one when the product of the row index and the column index as the group elements is equal to a given element, say E , and the remaining entries are zero. In fact, the similarity transformation matrix X can be calculated from the multiplication table of the group, where the arrangement of rows and columns is the same as those in X , in the following way: The matrix entry in X is one if its position is the same as that of a given element, say E , in the multiplication table, otherwise it is zero. For the group D3, we may choose X by the positions of the identity E in the multiplication table. The reader is encouraged to write the matrix form of the generators of D3 in two equivalent regular representations according to the multiplication table of D3, and to check whether they are related by the similarity transformation X.
The multiplication table of
I E E D F
A A B C E D F B B C A F E D C C A B D F E
D 3
'1 0 0 0 0 0'
D F , D F E F E D A C B B A C C B A
X=
0 0 0 0 0
0 1 0 0 0
1 0 0 0 0
0 0 1 0 0
0 0 0 1 0
0 0 0 0 1
For the different choice of the given element, we can find g linearly independent similarity transformation matrices X , where g is the order of the group. In fact, the reduced form of the regular representation is the direct sum of all irreducible representations, where the multiplicity of each
54
Problems and Solutions in Group Theory
irreducible representation is equal to its dimension m j . The matrix, which can commute with all representation matrices in the reduced form of the regular representation, contains Cj m3 = g parameters.
8 . C, is a class in a finite group G, and W is a vector in the group space
composed of the sum of all elements in C,. Prove that the representation matrix of W in an irreducible representation is a constant matrix, and calculate this constant. Solution. Let C, = (R1, R2, . . . , & , } be a class in G, where n(a) (I is the number of elements in the class. Denote by x: the character of RI, in the midimensional irreducible representation DJ(R). For any element S in G , we have SRI,S' E C,. Obviously, if R k # Ri, then SRkS' # SRiS'. Therefore, the set of SRI,S' is the same as the class C,, so that sws' = w.
Since D j ( S ) D i ( W ) D j ( S )  l = D j ( W ) , according to the Schur theorem, D j ( W ) = c l . Taking its trace, we obtain n(a)xi = mjc. Then,
9. Prove that the number of the selfreciprocal classes in a finite group G is equal to the number of the inequivalent and irreducible selfconjugate representations of G. In other words, the number of pairs of the reciprocal classes is equal to the number of pairs of the inequivalent and irreducible nonselfconjugate representations.
Solution. Let n be the number of the selfreciprocal classes Ci in G, and m be the number of pairs of the reciprocal classes C, and CL'. The number of classes in G is gc = n 2m. In any representation, the character xi of a selfreciprocal class is real, and the characters of two reciprocal classes are complex conjugate to each other, x: = x ,  I . On the other hand, any character in a selfconjugate representation is real, but two characters of a class in the pair of two nonselfconjugate irreducible representations are complex conjugate to each other. The complex characters only occur for the nonselfreciprocal class in a nonselfconjugate representation. The sum of two characters of a class in two irreducible representations which are complex conjugate to each other is real, and the difference of two characters is
+
Theory of Representations
55
pure imaginary for a nonselfreciprocal class and is zero for a selfreciprocal class. Being a function of the classes, the characters in the inequivalent and irreducible representations of G are linearly independent, and form a complete set of bases in the class space. Any class function can be expressed as a linear combination of characters of the inequivalent and irreducible representations of G. On the one hand, since the differences of two characters in the different pairs of nonselfconjugate irreducible representations, being a function of the classes, is linear independent, the number of pairs of nonselfconjugate irreducible representations cannot be greater than the number m of pairs of the reciprocal classes Ccyand (.'; 7 On the other hand, define m functions of the classes FB,
Being a function of the classes, Fp can be expressed as a linear combination of characters of all inequivalent irreducible representations of G. Due to the explicit form of F', the characters of the selfconjugate representations will not appear in the combination, and the characters of the nonselfconjugate representations will appear only in the form of difference of two characters in a pair of representations. Therefore, the numbers of pairs of nonselfconjugate irreducible representations cannot be smaller than rn. In summary, the number of pairs of nonselfconjugate irreducible representations of a finite group G is equal to the number m of pairs of the reciprocal classes in G. As a result, the number of the selfconjugate irreducible representations of G is equal to the number n of the selfreciprocal classes in G. This completes the proof.
10. If the group G is a direct product H1 @H2 of two subgroups, show that the direct product of two irreducible representations of two subgroups is an irreducible representation of G.
Solution. Suppose that R E H I , S E H2, RS = S R E G = H1 8 H2, D j ( H I ) is an irreducible representation of H I with the representation space L1, and D"(H2) is an irreducible representation of H2 with the representation space L 2 . ,C1 and L2 have no nontrivial invariant subspace for the groups H I and H2, respectively. First, we will show that the direct product of two irreducible representations of subgroups is a representation of the direct product group. In fact, if there is a correspondence from the elements RS and R ' S in G to the matrices D ( R S ) = @ ( R ) x D " ( S ) and D(R'S') = D j ( R ' ) x D " ( S ) ,
56
Problems and Solutions in Group Theory
respectively, we have the correspondence from RSR'S' = RR'SS' to
D(RS)D(R'S')= [ D j ( R )x D"S)] [ D j ( R ' )x Dk(S')]
=
[Dj(R)Dj(R')] x [D"S)D"S')]
Therefore, the set of D ( R S ) forms a representation of the group G, denoted by D ( G ) = @ ( H I ) x Dk(H2). representation space, denoted by L, is Its the direct product of the spaces L1 and L2. Let a and b respectively be the arbitrary vectors in the spaces C and L2. Then, their direct product 1 a x b is a vector in the space C. Second, we will demonstrate that this representation is irreducible. According to the definition of irreducible representation, we are going to show by reduction to absurdity that there is no nontrivial invariant subspace in its representation space. If there is a nonzero invariant subspace L ' in the space L with respect to D(G), where a x b E C', C' must contain all vectors in the subspace L1 x b because DZ(H1) x D j ( E ) c D ( G ) . Further, L' must contain all vectors in the space L1 x L2 = L because DZ(E) x D j ( H 2 ) c D ( G ) . This leads to C = L'. If D j ( R ) or D k ( S ) is replaced with another inequivalent irreducible representation, it is easy to prove by reduction to absurdity that the representation by the direct product is also changed into another inequivalent irreducible representation. If G is a finite group, the number of the inequivalent irreducible representations D J ( H 1 )is equal to the number of classes in the subgroup H I , and the number of the inequivalent irreducible representations D"(H2) is equal to the number of classes in the subgroup H2. While the number of classes in the direct product group G is equal to the product of the numbers of classes in two subgroups, which is just equal to the number of the inequivalent irreducible representations of G obtained from the direct product of representations. Therefore, each irreducible representation of the direct product group can all be expressed as the direct product of the irreducible representations of two subgroups. 11. Let each element in D3 be the coordinate transformation in twodimensional space:
Theory of Representations
57
where R is equal t o the representation matrix in the twodimensional representation DE(D3). For the generators D and A in D3, we have
The fourdimensional function space spanned by the following basis functions is invariant in the group D3:
$l(XC,Y>
= x3,
$2(X,Y>
= X2Y,
+3(X,Y)
= XY2,
744(X,Y>
= Y3.
Calculate the representation matrices of the generators D and A of D3 in this representation. Then, reduce this representation into the direct sum of the irreducible representations of D3, and construct new basis function which is the linear combination of the original basis functions and belongs to the irreducible representation.
Solution. First, according to the formula
PR$p(X)
= $p(Rlx) =
c
U
+u(x)Dup(R),
one is able to calculate the representation matrices of the generators D and A of D3 in the fourdimensional space spanned by the given basis functions, where Rlx means
For the element D , we have
2"
= (x
+ &y)
/2,
yt/ = (&x
 y) /2,
58
Problems and Solutions in Group Theory
Hence,
3&
For the element A, we have
3
&I
Hence,
1 0 0 0
D ( A ) = (O" 0 1 O ) 0 0
0 0 01
The characters are
*
+ x(A) = O = xA1(A)+ xA'(A) + xE(A).
The similarity transformation matrix X satisfies
x(D) = 1 = XA1(D) XA"D) + XE(D),
x
D(D)X = 2
(
2 0 0 0
0 0 0 2 0 0 0 1& 0 &  1
)
,
D(A)X = X
(
1 0 0 0 010 0 0 0 1 0 0 0 01
The second equality gives
Theory of Representations
59
Substituting it into the first equality, we have
The solutions are a2 =  3 ~ 1 ,bl = 3b2 and choose three constants, we obtain 1 0 10
c1
= c2 = dl = d2. Arbitrary
x=(
3 O
0 01) 10
0 101
The new basis , which belongs to the pth row of the irreducible repref q sentation I?, can be calculated by X ,
12. Calculate the characters and the representation matrices of the proper symmetry group 0 of a cube with the method of the quotient group and the method of coordinate transformations. Solution. The group 0 is the proper symmetry group of a cube, composed of three fourfold axes, four threefold axes and six twofold axes. Its order is g = 24. The generators of three fourfold axes are the rotations around the directions of three coordinate axes through 7r angle, denoted by T z , Tyand T,, respectively. Four threefold axes are along the following four directions with the generators Rj:
60
Problems and Solutions in Group Theory
The twofold axes are along the following six directions with the generators
Sk:
s : 1
s 3
s 5
(ex + e y > / f i ,
s 2
: (ey f e J J 2 , : (ez + e z > / f i ,
s 4
s 6
: (ez  e y > / d , : (ey  e z > / J 2 , : (ez  e z ) / f i 
T is an invariant subgroup of 0 with the index 2. Through the direct calculation we obtain the coset formulas of T:
E
rightmultiply by SI leftmultiply by SI T Z T,” T: TZ SZ T2
S 2
R1 T: T Z
Rz
S5
R3 Ty T X
R4
S6
S 3
R: Tx Ty
Ri
S 4
55
Ri T Z Ty”
Ri
S3 S6
SI T: SI Tz
S 4
the following formulas are just the opposite to the second Due to set of formulas:
Now, rightmultiplying S on the multiplication table of T given in Problem 1 16 of Chapter 2, we obtain the rightupper part of the multiplication table of 0. Then, leftmultiplying S1 on the upper part of the multiplication table of 0, we obtain its lower part. In the calculation, we only make the replacement of elements according to the above coset formulas of T.
The rightupper part of multiplication table of 0
Theory of Representations
T h e leftlower part of multiplication table of 0
61
S Z
s 4
s 6
T h e rightlower part of multiplication table of 0
T,” Rf
R4
2
Rl
R4
Ty”
The group 0 has five classes: C1 = {E}, 2 = {TZ, t ,T:}, C = C T 3 {Tz,T2,T,,T,”,Tz,T,3),,R $ l 5 j 5 4}, and C5 = { S k ,1 5 k 5 C = {Rj 4 S}. From l 2 l 2 22 32 32 = 24, we know that the group 0 has five inequivalent irreducible representations, denoted by A, B,E,TI,and T2,with dimensions 1, 1, 2, 3, and 3, respectively. The quotient group O / T is isomorphic onto V2. From the quotient group, we obtain two onedimensional representations: The identical representation D A = 1 and the antisymmetric representation D with the characters:
+ + + +
XB(Cl) XB(C2) XB(C4) 1, = = =
XB(C3) X B ( C 5 ) = 1. =
For a onedimensional representation, the character is the same as the representation matrix.
62
Problems and Solutions an Group Theory
The group 0 contains another invariant subgroup D2 of order 6 , composed of four elements E , TZ, Ty" and TZ. Its cosets are R1D2, R:D2, S1D2, S3Dz and S5D2. As we know that two cosets RH and SH are different if RlS does not belong to H . Now, it is easy to check that these five cosets all are different. The quotient group contains two threeorder elements (R1D2 and R:D2) and three twoorder elements (S1D2, S3D2 and S5D2) so that it is isomorphic onto the group D3. Let R1D2, R:D2 and SlD2 respectively correspond to D , F and A in D3. S5D2 has to correspond to B in D 3 , because RlSl = Ty" = S5T: E S5D3. Then, S3D2 corresponds to C in D 3 . In constructing the multiplication table of D 3 , the only independent multiplication formula is DA = B. The remaining formulas can be derived from it and the orders of elements. Therefore, the correspondence we just established is invariant to the product of elements. The group D3 has a twodimensional irreducible representation, which is a nonfaithful representation of 0. The elements in 0 belonging to the invariant subgroup D3 correspond to the same representation matrix in this representation, so do those belonging to its coset. Note that R1D2 = { R i ,R2, R3, R4}, R:D2 = {R:,Ri7 R j } , S1D2 = {S17 Rg7 S2,Tz,TZ}, s3D2 = {s37s47Tz,T23)7 and s5D2 = {s5,s6,Ty,Ty3}*
f ( E ) = 2,
XE(R1)= x E ( R ? ) 1, =
XE(S1)= XE(S3)= XE(S5)= 0,
D~(R:= ) 2
D"(S5) =
f (2
( ' "> '
a1
D E ( S l )=
7)
,
('0  1 ) '
1
fi
We can establish one threedimensional representation of 0 by the coordinate transformation matrices. In fact, being a rotation in the threedimensional space, the representation matrices of some elements in 0 are easy to obtain:
Theory of Representations
63
DT1(R1) =
(I..;)
001
)
DT'(S1) =
(0: 01)
10 0
.
Thus, the characters of classes in 0 are
XT'(C1) 3) XT1(C2) 1) = = XT1(C4) 0, XTl(C5)= 1. =
XT1(C3) 1, =
Another threedimensional irreducible representation DT2 (0)can be obtained by the direct product of the antisymmetric representation D B (0) and the threedimensional representation DT1 : (0)
XT2(C1) 3) XT2(C2) 1, = = XT2(C4) 0, XT2(C5) 1, = =
XT2(C3) 1, =
13. The multiplication table for a group G of order 12 is as follows.
I
E
E E A B C D F I J K L M N A B E K N D M I L C J F
A
A
B C
D F I J K L
M
N
B B E A L F N J M C K I D
C C I N D E K L B M A F J
D D L J E C M A N F I K B
F F K M B L I E D N J C A
I I N C M J E F L A D B K
J J D L I M B N K E F A C
K K M F N A L C E J B D I
L L J D F B C K A I E N M
M M F K J I A D C B N E L
N N C I A K J B F D M L E
a) Find out the inverse of each element in G; b) Point out which elements can commute with any element in the group; c) List the period and the order of each element; d) Find out the elements in each class of G; e) Find out all invariant subgroups in G. For each invariant subgroup, list its cosets and point out onto which group its quotient group is isomorphic; f ) Establish the character table of G;
64
Problems and Solutions in Group Theory
g) Decide whether G is isomorphic onto the symmetry group T of the tetrahedron, or the symmetry group D6 of the regular sixsided polygon.
Solution. a) E' = E , A' = B , C' = D , Fl = I , Jl = K , L' = L , M  l = M , and Nl = N . b) Only the identity E can commute with any element in G. c) The order of the identity E is one. The orders of L , A and N are 4 two, and the orders of A , B , C , D ,F , I , J and K all are three. d) The group G contains four classes, { E } ,{ L , M , N}, A , C, F, J } , { and { B , D , I , K } . The first two are selfreciprocal classes, but the last
two are the mutual reciprocal classes. e) G contains only one nontrivial invariant subgroup { E , L , M,N } with index 3. Its cosets are { A , C, F, J } and { B , D , I , K } , and its quotient group is isomorphic onto the cyclic group Cs. f ) From the number of the classes in G, we know there are four inequivalent irreducible representations in G. Since l2 l2 l 2 32 = 12, there are three onedimensional representations and one threedimensional irreducible representation in G. Three onedimensional representations can be calculated from the quotient group of the invariant subgroup. The characters in the threedimensional representation of G can be determined by the orthogonality.
+ + +
g) From the orders of elements and the classes in G, we know that G is isomorphic onto the group T, but not isomorphic onto the group D6.
14. Calculate the character table of the group G given in Problem 17 of Chapter 2. Solution. The group G in Problem 17 of Chapter 2 has 12 elements, divided into 6 classes. Since l 2 + l 2 + l2 + l 2 + 22 + 22 = 12, there are
four onedimensional and two twodimensional inequivalent irreducible representations in G. There are three invariant subgroups in G. The invariant subgroup { E ,K , N , A , M , } has index 2. Its coset is { B ,I , D, F, C, J } . L Its quotient group is isomorphic onto the twoorder inversion group V2. From the irreducible representations of the quotient group we obtain
Theory of Representations
65
two onedimensional representations, including the identity representation D ( l ) ( R )= 1. The invariant subgroup { E ,N , M } has index 4. Its cosets are { A ,K , L } , { B , C , D } and { F, I , J } . The quotient group is isomorphic onto the cyclic group C4,where { A , K , L } is a twoorder element. From this quotient group we obtain four onedimensional representations, including two which are known. The invariant subgroup { E ,A } has index 6. Its cosets are { B , F } , { C , I } , { D , J } , { K , M } and { L , N } . The quotient group is isomorphic onto the group D3, where { K ,M } and { L ,N } are threeorder elements. From the twodimensional irreducible representation of D3 we obtain the irreducible representation D5 of G. The representation matrices for the generators K and 23, which correspond respectively to D and A in D3, are
Another twodimensional irreducible representation D6 of G is the direct product D3 x D5,
D 6 ( B )= i D 5 ( B ) ,
D 6 ( K )= D5(K).
Thus, we obtain the character tables of G.
E
x '
x2
x3
x4
x 5
x6
1 1 1 1 2 2
A 1 1
1 1 2 2
BCD 1 1 i i 0 0
FIJ 1 1 i i 0 0
KL 1 1
1 1 1 1
MN
1 1 1 1 1 1
3.3
Subduced and Induced Representations
Let the order of a finite group G be g , and H = {TI = E , T2, . . . , Th} be a subgroup of G with the order h and the index n = g / h . Denote its leftcosets by R,H, 2 r n. For unification, the subgroup is denoted by R1H with R1 = E. Although Rj are not determined uniquely, we make choice of them arbitrarily. Then, any element in the group G can be expressed as RTTt. Let D j (G) be an mjdimensional irreducible representation of G. The set of those matrices Dj(T,) corresponding to the elements in H
*
< <
66
Problems and Solutions in Group Theory
constitutes a representation of the subgroup H , denoted by D j ( H ) , which is called the subduced representation from an irreducible representation Dj(G) of the group G with respect to the subgroup H . Generally, the subduced representation is reducible, and can be reduced with respect to the irreducible representations Dk( H ) of the subgroup H
(3.14)
where m is the dimension of the representation D k ( H ) of H , and E ( P ) is k the number of elements contained in the class of H .
*
c p
Denote by q!Jp the
bases in the representation space of
Dk(H)
We define an extended space of dimension nmk with the bases qTp= = +/.L.The extended space is invariant to the group PR,+~, where G. We can calculate an n%kdimensional representation Ak(G) of G in the following way. For any given element S in G, we calculate SR, for each R,, which can be expressed in the form of RUTt, where u and t are completely determined by S and T . Because
we obtain
This representation Ak(G) is called the induced representation from the irreducible representation D k ( H ) of the subgroup H with respect to the group G. In general, the induced representation is reducible and can be reduced with respect, to the irreducible representation @(G) of G:
Theory of Representations
67
where n(a) is the number of the class C of G. , Calculate the character xk(S)of the element S in the induced representation A k ( G ) . Let S belong to the class C, in G. In general, some elements in C, belong to the subgroup H , and constitute a few classes of the subgroup H , denoted by It is possible that no element in C , belongs to the subgroup H . For this case we say that no p exists. F'rom Eq. (3.15), the diagonal element of A k ( S )may appear only when T = u, i.e., SR, = R,Tt. Namely, xk(S)is nonvanishing only when the class C, contains a few elements belonging to the subgroup H . Denoting by lcp the number of the different R, satisfying SR, = R,Tt, where Tt belongs to the class c with the character j$ in the irreducible representation D k ( H ) of p H , we have = Ca q$. h o r n Problem 14 in Chapter 2, the number of elements R in G satisfying S R = RTt is m(a) = g / n ( a ) .Such elements R, in general, can be expressed in the form of R,T,, so SR, = R, (T,TtTgl), while xk(T,TtTL1)Zk(Tt). = On the other hand, the number of TP in the subgroup H which satisfy TYTtT;l = Tt is ~ ( p = h/%(P). If R,Tz satisfy S(R,T,) = (R,T,)Tt, ) then R,T,T, also satisfy this formula. However, the latter does not make new contribution to the characters x"(S), ~p = m ( a ) / E ( P ) , so
*
cp.
xk
(3.17)
By the way, we do not need to consider the case of SR' = R'Ttt with another Tt/ E ED, because its contribution to the characters xk(S) been has calculated. In fact, letting Ttt = T,TtT;l, we have S(R'T,) = (R'T,)Tt. Note that the elements Tt in the class of H all belong to the class C, of G. Obviously, the different classes Ca correspond to the different classes Therefore, the sum over the class a in Eq. (3.16) is equivalent to the sum over all classes ??p in H . Due to xj(S) = xj (Rl S R ) , it is easy to show from Eq. (3.17) that two multiplicities a j k in Eq. (3.14) and b j k in Eq. (3.16) are equal:
cp
cp.
The formula (3.18) is called the F'robenius theorem.
68
Problems and Solutions in Group Theory
15. Calculate all inequivalent irreducible representations of the group D2n+l with the method of induced representation.
Solution. The group D2n+l contains a ( 2 n + 1)fold axis, called the principal axis, and ( 2 n + 1) equivalent twofold axes, located in the plane orthogonal to the principal axis. Denote by C2n+l the generator of the (2n+ 1)fold axis and by C l the generator of one twofold axis. 2 and C2l are two generators of D2n+l. Due to the orthogonality, Czn+1 = C2’ CFi+l. The C2‘ order of Dzn+l is g = 4 n 2. The number of the classes in D2n+l is gc = n + 2. There are two onedimensional and n twodimensional inequivalent representations of D2n+l. The group D2n+l contains an invariant subgroup C2n+l with index two. Its coset consists of all twofold rotations whose axes are orthogonal to the principal axis. The quotient group is isomorphic onto the twoorder inversion group V2, which gives two onedimensional inequivalent irreducible representations of D2n+l, respectively denoted by A and B:
+
DA(C2n+l) = DB(Gn+1) = DA(Cz’)= 1,
DB(C2’)= 1.
The invariant subgroup C2n+l is a cyclic group. It has ( 2 n + 1) onedimensional inequivalent representations. Denote the basis in each onedimensional representation space by @,
Extending the onedimensional space by defining another basis we have
qj 5
= C2l@,
2: C+ is used. Thus, we obtain n twowhere the formula C2n+lC2/= Cl ; , dimensional inequivalent irreducible representations of the group D2n+l, denoted by E j , 1 5 j 5 n,
The remaining representations with the different j are either equivalent to one of the above representations or reducible.
16. Calculate all inequivalent irreducible representations of the group D2n with the method of induced representation.
Theory of Representations
69
Solution. The group D272. contains a (2n)fold axis, called the principal axis, and (2n) twofold axes, located in the plane orthogonal to the principal axis. The twofold axes are divided into two sets, each of which contains n equivalent twofold axes. They form two classes, respectively. Denote by C2, the generator of the (2n)fold axis and by C2' the generator of one twofold axis belonging to one class. Define C2" = C2nC2/ = C2/CT,, belonging to another class. The angle between two twofold axes for and C2n is x/n. C2n and C t are two generators of D2,. The order of D2n is g = 4n. The 2 number of the classes in Dan is gc = n + 3. There are four onedimensional and n  1 twodimensional inequivalent irreducible representations of D2,. The group D2, contains four invariant subgroups. One subgroup is C2n composed of all rotations around the principal axis. Its index is two, and its quotient group is isomorphic onto the twoorder inversion group V2, which gives two onedimensional inequivalent representations of D2,, denoted by A1 and A2:
The subgroup C, of Can composed of the even powers of C2, is also an invariant subgroup of D2n. Its index is four. One of its coset consists of the odd powers of C2,. The remaining cosets are two classes of the twofold rotations, respectively. The quotient group is isomorphic onto the fourorder inversion group V4, which gives four onedimensional inequivalent representations of Dan including two known. The new representations are denoted by B1 and B2:
The set composed of C, and one class of the twofold rotations also forms an invariant subgroup of D2,. These two subgroups are both the twoorder groups. The representations of D2, provided by their quotient groups are nothing but A l , B1 and A1 , Bz, respectively. The invariant subgroup C2, is a cyclic group. It has (2n) onedimensional inequivalent representation. Denote the basis in each onedimensional representation space by $ j ,
Extending the onedimensional space by defining another basis we have
q '= C2/$ j , Y
70
Problems and Solutions i n Group Theory
Thus, we obtain ( n  1) twodimensional inequivalent irreducible representations of the group D2n, denoted by E j , 1 5 j 5 n  1,
The remaining representations with the different j are either equivalent to one of the above representations or reducible.
17. Calculate all inequivalent irreducible representations of the symmetry group 0 of a cube with the method of induced representation.
Solution. In Problem 12 we have calculated all inequivalent irreducible representations of the group 0 with the method of the quotient groups and the method of coordinate transformations. In the present problem we will calculate them with the method of induced representation. In Problem 12 we have given the multiplication table of the group 0. Now, we use the same notation as that used there. The group 0 contains a subgroup D4 of order 8, composed of one fourfold axis along the z axis and four twofold axes in the zy plane. The generator of the fourfold axis is denoted by T', and four twofold rotations are denoted by T:, S1, i , and S2. The angle of two neighboring twofold T axes is 7r/4. The index of D4 in 0 is 3. The characters table of the subgroup D4 is as follows.
The characters table of 0 The characters table of
D4
1
1 1 1
1 1
1
1 2 1
We calculate the character of the induced representation from B1 of with respect to 0 by Eq. (3.17),
D4
24 x 2 x (1) 24 x 1  3, XB1(TZ) = = 1, 8 8x6 24 X B V 3 = 8x3 (1 x 1 + 2 x 1) = 3 , 24 x 2 x (1) = 1, xB'(R1) = 0, X B 1 (S1)= 8x6 1 x 32 + 6 x (1)2 3 x (3)2 + 6 x (1)2 + 0 = 48.
X y E )=
~
+
Theory of Representations
71
Because the characters is orthogonal to that of the identical representation
A , this representation is the direct sum of a onedimensional representation B and a twodimensional representation E of the group 0.Denote by $1 the basis of the representation B1 of Dq. Extend the space by defining new bases $2 = R1$1 and 4 3 = RT$1. By making use of the multiplication table of 0 given in Problem 12, we have TzR1 = S3= RTT; and T,Rq = T i = RIS1. Thus,
Tz41 = 41,
Tz$2
= 43,
Tt.43 =  4 2 ,
D(Tt.)=
(
1 0 0 0 0 I), 0 1 0
0 0 1 D(R1) = (1 0 0 ) . 0 1 0
We obtain the reduced form of the direct sum of B and E:
We calculate the characters for the induced representation from A2 of D with respect to 0 by Eq. (3.17) 4
24 x 1 24x2~1 xA2(E>  3, X A 2 ( T Z ) = 8 6 = 1, = t4 xA2(T;)= (1 x I + 2 x (1)) = 1, 8x3 24 x 2 x (1) = 1, XA"S,> = XA2(R1) = 0, 8x6 1 x 32 6 x l2+ 3 x (1)2 6 x (1)2 + 0 = 24.
+
+
This is the irreducible representation TI of 0. Denote by $1 the basis of the representation A2 of Dq. Extending the space by defining new bases $2 = R1$1 , $3 = R:$I , we have
R1$1 = $2, TZ$l = $ 1 7
R1$2 = $3, T d 2 = $3 7
R1$3 = $1, Tz+3 = $2.
72
Problems and Solutions in Group Theory
By a simple similarity transformation (the cyclic of 1,2,3), this form of representation coincides with the form given in Problem 12. We calculate the characters for the induced representation from B of 2 Dq with respect to 0 by Eq. (3.17)
24 x 1 24
XB2((TZ) =
(1 x 1 2 x (1)) = 1, 2 4 x 2 ~ 1 = 1, XB"(R1) = 0 , X B 2 ( S , )= 8x6 1 x 32 + 6 x (1)2 + 3 x (1)2 + 6 x l2 = 24. +0
+
24 x 2 x (1) = 1, 8 6
This is the irreducible representation T of 0. The characters are the same 2 as those in TI except for the signs of the characters of Tz and S1. For the group 0, 7'2 = TI x B . The calculation is similar.
18. The regular icosahedron is shown in Fig. 3.1. The opposite vertices are denoted by Aj and Bj, 0 5 j 5 5. Choose the coordinate frame such that the origin is at the center 0 of the regular icosahedron, and the x axis is in the direction from the vertex Bo to the vertex Ao. Aj are located above the z plane. y
A3
B5
B O Fig. 3.1 The regular icosahedron.
Theory of Representations
73
In the regular icosahedron, there are six fivefold axes around the directions from Bj to A, with the generators Tj, 0 5 j 5 5. Except for one fivefold axis ( j = 0) which is along the positive z axis, the polar angles of the remaining fivefold axes all are 01, and their azimuthal angles are 2 ( j  1)7r/5, respectively. There are 10 threefold axes along the lines connecting the centers of two opposite triangles with the generators R j , 1 5 j 5 10. The polar angles of the threefold axes are 0 2 when 1 5 j 5 5, and 03 when 6 5 j 5 10. Their azimuthal angles respectively are ( 2 j  1)7r/5. There are 15 twofold axes along the lines connecting the central points of two opposite edges with the generators Sj, l 5 j 5 15. The polar angles of the twofold axes are 04 when 1 5 j 5 5, 05 when 6 5 j 5 10, and 7r/2 when 11 5 j 5 15. Their azimuthal angles are 2 ( j  1)7r/5 when 1 5 j 5 5, ( 2 j  l ) n / 5 when 6 5 j 5 10, and ( 4 j  3)7r/lO when 11 5 j 5 15, respectively. t a n e l = 2, tano2 = 3  &, tano3 = 3
+ &,
tan04 = (&  1)/2,
t a d 5 = (&
+ 1)/2.
(3.19)
All axes are the nonpolar axes, and any two axes with the same fold are equivalent to each other. The proper symmetry group of an icosahedron is denoted by I, which contains 60 elements and five classes. The classes, denoted by E , C5, Cz, C3 and C2, contain 1, 12, 12, 20, and 15 elements, respectively. Calculate the character table of the proper symmetric group I of an icosahedron with the method of induced representation.
Solution. Since l2+ 32+ 32 + 42 + 52 = 60, the dimensions of five inequivalent irreducible representations respectively are 1 , 3 , 3, 4 and 5, denoted by A , T I ,T2, G and H . The representation A is the identical representation.
The character table of T The character table of
D5
A2
1 P
0
The group I does not contain any nontrivial invariant subgroup. There
74
Problems and Solutions in Group Theory
are two bigger subgroups T and D5. The subgroup T consists of three twofold axes (SS,5’12 and S1) and four threefold axes (Rc, RZ’, R4, and R,d.), the subgroup D5 contains one fivefold (To)and five twofold axes and ( S j , 11 5 j 5 15). Their character tables are given in the Table. For the induced representation from the identical representation A of T with respect to I, we have from Eq. (3.17)
XA(E)= 5,
XA(C5)= XA(CZ)= 0 ,
XA(C2)=
6 0 x 3 ~ 1 = 1, 12 x 15
XA(C3)=  4 x l + 4 x 1 ) = 2 , 6o ( 12 x 20 1 x 52 + 0 + 0 + 15 x l2+ 20 x 22 = 120.
Because there is no negative character, this representation must be equivalent to the direct sum of an identity representation A and a 4dimensional representation G. Subtracting the characters of A , we obtain the characters of the irreducible representation G of I:
XA(E) = XA(C5) = XA(CZ)= XA(C2)= XA(C3)= 1, xG(E) = 4, xG(C5) = xC(C5”) 1, = XG(C2) = 0, XG(C3)= 1.
For the induced representation from the representation E of T with respect to I, we have
60 x 3 = XE(C5)= XE(CZ) = 0 , XE(C2) =  1, 12 x 15 60 XE(C3)=  x bJ + 4 x b J 2 ) = 1, (4 12 x 20 1 x 52 + 0 + 0 + 15 x l2+ 20 x (1)2 = 60.
X E ( E )= 5,
It is the fivedimensional irreducible representation H of I:
XH(E) = 5,
XH(C5) XH(C5”) 0 , XH(C2)= 1, XH(C3)= 1. = =
By the induced representations from the irreducible representation of T, it is impossible to distinguish two classes C5 and Cl. For the induced representation from El of D5 with respect to I, we have
60 x 2 6 0 x 2 ~ ~ XEl(E) =  12, XE1(C5)= 10 x 12 = P, lox 1 60 x 2 x (p’) = p 1 x”l(C,”) = X E 1 (C2) = X E 1 (C3) = 0, 10 x 12
7
p = (A 1)/2,
p1 = (fi 1)/2,
+
1 x 122 + 12 x p 2 + 12 x (P’)~ 0 +
+ 0 = 180.
Theory of Representations
75
It is the direct sum of three irreducible representations. Because the characters are not orthogonal to those of the representations G and H , so this representation contains one representation G and one representation H . Subtracting the characters of G and H , we obtain the characters of the threedimensional irreducible representation T I of I:
xT1(E)= 3,
XT1(C2)
XTl(C5)
=pl,
XTl(C5”)= p,
= 1,
XT1(C3)
= 0.
The characters table of the group 1
By the induced representation from E of D5, we can calculate the char2 acters of T2 similarly. In comparison with the characters of TI, only the characters of C and Cz are in5 terchanged. Finally, the character table of the group I is listed in the table.
G
p 1
p1 1 0
1 0 1
1
19. Calculate the reduction of the subduced representation from each irreducible representation of the group I with respect to the subgroups Cg, D5 and T.
Solution. The character tables of I, D5 and T have been given in Problem 18. C5 is a cyclic group, whose character table is easy to write. The important thing in calculation is to identify which class of I each element in the subgroup belongs to. Then, the character formula (3.5) can be used to calculate the reduction of the subduced representation from each irreducible representation of the group I with respect to the subgroups C5, D5 and T. Here we neglect the calculation process, but only list the results.
The reduction of the subduced representation from each irreducible representation of I with respect t o the subgroup C5
E
A A2
A3
c 5
c g
1
r12
714
c:
1 r13 77 r14 VZ 1 P p’ 1 0
c“ 5
1 r14 r13 r12 71 1 P’ p 1
0
Reduction of the subduced representation
1 1 1 1 1 l 3 3 4 5
1
1
rl
A4 A5 A
v3
714
v2
rl
713
Ti T 2 G
H
l p’ p 1 O
1 P p‘ 1 0
76
Problems and Solutions in Group Theory
The reduction of the subduced representation from each irreducible representation of I with respect t o the subgroup
D5
I
A1
E
1
2C5
1
2Cz
1
5C4
1
I
Reduction of the subduced representation
A2 El E2
A
1
2
2 1
1 p
pl 1 p’
1 p’
P 1
1
0
0
1 1 1 0 1
Ti
T2
3 3
4
p
1 0
p p’
1 0
G H
5
A1 A2 @ Ei A2 @ E2 El @ E2 A1 @ Ei @ E2
I
E
1 1 1
3c2
1 1 1 1 1 1 1
4C4
1 w w w 2
4C4’
1 2 w 0 1
I
Reduction of the subduced representation
3
1
T 2
H
3 3
4
0
1
5
0 1 0 0 1 1
0 0
1 1
A1 T T B A1 C T E$E’$T
w
q = exp{i2~/5},
p = q + q 4 =(&1)/2,
= exp{i2~/3},
p1 =7+q3=(&+1)/2.
20. Calculate the reduction of the subduced representation from the regular representation of the improper symmetry group I h of an icosahedron with respect to the subgroup Csi, D5d and T h .
Solution. The I h group is an improper point group of Itype. It is the direct product of the group I and the twoorder inversion group V2 = { E , o}. Any irreducible representation of I h can be expressed as the direct product of two irreducible representations respectively belonging to I and V2. Since the space inversion o is commutable with all elements in I h and its square is equal to the identity, the representation matrix of o in an irreducible representation of Ih is a constant matrix fl. Based on D ( o ) = 1 and 1, the irreducible representations of I h are divided into rg and ru)where r denotes the irreducible representations of I. g and u are the abbreviation of the Germanic gerade and ungerade, respectively. The subgroups C5i, D5d and T h are also the improper point groups of
Theory of Representations
77
Itype, their irreducible representations are the direct product of an irreducible representation of V2 and an irreducible representation of respectively the proper point groups C5, D5 or T. Similarly, based on D ( a ) = 1 and 1 the irreducible representations of the subgroups are also divided , into yg and y, where y denotes the irreducible representations of C5, D5 or T. Therefore, the reduction of the subduced representation from the regular representation of I h with respect to the subgroup C5i, D5d and T h is similar to the reduction of the subduced representation from the regular representation of I with respect to the subgroup C5, D5 and T, where the only difference is that all the representations are duplicated by the subscripts g and u. Remind that the character of each element in the regular representation is vanishing except for the character of the identity which is equal to the order of the group. In the reduction of the subduced representation from the regular representation of a group G of order g with respect to its subgroup H of order h, the multiplicity of an irreducible representation of H is equal to its dimension multiplied by n = g/h. The results of the reduction are as follows. For group Csi,
For group
DSd,
For group
Th,
In all formulas, the series in the bracket is just the reduction series of the regular representation of the subgroup.
21. Calculate the representation matrices of R and s of 1 in its irre6 2 ducible representations by R6 = SIT:SITcl and S12 = ToS1TiRG [see Problem 4 in Chapter 41, where the representation matrices of the generators TOand S are [see Problem 12 in Chapter 41: 1
DT1(To)= diag {q71,q'},
DG(To)= diag{q2,q7q1,q2},
DTZ(To)= diag{q2717q2} DH(To)= diag{q2,q,l,q1,q2},
78
Problems and Solutions an Group Theory
DT'(S1) = 
d3
p
Jz p1
1
p1 p 1 1 p1
J2
p
pl
1 DG(S1) = 
lh
p p1
1
p1 P 1
1
P2 2p1 2P P2 2p' p2
 p2
1 p
2p P2
P ( S 1 )= 
1 5
fi fi
2p
& 5 4
fi
l/G
1,
p2
$ 3
1
fi
p2  2p1 2p' p2
Solution. Any element in I can be expressed by the product of two elements respectively belonging to the subgroup C5 and T: To" (RgS;Sf2), where TOand S1 are two generators of I and R and S2 can be expressed 6 as the product of the generators. In this problem we are going to calculate the representation matrices of R and S2 explicitly. In the calculation 6 we express R and S12as a product of two elements and make use of the 6 property that the representation matrix of To is diagonal. Remind that q = exp{  i 2 n / 5 } satisfies the following useful formulas:
C qa=o,
4 a=O
a+5n
=qa,
1 + 2 ~ + 2 ~ ~ =  1  2 ~ p1 =  q 2  q 3 = (&
p =q
+ q 4 = ( A 1)/2,
+ 1)/2.
For the identical representation, oA(To)= o A ( S 1 ) = D A ( & ) = DA(S12)= 1. The results for other representations are as follows.
Theory of Representations
79
' 0 0 0 0 1 0 0 010 0 0 1 0 0 . 010 0 0 , 1 0 0 0 0
3.4
1
The ClebschGordan Coefficients
If a quantum system consists of two subsystems, the wave function of the total system can be expressed as the combination of the products of the wave functions of two subsystems. Denote by G the common symmetry group of the system and two subsystems. We choose the wave function of each subsystem belonging to a given row of a given irreducible representation of G
*
Then, the wave function for the total system is expressed as their product, transforming according to the direct product representation:
80
Problems and Solutions in Group Theory
The character in the direct product representation is equal to the product of the characters in two representations. The direct product of two irreducible representations is reducible in general, and can be reduced by a similarity transformation [see Eqs. (3.1012)] :
For a finite group or a compact Lie group, all representations in Eq. (3.22) are chosen to be unitary, then Cjk can be chosen to be a unitary matrix with determinant one. The series in the righthand side of Eq. (3.22) is called the ClebschGordan (CG) series, the matrix entries in the similarity transformation matrix Cjk are called the CG coefficients. Rewrite Eq. (3.22) in the form of characters, we may calculate the multiplicities a~ by the orthogonal relation (3.5) of characters:
(3.23)
Substituting Eq. (3.23) into Eq. (3.22), we are able to calculate the CG coefficients CPv,JMr. jk When U J > 1, the parameter T is needed to distinguish these U J irreducible representation D J ( G ) . The method has been demonstrated in section 2 in detail. The total wave function 2) with the given symmetry is combined from the wave functions 2) by the CG coefficients C;:,,,,:
@hT(l, !Pg(l,
P"
(3.24)
In physics, it is convenient to express Eq. (3.24) in the Dirac notations. $ i ( l ) $ i ( 2 ) is denoted by Ij, p ) Ik, v), where the indices j and k are sometimes neglected. @Lr(l, is denoted by (IJ,( T ) , M): 2)
Theory of Representations
81
The action of the symmetric operator PR on the wave function is
First, we choose R to be the generator A whose representation matrices are diagonal such that Eq. (3.26) becomes an eigenequation, and gives some relations of the CG coefficients. Sometimes, the method of projection operators is useful in this step. Then, the CG coefficients can be further determined in terms of the remaining generators. Before completion of the calculation, some undetermined parameters should be chosen arbitrarily.
*
The inverse transformation of Eq. (3.24) is
y (1)$,” contains some functions ! $ (2) (1,2) belonging to different irreducible representation. In terms of the projection operator (3.27) one can pick up from $$(1)$:(2) the part belonging to the M t h row of the irreducible representation D J ( G ) . In fact,
@hr
The projection operator is helpful in the calculation of the ClebschGordan coefficients.
22. Please use the projection operators to reduce the regular representation
of group T. Solution. In the notation for the group T given in Problem 16 of Chapter 2, R1 is a threeorder element. Its eigenvalues are denoted by u p , where w = exp{i2~/3}, = u p . The value of p can be taken as module 3, namely p + 3 is equal to p. Taking the representation matrix of R1 to be diagonal, we can designate the rows and columns of each irreducible representation by the power p of the eigenvalue u p for R1. By making use
82
Problems and Solutions in Group Theory
of the projection operator for the cyclic subgroup of R1, we can calculate the common eigenfunctions iak; for left or rightmultiplying of R1:
where ca is the normalization factor and can be chosen for convenience. Take R(’) = E and c1 = 9 ,
= 9PpEPu= (1 w”p
= 3Spu { E
+ Rid’” + R : w ~ } .
+
+
W P  ~ )
{ E + R1wp
+ R:wp)
Then, taking an element R(2) which does not appear in we have
a,,, , say R(2)= T:, (1)
and a$. One will not Now, all elements in T appear in the functions obtain any other linearly independent function @$) by choosing new R(”). Generally, QfJ does not belong to a given irreducible representation, and can be combined to new basis functions belonging to the irreducible representation by the class operator W = R1 R 2 R3 Rq, which take constant values 4Xj(Rl)/mj (see Problem 8) in an irreducible representation. In the representation A, E , E‘ and T of the group T, the values of W are 4, 4w, 4w2 and 0 , respectively. In fact, for given subscripts p and v , we calculate the matrix form of W in the basis functions and then diagonalize it. Due to completeness of the function bases, we only need to list those terms of E and T; in the calculation
@rJ
+
+ +
(at;,
,
W@FJ= 3wp { E Waf; =
=
SP”WP
+ T:} + { E + T:} w” { 1 + { @FJ+ 3@;J
= wp
{ @?J + 3 Q f J )
W P  ~ }
w”p
+
+  .
1.
When p = v , the matrix form of W in these basis functions is
Theory of Representations
83
The eigenvalues are 4wp and 0, and the transpose of the corresponding eigenvectors respectively are ( 1 / 3 , 1 ) and (1,1). Combining the basis functions according to the eigenvectors, we obtain the irreducible basis functions which respectively belong to representations A, E , E‘ and T .
When p # u , there is only one basis function representation T :
@fJ.
belongs to the
In the calculation, being the eigenfunctions, each basis function can contain an arbitrary constant factor. For a onedimensional representation, this factor is the normalization factor, which is irrelevant to the representation matrix and can be omitted. But for the threedimensional representation T , the choice of the factor is related to the representation matrix. In order to make the representation unitary, we choose the basis functions to be normalized, COO = 1/6 and c10 = c20 = 1/3. In calculating the representation matrix of another generator T with respect to the basis function ’ we only need to list those terms of E , T:, and T:, and notice T:Ty” = T:.
QTo,
QgTg = (3E + 3R1 + 3Rf  T:
S;lb = (7‘;
 Ty” T:  Rq  R3
 R  R:  R  R i } / 6 , 2 g
+ R2 + R: + w2 (T: + R4 + Rg)
+u(Ti+R3+Ri)}/3,
84
Problems and Solutions in Group Theory
*To = {T:
+ R2 + Rz + w (Tz + R4 + Ri) + w2 (Ty”+ R3 + R ; ) }/3.
Thus,
The representation matrices of the generators are
T D
(R1)= (0 w 0 ) ,
0 0 w2
1 0 0
D T
The factors cpY in other basis functions are determined by the representation matrices and the following formula:
qvT; =
c
P
D;p(T,2)qp.
+ R4 + Ri + w ( T i + R3 + Ri) + w2 ( T i + Ra + R i ) }/3, *&= {T: + R4 + Ri + w (Ty”+ R2 + R:) + (T’: + R3 + R:)} /3,
= {T:
d2
= { 3 E  T  Ty” T Z Z
+ w (3R:  R;  R:  R:)
 Rd)}/6,
+ w2 (3R1

R2
 R3
Theory of Representations
85
Usually, the threedimensional irreducible representation T of the group T is calculated by the coordinate transformation method in threedimensional space. The results are as follows:
0 0 1 DT(R1)= ( 1 0 o ) , 0 1 0
DT(T,2>=
( .).
1 0 0 0 1 0 0 1
We can find a similarity transformation X to relate two equivalent representation. For the generators R1,
we have X =
(" ").
a bw2
a bw
cw2
Substituting it into DT(T,")X = X D T ( T 2 ) ,
a a
a+2b+2c 2ab+2c 2a+2bc 2bw2 2cw 2a  bw2 2cw 2a 2bw2  cw 2bw 2cw2 2a  bw 2cw2 2a 2bw  cw2
bw
cw2
+ +
+ +
+ +
+ +
we obtain a = wb = w2c =
m,
x  .  q ; 21 f 1 1
x='(l
A
1 w2 ww2 ) , w 1 1 1
;).
23. Please use the projection operators to reduce the regular representation of group 0.
86
Problems and Solutions an Group Theory
Solution. We follow the notations and formulas given in Problem 16 of Chapter 2 and Problem 12 of this Chapter. Just like the method used in the preceding problem, we first calculate the common eigenfunctions @$ = c, PpR(,)PVof left or rightmultiplying R1 in terms of the projection operators. Letting R ( l )= E and R(2) T:, we have =
There still exist some elements in 0 which do not appear in the functions @$ and Letting R(3) Tz and R(4)= S2, we have =
@pJ.
where 6,(”) is equal to 1 when ( p v) is a multiple of 3, otherwise it is zero. Thus, all elements in 0 appear in the four kinds of basis functions, so they are complete. Now, we will combine the basis functions by making use of the class operator W = Tz T: Ty TZ T, T: such that they belong to the irreducible representations. W takes a constant value 4xj(Rl)/rnj(see Problem 8 ) in an irreducible representation. Through calculation, the constants in the irreducible representations A l , A2, E , TI and T2 of 0 are
+
+
+ +
+ +
Theory of Representations
87
6 , 6, 0, 2 and 2, respectively. In fact, for given subscripts p and v , we calculate the matrix form of W in the basis functions at?, and then diagonalize it. Due to completeness of the basis functions, we only need to list those terms of E , T:, T, and S2. In the calculation of leftmultiplying W on the basis functions the following formulas obtained from the multiplication table will be useful:
@E,
Hence,
When p = v = 0, there are four bases, a::’ @ , ; I , and The matrix form of W in the basis functions and its eigenvectors for the eigenvalues 6, 6, 2 and 2 are
@g).
The corresponding eigenfunctions respectively belong to the representations
88
Problems and Solutions in Group Theory
A l , A2, TI and T2:
Each eigenfunction may contain a constant factor. For onedimensional representation, this constant is the normalized factor, which is irrelevant to the representation matrix and may be omitted. But for twodimensional or threedimensional representations, the choices of factors are related to the explicit forms of the representation matrices. When p = v # 0, there are three bases, a( 1 ) ,,@ ( 2 ) and ,, , The matrix form of W in the basis functions and its eigenvectors for the eigenvalues 0, 2 and 2 are
@rJ.
The corresponding eigenfunctions respectively belong to the representation
Theory of Representations
89
E , TI and T2:
q,= a,, { $ 9 3 + QfJ} = a,, ( E + T i + T l + T; + w,
+w,(R;+R;+R;+R:)},
QT1
(R1+ R2
+ R3 + R4)
, = b, { @FJ +fJ, ,
=
bp,
2WWt3,)
( 3 E  T:  Ty” T: (3R1  R2
 R3
 25’1
1
 2S3  2S5  2Tz)
+
LJ’
 R4
 2T:  2’ T
+ u p(3R:  Ri  Ri  Ri 2Tz 2Ty  2 T z ) } ,
“2= c,,
{! a;

= cp, ( 3 E  T:  T i
+ w’
+ 2S1 + 2S3 + 2S5 (3R1  R2  R3  R4 + 2T: + 2Ty”+ 2T:)
 T:
@pi + 2 W W f J 1
When p # v and p v = 3, there are three bases, af, @f; and ipp!. ; The matrix form of W in the basis functions and its eigenvectors for the eigenvalues 0, 2 and 2 are
+
0 13
:),
(i). (!J (3)
The corresponding eigenfunctions respectively belong to the representation E , TI and T2:
90
Problems and Solutions in Group Theory
(2@$  @f;
+a : ; }
+ 2Rg  Tz  T:  5’1 + 35‘2 + (2Ti + 2R4 + 2Ri  Ty  Ty” 5’5 + 3S6) + (2T: + 2R2 + 2Ri  Tx  T:  S3 + 3 5 4 ) ) .
(2T: + 2R3
W’
W ’
When p = 0 and v # 0, there are two bases, @$ and Q O3u . The matrix ( ) form form of W in the basis functions and its eigenvectors for the eigenvalues 2 and 2 are
The corresponding eigenfunctions respectively belong to the representation TI and T2:
;* :
{@g) = bOv {T: + R4 + R  T  Ty” S Z X 1 + o. (TZ + R2 + R:  T,  T:  S,)
= bOv
 L P @ g ) }
+ d (Tz + R3
!Pov  cov T 2
+ Rg  3”’
 T:  S 3 ) } ,
{@I;“? + 1 = coU {T: + R4 + R + Tx+ Ty”+ Z + w’ ( T i + R2 + R: + T, + T: + 55) + w v (T: + R3 + Ri + Ty + 3”: + S3)} .
W,@g)
S 1
When u = 0 and p # 0, there are two bases, @$) and @$I. The matrix form of W in the basis functions and its eigenvectors for the eigenvalues 2 and 2 are
(  2 L
2w’) 0
,
(;’)
( ’ )* u’
The corresponding eigenfunctions respectively belong to the representation
TI and
7’2:
Theory of Representations
91
+up
(Tl+R3 + Ri + Tz + S5 + T’)}
.
Now, we calculate the representation matrix of the generator T,. The representation matrix of the generator R1 is known. First, for the repreto make sentation E , we take the normalized bases, a l l = a21 = the representation unitary,
,/m,
We only need to list the terms of E and T,:
Hence, the representation matrices are
According to these representation matrices we can calculate the remaining factors a12 = a22 =
dm:
92
Problems and Solutions in Group Theory
In comparison with the representation matrices calculated in Problem 12 with the method of the quotient group,
the similarity transformation X satisfying X  ’ D E ( R ) X = E ( .R )is D
d@
Second, for the representation T I ,we take the normalized bases, boo = and blo = b20 =
dm,
A: *
= d m ( 3 E + 3 R 1 +3Rf  Tz T:  TZ  R2  R:  R3  R  R  R g 4 Z Tx T: + Ty Ty” Tz T: S1 S3 S  3S2  3Sq  3Sc} , 5 Q; = T {TZ R2 Ri  Ty  T:  S1 w2 (Ti + R + Ri  Tz T ” 5 ’ 3 ) 4 y+ W (T2 R3 R  Tx Tz  S5)} g = dWy{T; R 2 Ri  Ty  T:  S1 w (Ti R 4 R  Tz  Ty” 5 3 ) g w2 ( T i + & + R: Tx  T:  S 5 ) } .
+ + + + + + + d m + + + + + + + + + +
+
)
+
We only need to list the terms of E , T: and T:. Notice that TzS2 = SIT, =
T; .
Hence, the representation matrices are
Ti D
(R1)= ( 0 w 0 ) , o T ’ ( T z ) = (  2 w
0 0 w2
1 0 0
1
2w 2w2
y2).
2w
2w2
1
They are symmetric. The factors in the remaining basis functions can be
Theory of Representations
93
calculated by the following formulas:
SPFhT, =
1 J18 {WE  w2T;  T: +  . } = 1 { 9;;  2 ~ 9 ; ;  2w29:;},
Thus,
=
d m { T j+ R4 + R;  Tx Ty”  S1 + w2 (TZ + R2 + Ri  T,  TZ  S5)
+W
+ + Ri  Ty  T;  S 3 ) } , Q r i = d m { T : + R4 + R;  Tx  T i  S1 + w ( T i + R2 + Ri  T,  T:  S5)
( T i R3
9; = :
+ w2 (T: + R3 + Rz  Ty  T:  S 3 ) } , d m {3E  T:  T;  T;  2S1  2S3  2S5
+ w 2 (3R1  R2  R3  R  2T:  2T;  2T:) 4
+W
(3R:  R;
 Ri
 Ri  2Tx  2Ty  2 T z ) } ,
QTl 12 
d m { 2T; + 2R3 + 2Rg + T, + T: + S1  32 s LJ2 (2Ty” 2R4 + 2R; + Ty + Ty”+ Ss  3S6) + + w (2T: + 2R2 + 2R,”+ Tx + T: + S3  3S4)}, = I/m (2T: + 2R3 + 2Ri + T, + T: + S  32 1 s (2Ty” 2R4 + 2Ri + Ty + Ty”+ S  356) 45 + w 2 (2T: + 2R2 + ZR,”+ Tx + T: + S3  3S4)},
W
9 ; = I / m { 3 E  T:  T i  T: ;
 2S1  2S3  2S5
+ w ~ ( ~ R ~  R ~  R $  R ~  ~ T ~ 
94
Problems and Solutions in Group Theory
In comparison with the representation matrices calculated in Problem 12 with the method of coordinate transformation in threedimensional space,
0 0 1 DT1(R1)= (1 0 0 ) , 0 1 0
DT1(Tz) =
(:
1 0 0 0). 0 0 1
The similarity transformation Y , satisfying YlDT1 (R)Y = BT1R )is (
Y=
k(l
1 w2 w 1 w w2 1
1).
1 1 1
Yl=%(:.
w,’ ;)@%
At last, for the representation T2, if we choose cpY = b p Y , the coefficients in the expansions of the basis functions are the same as that of except for the terms related to the elements in the coset of the subgroup T, which change signs. So, the representation matrix DT2((R) just the is direct product of D T 1 ( R )and D A 2 ( R ) .
@%
24. The multiplication table of the group G is given as follows.
\right left \
E E A B C F K M N
A A E N K M C F B
B B M E N K F A C
C
F
K K C F A B E N M
M M B C F A N K E
N N F A B C M E K
E A B
C
F K M N
C K M E N A B F
F N K M E B C A
a) Write the character table for the group G. b) If we know that the representation matrices of the generators A and B in a twodimensional irreducible representation D are
and two sets of basis functions qp and @Y respectively transform according to this twodimensional representation of G:
Theory of Representations
95
combine the product function q!~~$,, such that the basis function belongs to the irreducible representation of G. Solution. The group G consists of 8 elements, where the order of the identity E is one, the orders of A, B , C, F and K are two, and the orders of M and N are four. E and K are commutable with any element in G. G contains five classes, { E } , { K } , { A , C}, { B , F } and { M , N } . They are all selfreciprocal classes. Since l1+ l2 l2 l2 22 = 8, .the group G has four onedimensional and one twodimensional inequivalent irreducible representations. These representations all are selfconjugate. There are four invariant subgroups in G. The cosets of the invariant subgroup { E , K } are { A , C}, { B , F } and { M , N } . Its quotient group is isomorphic onto the group Vq. The indices of the remaining three invariant subgroups, { E , A, K , C } , { E , B , K , F } and { E , M , K , N } all are two. We can obtain four onedimensional representations of G from these quotient groups. The characters in the twodimensional representation can be obtained by Eqs. (3.56). The character table of G is as follows. The characters for the selfdirect product of the twodimensional representation are also listed in the table.
+ + +
$ $ x
x3
1 1 1 1 2
K 1 1 1 1 2
( A n
1 1 1 1 0 0
4
4
( B , F ) (M,W 1 1 1 1 1 1 1 1 0 0 0 0
From Eqs. (3.1011), one obtains:
Substituting the matrix forms of A and B into the above formula, one has
1 0
Xl(; 0 0 1  1 0 0
;
0 0
1 0 0
0
)
0 0  O 00 ) 1 1 0 0 01
Xl(O 0 0 0 o ) 1 1
0 1 0 0 1 0 0 0
x = ( o 1l o0
0 0 )
0 0 1 0 0 0 01

96
Problems and Solutions in Group Theory
The eigenvectors of the product matrices for the eigenvalues f l are: eigenvalue
1 1 1
1
Thus, the similarity transformation matrix X is
x=( 1 Jz
0
1 1 0 0 0 1 1 0 0 11
)

110
0
The new basis functions belonging to the irreducible representations are calculated by X:
*(l)
= ($141
= ($142
+ $ 2 4 2 ) /a, + $ 2 4 1 ) /Jz,
!P(2) = ( $ 1 4 1
*1(4)
 $242)
/Jz,
/4.
= ($142  $241)
25. Calculate the unitary similarity transformation matrix X for reduc
ing the selfdirect product of the threedimensional irreducible unitary representation DT of the group T:
xl { D T ( R )x D T ( R ) }x =
c
UjDj(R>.
Solution. This problem will be solved by two methods. One is to calculate the similarity transformation matrix directly. Another is to calculate the expansions of states in the product space. These two methods are the same in principle. However, when the dimension of the representation is quite large, the first method may not be convenient. The aim of this problem is to show the relation between these two methods, and to demonstrate the merit of the second method. To begin with, we solve this problem with the first method. Take the generators of the group T to be T: and R1. Their representation matrices in the threedimensional representation T are
DT(T,”)=
(
1 0 0 0 1 0 ) 0 0 1
,
DT(R1)
=
(; :i)
0 1 0
.
Theory of Representations
97
Denote by D ( R ) the selfdirect product representation of DT,
The multiplicities
aj
can be calculated with the character formula (3.11):
Namely, X'D(R)X = D A ( R )@ D E ( R )@ D E ' ( R )@ D T ( R )@ D T ( R ) . Let R = T; in the formula of the similarity transformation.
D(T:) = diag {1,1,1,1,1, 1, 1, 1,l)
XlD(T;)X
= diag { l , l , l , 1,1,1,
1,1,l).
We conclude that for the lst, 2nd, 3rd, 6th and 9th columns of X , the matrix entries in the 3rd, 6th, 7th and 8th rows are zero, while for the 4th, 5th, 7th and 8th columns, the matrix entries in the lst, 2nd, 4th, 5th and 9th rows are zero. Then, let R = R1. The action of DT(R1) is to cycle the matrix entries in three rows as 1 + 2 + 3 + 1. In the direct product space, the rows (columns) can be designated by two indices ab, and the action of D(R1) is to change the 12th component in the vector to the 23th position, etc. If we change the designation of the rows (columns) by one index (u running from 1 to 9, corresponding to the order of the indices ab as 11, 12, 13, 21, 22, 23, 31, 32, 33, then the action of D(R1) is
11 + 22 15
+33 + 12 +23 + 31 + 12, 11, + 9  + l , 26  7 2, i 13 + 21 + 32 + 13,
34
+ 8 3.
The matrix after the similarity transformation is
98
Problems and Solutions in Group Theory
‘10 0 0 0 0 o w 0 0 0 0 o o w 2 0 0 0 0 0 0 0 0 1 XlD(R1)X = 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 ,oo 0 0 0 0
0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 1 0 0
where w = exp{i2~/3}. W 3 have known that the thirG and the 6t il matrix entries in each of the first three columns of X are zero. The three column matrices are also the eigenvectors of D(R1). From the above cyclic property, their matrix entries in the 2nd, 4th) 7th and 8th rows must also be zero, namely only the matrix entries in the lst, 5th and 9th rows can be nonzero. The values of those matrix entries can be determined from the eigenequation:
From the unitary property of X , for the 6th and 9th columns, only the matrix entries in the 2nd and 4th rows may be nonvanishing. Since the multiplicity of the representation T in the reduction of the direct product representation is two, we can simply choose the matrix entries in the 6th and 9th columns to be zero except for x 6 = X49 = 1. Substituting them 2 into the formula for the similarity transformation on D(R1), one obtains
D(R1)X.g = X.7,
The solutions are
x4 6
D(R1)X.7 = x.8.
= x 5 = x 7 = x 8 = 1, 7 8 3
and the remaining matrix entries all are zero. Finally, the matrix X is
Theory of Representations
99
’mm~
0
0 0
0 0
0 0
0
0 0
w”J3
0 0 0 0 0 0 1 0 0 0
0
X=
0 0 0
0 0 0 0 1 0 0 0 0 0 0 1 w/J3 0 0 0 0 0 0
0
0
0 0
,m
0
u/&
0
w2/&
1 0 0 0
0 1 0 0
0 0 0 0
0 0 1 0
0 0 0 0
0 0 0 0
It can be checked that X does satisfy the similarity transformation formulas. Now we turn to the second method. For convenience we use the Dirac notation to express the basis state in the representation space before and after the similarity transformation. The representation space before the transformation is the direct product of two subspaces, and the basis state is denoted by [ T , p ) l T , v ) .Sometimes, two T in this basis state can be omitted. The actions of the generators T: and R1 on the basis state are:
The representation space after the transformation is the direct sum of five subspaces, where the basis state is respectively denoted by IIA), llE), IlE’), llT, ( l ) , p ) and llT, ( 2 ) , p ) , p is 1, 2 or 3, satisfying
where the index T = 1 or 2 is used to identify two representations T . The basis state which is invariant in the action of T: must be in the following form:
The basis state which changes its sign in the action of T; must be in the
100
Problems and Solutions in Group Theory
following form:
Applying R1 to the basis state IIA), we have
By comparison, we obtain cf = c t = 0, and cf = cf = cf. After normalization, cf = and the normalized basis state is
0,
Applying
R 1
to the basis state IIE), we have
By comparison, we have cf = cf = 0, and cf = cfw = c f u 2 . Taking cf = we obtain the normalized basis state
0,
Similarly, we have
Applying R1 to the basis state llT, (r),3),we have
Theory of Representations
101
By comparison, we have and
We may take
Applying R1 to them, we have
These results are consistent with that obtained by the first method.
26. Calculate the CG series and the CG coefficients for the direct product
representation of each pair of two irreducible representations of the group 0.
Solution. In Problem 17, we have calculated the representation matrices of the generators T, and R1 of the group 0 in five inequivalent irreducible 1 ' representations A , B, E , 2 and Tz. Now take a similarity transformation to diagonalize the representation matrices of R1:
DT1(Tz)= DT2(Tz) = 3
DT1(R1) =DT2(R1) =
2w2 2w 1 2w 1 2u2) 1 2w2 2w
,
(;
0 1
0
w:l)
,
where w = exp{ i27r/3}. Since the representation matrix of R1 is diagonal, and the diagonal elements wm are different in each irreducible representation, this power m can be used to designate the row (column) in each irreducible representation. Notice that the value of the power m is module 3, m 3 = m. The representations A and B are onedimensional, so that the index 0 for the row (column) can be omitted. The index for the row (column) in the representation E takes the values 1 and 1. The index in the representations 571 and T takes the values 1, 0 and 1. 2
+
102
Problems and Solutions in Group Theory
The direct product of the identical representation A and any representation is still equal to that representation. The direct product of the antisymmetric representation B and any representation is easy to calculate:
where 0 3 is the Pauli matrix. The ClebschGordan series for the direct product representations is listed in the Table, and the ClebschGordan coefficients are calculated in the following.
T h e representation of 0
II E 3T: 6Tz 8R1 A 1 1 1 1 1 B 1 1 1 1 E 2 2 0 1 3 1 1 0 Tl 3 1 1 0 T2 4 4 0 1 ExE 6  2 0 0 E X Ti N_ E x T2 1 0 T ~ x T ~ E T ~ x 9T ~ 1 9 1 1 0 Tl x T2
6S1
I
I
T h e reduction
1 1 1 0 1 1 0 A@B@E 0 Tl @ T 2 1 A@E@Tl@Tz 1 B@E@Tz@Tl
(1)ExEzAeBeE. Since the representation matrices of R1 are all diagonal, the eigenvalue of R1 for the product state is the product of the eigenvalues of two states, i.e., the sum of rn‘s. Thus, we have:
Applying T, to them, we have
So, a1 = u2,
bl
= b2, and
c1
= c2. After normalization we have
Theory of Representations
103
(2) E x Ti N E C T2. B From the eigenvalues of R1, we have:
Applying Tz to them, we have
The solution is have
a1
= b2 = c2 = a2 = c1 = bl. After normalization we
For the representation of T2, we obtain a1 = b2 = c2 = by the similar calculation. After normalization we have
a2
= c1 = bl
T 2
For the CG coefficients in E x T2, one only needs to exchange 12 and ' in the above results. Namely,
104
Problems and Solutions an Group Theory
(3) Ti x Ti N T2 x T2 N A @ E @ Ti @ 2'2. From the eigenvalues of R1, we have:
Applying T, to them, we have
Thus, a1 = a2 = a3, bl = normalization we have
b2
=
b3
=
b4
= b
5 
be.
After
The states belonging to the representation TI must be orthogonal to the above states,
Applying T, to them, we have
Theory of Representations
105
Thus,
After normalization we have
From the orthogonality of states we have
Applying T, to them, we have
Thus, dl = & = d3. After normalization we have
106
Problems and Solutions in Group Theory
For the case of T x T2, the term lT1,p ) 12’1, v) in the expansion is replaced 2 with JT2, p)JT2, and the coefficients are exactly the same. Y), (4)
571
x T “B$E$T2$Tl. 2
Multiplying the antisymmetric representation B on the formula in the case (3), we obtain the formula for the present case. Note that the similarity transformation o3 appearing in the direct product of B and E. The results are as follows:
27. Calculate the CG series and CG coefficients in the reduction of direct
product representation in terms of the character table of the group I given in Problem 18: (1) D T ~ x DT1; (2) DT1 x DT2; (3) DT2 x DT2; (4) DT1 X DG; ( 5 ) D*2 x DG; (6) DT1 x D H ; ( 7 ) DTz x D H ; (8) DG X DG;
(9) DG x D H ; (10) D H x D H .
Solution. We will only list the results for this problem. Notice that the character for the direct product representation is equal to the product of characters of two representations. In Problem 23 the representation matrices of the generators T and S1 in each irreducible representation of the group I were given, where the representation matrix of T is diagonal. Since the order of T is five, the diagonal element of D ( T ) takes the power qm where q = exp{i2~/5). The power m is different in each irreducible representation of I, so this power rn can be used to designate the row (column) in each irreducible representation. The possible values of the power in each representation are as follows: 0 for the representation A , 1, 0 and 1 for the representation T I ,
Theory of Representations
107
2,O and 2 for the representation T 2 , 2 , 1 , 1 and 2 for the representation G, and 2, 1, 0, 1 and 2 for the representation H . Note that the power m is module 5 , m 5 = m.
+
The reductions of the direct product representations in 1
T h e reduction
A
Ti
T 2
1
3
G H
Ti x Ti Ti x T2 T x T 2 2 Ti x G T x G 2 Ti x H T x H 2 GxG G x H H x H
3 4 5 9 9 9 12
12
1 p1 p
1
0
P2
15
15
1 p2 p1 p 0
0
16 20 25
1
0 0
1 p p1 1 0 P2 1 p2 P p1 0 0 1 0 0
1
1
1 0 1 1
1
1 0 0 1 1 0 0 1
1 0 0 1 1 0 0 0 0 0 0 0 1 1
1
(1) DT1 x DT1 Y D A @ DT1 @ D H .
108
Problems and Solutions in Group Theory
(3) DT2 x DT2
fi D A @
DT2 @ D H .
(4) DT1 x DG N DT2 @ DG @ D H .
Theory of Representations
109
110
Problems and Solutions in Group Theory
Theory of Representations
111
(8) DG x DG
2 :
D A @ DT1 @ D T 2 $ D G $ D H .
112
Problems and Solutions in Group Theory
(9) DG x DH 2 DT1@ DT2@ DG @ 2DH.
Theory of Representations
113
(10) DH x DH E DA @ DT1 @ DT2 @ 2DG @ 2DH.
114
Problems and Solutions in Group Theory
Chapter 4
THREEDIMENSIONAL ROTATION GROUP
4.1
SO(3) Group and Its Covering Group SU(2)
In the real threedimensional space, a spatial rotation which keeps both the position of the origin and the distance of any two points invariant:
*
(i)
) : (
R11 R12 R13
"3)
= (R21 R22
R31 R32 R33
(zi)
,
 =R x ' ,:
I T J:I  gTg,
RTR = 1,
R* = R ,
is described by a real orthogonal matrix R. R is a proper rotation if detR = 1, and an improper rotation if detR = 1. The set of all threedimensional real orthogonal matrices with detR = 1, in the multiplication rule of matrices, satisfies the four axioms and forms the simplest nonAbelian compact Lie group SO(3). It is a very important Lie group from the viewpoints of both physics and mathematics. The element R ( n , w ) in SO(3) is a rotation around the direction n through the angle w , satisfying
R(n,w
+ 27r) = R(n, w ) = R(n,
R(n,x) = R(n,r).
27r
 w),
(44
When n is along the coordinate axis, the rotation matrix can be written in the exponential form of matrix:
0
0 sinw
0
cosw
115
116
Problems and Solutions in Group Theory
R(e2,u) =
(
cosw 0 sinw 0 1 0 sinw 0 cosw
cosw sinw 0 R(e3,w) = (sinow co;w where
) :)
=exp{iwT2},
= exp{iwT3},
0 i 0
oi
0
i 0 0
( T a ) ) ~ d iCabd. =
0 0 0
(44
They are the generators in the selfrepresentation of S0(3), satisfying the typical commutation relations for the angular momentum operators
*
The rotation S(cp, 0)
transforms
e3
to the direction n(0,cp). Since
The rotation R ( n , w ) around n through w can be expressed as
R(n, W) = S(cp,0)R(e3,~)S(cp, = exp {  i d 0)'
 T} .
(4.5)
Thus, the rotations with the same rotational angle LJ are conjugate to each other. The class of SO(3) is described by the rotational angle w. The SO(3) group is a simple Lie group because it does not contain any nontrivial invariant Lie subgroup. Define a vector 2 , which is along the direction n with the length w . Let the polar angle and the azimuthal angle of n be 0 and cp, respectively. R(n,w) can be described by the spherical coordinates (u,O,cp) of i3, or its rectangular coordinates ( ~ 1w2, wQ), which are the group parameters of , SO(3). The varied region of the group parameters is called the group space. The group space of SO(3) is a spheroid with radius T ,where in the sphere,
Threedimensional Rotation Group
117
two end points of a diameter describe the same rotation. Namely, the group space of SO(3) is a doublyconnected closed region. It is the reason why SO (3) has the doublevalued representations. The set of all twodimensional unitary matrices u with det u = 1, in the multiplication rule of matrices, form the Lie group SU(2). Its element can be generally expressed by
u(n,w ) = 1 cos(w/2)  i(3 n) sin(w/2)
*

=
(
cos(w/2)  i sin(w/2) cos e i sin(w/2) sin Be+ i sin(w/2) sin cos(w/2) i sin(w/2) cos e
+
)
7
(4*6)
satisfying u(h,wl)u(fii, wz) = u(n,w1
u(n,47r) = 1,
.(n,w)
+ w2),
(4.7)
u(ii,27r) = 1,
= u(ii,47r  w ) = u(ii,27r
 w).
Choosing the group parameters of SU(2) to be the vector G along the direction n(6,cp) with the length w , we obtain the group space of SU(2) to be a spheroid with radius 27r, where all points on the sphere describe the same element 1. The group space for SU(2) is a simplyconnected closed region, and the SU(2) group is a compact simple Lie group. The generators in the selfrepresentation of SU(2) are a,/2. There is a twotoone correspondence between Au(n, w ) E SU(2) and R(n, w ) ~ s O ( 3 ) :
u(n,w ) a a u ( n , w )  l =
c
3
abRba(n, ) . w
(4.8)
b= 1
So, the SU(2) group is the covering group of the SO(3) group. The subset of u ( n , w ) with the same w forms a class of SU(2).
1. Prove the preliminary formula by induction:
=
P + g [a, P] + +% [a, [a, PI] + *
1
1
*
where a! and P are two matrices with the same dimension. Then, show Eq. (4.8) and prove that SU(2) is homomorphic onto SO(3).
118
Problems and Solutions in Group Theory
Solution.
We are going to show by induction that the expression in the brackets is equal the following commutator with nmultiplicity
The above formula is true for n = 1 obviously. Now, if it is true for n  1, we will show that it is also true for n.
n
n1
nI
= o x
m=l
(1)""(n  l)!aml Panm (m  l)!(n  m)!
= anp
+
n1
c
m=l
(1)""(n  l)! [m (m)!(n  m)!
+ ( n m)] ampan" + (I)npan
= m=O
C
(l)n"n! ampQnm. m!(nm)!
This completes the proof. Further, due to (Tc)ba i E c a b , =
21
[Z  ii, a,] = i
C ncEcabOb= C
bc
3
06
(ii  T ) , ~ ~
b=l
b= 1
Repeating the commutation relation by n times, we have
n
b= 1
Threedimensional Rotation Group
119
By making use of the preliminary formula and
u(ii, w ) = exp{ iwa'
 n/2},
R(ii,w ) = exp (  i w n  T) ,
we obtain
Now, Eq. (4.8) is proved. Any u ( n , w ) E S U ( 2 ) can uniquely determine a matrix R ( n , w ) E SO(3) by Eq. (4.8). If both u1 E S U ( 2 ) and u2 E SU(2) determine the same R E SO(3) by Eq. (4.8), u,'ul can commute with three Pauli matrices g a , so it has to be a constant matrix. Due to their determinants, we have u1 = fu2. Therefore, Eq. (4.8) gives a twotoone correspondence between &u(ii,u)E SU(2) and R ( n , w ) E SO(3), which is obviously invariant in the product of elements. It means that SU(2) is homomorphic onto SO(3).
2. Expand R(n, w ) = exp (  i w n . T) as a sum of matrices with the finite
terms.
Hint: (n  T)3= n T.
0
Solution. From
one has
R(n, w ) = exp (  i w n
 T)
120
Problems and Solutions in Group Theory
This problem can also be solved with the method of similarity transformation:
ii . T = ST3Sl,
S = R(e3, cp)R(e2, ) , O
where 8 and cp are the polar angle and azimuthal angle of the direction n(O,cp), and S is the rotation transforming the X 3 axis to the direction A. Since R(e3,w ) can be expanded in the following form
R(e3,w ) = exp (iwT3) = 1  Tl + T: cos w  iT3 sin w ,
we have
R ( n ,w ) = SR(e3,w)Sl = 1  ST?SP1+ STzSl cos w  iST3Sl sin w = 1  (ii  T ) + (ii. T ) COW  i (ii.T) sinw. ~ ~
3. Check the following formulas in the group 0 [see the notation given in Problem 12 of Chapter 3 in terms of the homomorphism of S U ( 2 ) onto 1
SO(3): TzR1 = S3,
TzTx= R1,
RlR2 = Ri.
Solution. Since the elements in S U ( 2 ) are twodimensional matrices, the homomorphism of S U ( 2 ) onto SO(3) can be used to simplify the calculation
for the multiplication of elements in SO(3). The elements TZ, R1, R2, Tx, R and S3 in the group 0 correspond to those elements in the group S U ( 2 ) ) Z where a sign in u ( R ) is irrelevant to us:
Threedimensional Rotation Group
121
Now, replacing the elements R of 0 in the multiplication formulas with the corresponding elements u(R) of SU(2), we are able to check whether the multiplication formulas are correct up to a sign. u V Z ) u ( ~ 1 ) m 13 ) (1/2) {I  i (01 = < i 0= 4 7 5 (  i 0 2  i03) = 4 ~ 3 1 ,
U(T,)U(T,)
+ 0 2 + 03))
= m 1i 0 3 ) ( 
= (1/2) (1
i (01 + 0 2
+
m{ 1
 ia1)
03))
= u(R1),
U(Rl)U(R2) =
(1/2) (1  i (01
= (1/2) (1  i (a1
+ 0 2  0 3 ) ) = u(R$).
+ 0 2 + 0 3 ) ) (1/2) (1  i (01  0 2  0 3 ) )
4. Prove the following formulas for the group I in terms of the homomor
phism of SU(2) onto SO(3)
R6 = s l T i s l T r l ,
s12
= ToSiTiR6,
which were used in Problem 21 of Chapter 3.
Solution. According to the notations given in Problem 18 of Chapter 3, we are able to calculate the matrices u ( R ) in SU(2) which corresponds to elements of I, where a sign in u ( R ) is irrelevant to us.
where the polar angle and azimuthal angle of n are 03 and n/5, and the polar angle and azimuthal angle of m are 0 4 and zero, tan 03 = 3 &, and tan04 = (&  1)/2. Thus,
+
122
Problems and Solutions in Group Theory
Let A = S (T,SITcl) and B = (T:S1Tc1) S 1 2 . We are going to calculate 1 the matrix entries of u(A) and u ( B ) to show that they are both equal to u(&) . Because
C O S ( ~ T / ~= )

6  1 4 '
= = sin(2~/5) sin(3~/5)
we have

  [(lo 1 20
+ 2 d q 7& + l  (10  2 h )  4 5  1
4

1
+
i
40I/ic%
2
[(lo
+ 2&)(5
 &)
 (10  2&)2&]
where
d
m
m = 3&
 5 is used.
Threedimensional Rotation Group
123
4.2
Inequivalent and Irreducible Representat ions
In the threedimensional space, any rotation can be expressed as a product of three rotations around the coordinate axes
*
, where ca = C O S ~ s, = s i n a , etc. Three angles a , /? and y are called the Euler angles. There are two methods for calculating the Euler angles. One is based on the matrix form of R. The third column of R is a unit vector, where its polar angle is p and its azimuthal angle is a. The third row of R is also a unit vector, where its polar angle is /? and its azimuthal angle is ( T  y ) . The other is based on the relative position of the coordinate frames before and after the rotation R. If R transforms the coordinate frame K to K‘, then in the K coordinate frame, the polar angle and the azimuthal angle of the 2 axis of K‘ are /? and a , respectively. In the K‘ coordinate ‘ frame, the polar angle and the azimuthal angle of the 2 axis of K are ,8 and (7r  7 ) . The domain of definition for the Euler angles in SO(3) are
nSaQn,
o<p57r,
n<y<7r.
(4.10)
For the SU(2) group, 27r y 5 27r. When /? = 0 or n,only one parameter between a and y is independent. For the SU(2) group, its inequivalent irreducible representation can be designated by a halfinteger j = 0, 1/2, 1, 3/2, . . .,
<
*
=
*
(1)n
+ v)!(j  v)!(j + p ) ! ( j  p ) ! } 1 ’ 2 ( j + v  n ) ! ( j p  n)!n!(n v + p ) ! {(j
{cos(w/2)}2j+Y’2n { ~ i n ( w / 2 ) } ~ ~  ”, ” +
O
VP
n :
,ax(
),
e
m

,
r n i n ( j + ’p.  )
(4.11)
124
Problems and Solutions in Group Theory
The matrix d j ( w ) satisfies the following symmetric relations:
(l)P”d&&)
= d$JbJ) = dj,,(w) = dt”P(Ld))
d i p @ ) = (l)~P&/(p))
d” P (7r  w ) = (l)jW:&J) j
dj,,(27r) = (l)”ja,,,
= (l)j+”dj”,(W).
(4.12)
The character of the element u(n,w) in the representation Dj is
*
D j is a unitary representation of SU(2) with dimension (2j + 1). When
j is an integer, D j is a singlevalued representation of SO(3) and a non
faithful representation of SU(2). It is equivalent to a real representation by a similarity transformation. When j is a halfoddinteger, Dj is a doublevalued representation of SO(3) and a faithful representation of SU(2). It is a selfconjugate representation, but cannot be transformed into a real representation by a similarity transformation. Do(.) = 1 is the identical representation, D1I2(u) = u is the selfrepresentation of the SU(2) group. D 1 ( u )is equivalent to the selfrepresentation of the SO(3) group:
M’R(a, p, Y ) M = D’(a, p, 7)) M = The generators in the representation
Dj
’
(1;
‘i) O J Z O
.
(4.14)
are
(4.15)
Letting c = cos(p/2) and s = sin(P/2), we have
do(@)
= 1)
d”“(a)
=
( “) ,
s c
d’(@ =
i
c2 Jzcs
s2
Jzcs s2 c2  s2 JZcs Jzcs C2
)
,
Threedimensional Rotation Group
125
5 . Calculate the Euler angles for the following transformation matrices R,
S and T , and write their representation matrices in Dj of SO(3):
a)
~ ( a , p , r )1= ;
Jz
b)
J2
3 a  2
2d3
2&
& + 2 6 Jz6 2J2
&+2d3
2&
'2Jz
4Jz
Solution. a) From the third column of R, we have
m, s i n a = ( f i / 4 )  2 = m,
C O S = ~
(J2/4)
*
2=
a = 3n/4.
From the third row of R, we have
126
Problems and Solutions in Group Theory
Hence,
b) From the third column of S, we have
c o s a = (@4) s i n a = (Jz/4)
 Jz = &/2,
 Jz = 1/2,
a = 7r/6.
From the third row of S, we have
Hence,
c) From the third column or the third row of T , we have p = 7r, but a and p are undetermined. In fact, when ,f3 = 7 r , only ( a  y) is independent. For definiteness, we may choose a = 0 and calculate R(e2,7r)lT
= 1
Hence,
(
A
?i)
1
0
=R(e3,57r/6).
Threedimensional Rotation Group
127
6. Calculate the Euler angles for the following rotations R, S and T , and write their representation matrices in Dj of SO(3). a) R is a rotation around the direction n = el sin8 e3 cos8 through an acute angle 8; b) S is a rotation around the direction n = (el e2 e s ) /& through 27r/3; c) T is a rotation around the direction n = (el e2) /fi through 7r.
+ + +
+
Solution. If a rotation R transforms the coordinate frame K to the frame K ' , the polar angle and the azimuthal angle of the 2' axis of K' in the coordinate frame K are p and a , respectively. In K', the polar angle and ) the azimuthal angle of the 2 axis of K are /3 and (i7  y , respectively. a) The X' axis and the 2' axis of K' are both in the X Z plane of K , and the angle between two Z axes is 28, so p = 28 and a = 7r  y = 0,
b) The X', and 2' axes of K' are located in the Y , 2 and X axes of Y' K , respectively, so p = 7r/2, a = 0 and 7r  y = 7r/2,
c) The 2 axes of K' is along the negative 2 axis of K , so that p = 7r ' and only ( a  7)is independent. Since the Y' axis of K' coincides with the X axis of K , we can choose y = 0 and obtain a = 7r/2.
7. Calculate the Euler angles for the rotations TO,T2, R1, R2, R6, S1, S2,
$3, 5 ' 1 1 , and S12 in the icosahedron group I, where the notations for the elements are given in Problem 18 of Chapter 3.
Solution. The key for solving this problem is to determine the directions of the coordinate axes after rotation. The Euler angles of TOare obviously /3 = 0 and a y = 27r/5. One may choose a = 0 for definiteness. After the rotation T2, the 2 axis is rotated to the direction O A l , so we have a = 0 and p = 81. The Y axis is rotated to the direction pointing from 0 to the central point D1 of the edge A2A3. The angle between two planes AlOAo and AlODl is 37r/10. So, 7ry = n/2+37r/10, and y = 7r/5. After the rotation R1, the 2 axis is rotated to the direction of O A l , so we have cy = 0 and ,8 = 81. The Y axis is rotated to the direction pointing
+
128
Problems and Solutions a Group Theory n
from the origin 0 to the central point 0 2 of the edge AoA4. The angle between two planes AlOAo and A 1 0 D 2 is 7r/10. SO,7r  7 = 7r/2  7r/10, and y = 37r/5. After the rotation R2, the 2 axis is rotated to the direction of OA2, so a = 27r/5 and ,f3 = 81. The Y axis is rotated to the direction pointing from 0 to the central point 0 3 of the edge A3A4. The angle between two planes A2OAo and A 2 0 0 3 is 3n/10. So, n  y = n/2 3n/10, and y = n/5. After the rotation Rg, the 2 axis is rotated to the direction OB3, so we have Q = n/5 and p = 285. The Y axis is rotated to the direction pointing from 0 to the central point Q of the edge AoAl. The angle between two planes B30Ao and B3OQ is 7r/10. So, n  y = n/2 n/10, and y = 2n/5. From the result of the preceding problem, the Euler angles for S1 are p = 284, a = 0 and y = n. After the rotation S2, the 2 axis is rotated to the direction OA2, so we have a = 2n/5 and p = 81. The Y axis is rotated to the direction pointing from 0 to the central point 0 4 of the edge AoA5. The angle between two planes A2OAo and A 2 0 0 4 is n/lO. So, n  y = 7r/2  n/lO, and y = 3n/5. After the rotation s6, the 2 axis is rotated to the direction OB4, so we have a = n/5 and p = 285. The Y axis is rotated to the direction pointing from 0 to the central point 0 5 of the edge A1A5. The angle between two planes B4OAo and B40D5 is 3n/10. So, n  y = n/2  3n/10, and y = 47r/5. After the rotation ,511, the 2 axis is rotated to the direction of the negative 2 axis, so we have p = 7r and only a  y is independent. We may choose y = 0. The Y axis is rotated to the direction pointing from 0 to the central point D6 of the edge B3A5. The angle between OD6 and the Y axis in the XY plane is 47r/5. So, Q = 47r/5. S12 is a rotation around the Y axis through n,so its Euler angles are ,8 = 7r and a = y = 0.
+
+
8 . Express the representation matrix D j ( n , w > of a rotation around the direction n(8, cp) through LJ in terms of Dj(e3,a ) and d j ( p ) .
Solution. Letting S be a rotation with the Euler angles a = cp, /3 = 8 and y = 0, we have SR(e3,w)S' = R ( ~ , L J ) ,
Threedimensional Rotation Group
129
9. Calculate all the matrix entries d;,(w) by Eq. (4.11) where j = 1/2, 1, 3 / 2 , 2, 5/2 and 3.
Solution. Due to the symmetric relation (4.12) of d j ( w ) , we only need to calculate the matrix entries d;p(w) with 0 5 v 5 p. The remaining matrix entries can be calculated by the following formulas
It is useful to obtain the formulas for dLp(w) with p = j , j  1, j  2 , and
0 from Eq. (4.16):
For convenience, we use the brief notations c = cos(w/2) and s = sin(w/2). dip(7r  w ) can be calculated from d;,(w) by interchanging c and s.
d:,(w) = c2, dAl(w) = J z c s , d3/2
(3/2)(3/2)(w)
= c39
dAo(w) = c2  s2, d3/2
(1/2)(3/2)
(dl
=
d3/2
= c3  2cs2, dq2(w) = 2c3s,
(1/2)(1/&4
4 2 ( 4
= c4,
d&(w) = fic2s2,
dql(w) = c4  3c2s2,
&(w)
= c4  4c2s2
+
54,
d & ( w ) = J G C S (c2  2), d5/2 ( 5 / 2 ) ( 5 / 2 )( W ) = c 5 7
130
Problems and Solutions in Group Theory
d5/2 d5/2 d5/2
(1/2)(3/2)
(3/2)(5/2) (O) (3/2)(3/2)
= J5c4 s , ( w ) = c5  4c3s2,
d5/2 (w) = m c 3 s 2 , (1/2)(5/2)
( w ) = d2 (2c4s  3c2s3),
d5/2
(1/2)(1/2)
(4= c5  6c3s2+ 3cs4,
d;,(w) = c 6 ,
d l 3 ( w ) = &c5s, d i 3 ( w ) = 2&c3s3, dY2(w) = (c5s  2c3s3), d ; l ( w ) = c6  8c4s2 6c2s4,
d : 3 ( ~= &c4s2, ) d l 2 ( w ) = c6  5c4s2,
+
d i 2 ( w )=
J30(c4s2 c2s4),
& ( w ) = 2 4 (c5s  3c3s3
+ cs5) ,
d&(w) = c6  9c4s2 9c2s4 s6.
10. Reduce the subduced representation from the irreducible representation D3 of SO(3) with respect to the subgroup D3 and find the similarity transformation matrix.
+
Solution. In terms of the formula x 3 ( w ) = sin(7w/2)/sin(w/2) for the representation D3 of S0(3), the characters of the elements with the rotaand tional angles 0 (the identity E ) , 2 ~ / 3 T are calculated respectively to = be x 3 ( E )= 7, x 3 ( 2 r / 3 )= 1 and x3(r) 1. The subduced representation from D3 with respect to the subgroup D3 is reducible, and the multiplicities u j of the irreducible representations A l , A 2 and E in it can be calculated by Eq. (3.11):
U A = ~
(1/6) (1 x 7 x 1 + 2 x 1 x 1+ 3 x (1) x 1) = 1,
U A ~ = UE
(1/6) (1 x 7 x 1 + 2 x 1 x 1+ 3 x (1) x (1)) = 2,
= (1/6) {I x 7 x 2
+ 2 x 1 x (1) + 3 x (1) x 0} = 2.
Namely,
The dimensions of the representations on two sides of the equation are both 7. In order to obtain the similarity transformation matrix for reduction, we need the matrix forms of the generators D and A of the subgroup D3 before and after the similarity transformation. The forms after the transformation have been given in Problem 11 of Chapter 3. The forms before
Threedimensional Rotation Group
131
the transformation depend upon the Euler angles of two generators:
D:,(A)
= D : p (  ~ / 2 ,T , ~ / 2 = D:,(O, )
T ,T )
= bv(p),
D 3 ( D )= D3(&3, ~ / 3 ) diag(1, w2,w, 1,w2, w , l}, 2 =
w = exp{i2~/3} = (1  i&)
/2,
w2 = (1
+ i&)
/2.
X'D3(A)X == diag(1, 1, 1,l, 1,1, l}, / 2 0 0 0 0 2 0 0 0 0 2 0 1 XlD3(D)X = 0 0 012 0 0 0 ~ 0 0 0 0
\ o o o o
0 0 0 &0 0 0
0 0 0 0 1 00  1 4 a1
0 0 0
The row index of X is the same as that of D 3 , which runs over 3,2;  , (3). The column index of X is the same as that of the reduced representation, denoted by the representations and their rows after reduction, which runs over Al, A2(1), A2(2), [E(1)1], [E(1)2], [E(2)1] and [ E ( 2 ) 2 ] , where the digits in the bracket are used to distinguish the multiple representations. D3(D) and X  l D 3 ( A ) X are diagonal, and D3(A) is a block matrix containing three ol and one digit 1. The first three columnmatrices of X [designated by A l , A2(1) and A2(2)] are the eigenvectors of D 3 ( D )for the eigenvalue 1 and the eigenvectors of D3(A) for the eigenvalues 1, 1 and 1, respectively, so that only the lst, 4th and 7th components (designated by 3, 0, and 3) are not vanishing:
The 4th and 6th columnmatrices of X (designated by [E(1)1] and [E(2)1]) are the eigenvectors of D3(A) for the eigenvalue 1. Due to the orthogonality, their l s t , 4th and 7th components are vanishing and the remaining components can be chosen as
Substituting them into the similarity transformation formula on D3( D ) ,we can calculate the remaining colomns in X. In the calculation we are only interested in the rows where the matrix entries are not vanishing. For the 4th column of X (designated by [E(1)1]), are only interested in the 2nd we
132
Problems and Solutions in Group Theory
and 6th rows (designated by 2 and 2):
For the 6th column of X (designated by [E(2)1]), are only interested in we the 3rd and 5th rows (designated by 1 and 1):
Finally, we obtain the similarity transformation matrix X : 1 0 0 0 0 0 0 0 l i 0 0 0 0 0 0 1 x=1 0 o J z o o o 0 0 0 0 0  1 0 0 0 1i 0 (11 0 0 0 0
'1
Jz
0 0 4 0 4 0 0
X . , ~ ( 1 ) 2 X . , E ( 2 ) 2 are and
the eigenvectors of D 3 ( A )for the eigenvalue 1.
1
11. Reduce the subduced representations from the irreducible representations D20 and D1* of SO(3) with respect to the subgroup I (the proper symmetry group of the icosahedron) , respectively.
Solution. From the character formula for Dj(SO(3))
we calculate the characters for the elements with the rotational angles 0 (the identity E ) , 2n/5, 47r/5, 27r/3, and 7r, respectively
x20(E)= 41,
sin(4i+) = 1) sin ( n /5) sin (82n/5) sin (417r/3) = 1 = 1, x20(27r/3)= sin(7r/3) x20( 4 ~ / 5 ) sin(27r/5) = x20(27r/5) =
Threedimensional Rotation Group
133
X20(7r) =
sin (41 / 2) 7r = 1, X18(E) = 37, sin(7r/2) sin(377r/5)  sin(27r/5) = 2 cOS(T/~) pl, = x18(27r/5) = sin(z/5) sin(7r/5) sin( 74n/5) sin(47r/5) = 2 C O S ( ~ T = p, /~) x'*(4n/5) = sin(27r/5) sin(27r/5) sin(377r/3)  sin(.rr/3) x18(27r/3) = = 1, sin ( n /3) sin(3fi77in$j3) sin(.rr/2) p(") =   1. = sin(7r/2) sin(7r/2)
The multiplicity u{ of each irreducible representation I of I can be calcu' lated with the character formula (3.11) and the character table of I (see Problem 18 in Chapter 3). For the representation D20, we have
a ; '
= (1/60) (1 x 41 x 1 12 x 1 x 1 12 x 1 x 1 20 x (1) x 1 15 x 1 x 1) = 1,
+
+
+
+
a$! = (1/60) (1 x 41 x
3
+ 12 x 1 x p' + 12 x 1 x (  p )
a20 Tz

U ?
=

+ 15 x 1 x (l)} = 2, (1/60) (1 x 41 x 3 + 12 x 1 x (  p ) + 12 x 1 x pl + 20 x (1) x 0 + 15 x 1 x (1)) = 2, (1/60) (1 x 41 x 4 + 12 x 1 x (1) + 12 x 1 x (1) + 20 x (1) x 1 + 15 x 1 x 0} = 2,
x0
+ 20 x (1)
u20 H
(1/60){1 x 41 x 5 + 1 2 x 1 x 0 + 1 2 x 1 x 0
+ 20 x (1)
x (1)
+ 15 x 1 x 1) = 4.
For the representation Ill8, we have
= (1/60) (1 x 37 x 1
+ 12 x (p)'
x 1
+ 12 x p x 1
+ 2 0 x 1 x 1 + 1 5 1 x I } = 1, ~
a;: = (1/60) { 1 x 37 x 3
u18 T2
+ 12 x (p)'
x pl
+ 12 x p x (  p )
+ 20 x 1 x 0 + 15 x 1 x (1)) = 1, (1/60) (1 x 37 x 3 + 12 x (p)' x (  p ) + 12 x p x p' + 20 x 1 x 0 + 15 x 1 x (1)) = 2,
134
Problems and Solutions in Group Theory
UE =
uf=
(1/60) (1 x 37 x 4 + 12 x (p)'
320 x 1 x 1 + 1 5 x 1 x 0) = 3 ,
x (1)
+ 12 x p x (1)
( 1 / 6 0 ) { 1 ~ 3 7 ~ 5 + 1 2 ~ (  p~) 0~ + 1 2 ~ p ~ O +20~1~(1)+15~1~1}=3.
Thus,
The dimensions of the representations on both sides of each equation are the same.
12. Reduce the subduced representations from the irreducible representations D1, D 2and D3 of SO(3) with respect to the subgroup I, respectively, and calculate the similarity transformation matrices. From the results, calculate the polar angles of the axes in the icosahedron.
Solution. The representation matrices of the generators of I in each irreducible representation have been given in Problem 21 of Chapter 3, and the polar angles for all twofold axes, threefold axes and fivefold axes in I have been listed in Problem 18 of Chapter 3. In the present Problem we are going to show how to obtain these results. We need to calculate the irreducible bases in the group space of I by the projection operator method, just like we have done for the groups T and 0 in Problems 22 and 23 of Chapter 3. However, in the calculation we have to use the multiplication table of I, which is neglected in this book due to limited space. In the following we will only explain the calculation method by using the results for the multiplication of elements. The reader can find the multiplication table in the textbook Group Theory for Physicists, Chap. 3 [Ma]. However, it is suggested to pay more attention to the calculation method introduced here by accepting the multiplication results. Since the representation matrix of TOwe choose is diagonal and T i = E , each diagonal matrix entry of TOmust be a power of q = exp{i2~/5}. From the character of TOin each representation, we have
DA(To)= 1, DT1(T0) diag{q,l,q'}, =
DTZ(To)= d i a g { ~ ~ , l , q 
DG(To)= diag{q2,q,q1,q2},
DH(To)= diag{q2,q,l,q1,q2}.
This power p can be used to designate the row and column. Note that p is
Threedimensional Rotation Group
135
module 5, p + 5 = p. We first calculate the vectors jection operator PI:
@Fin the group space of I by the pro
where c is the normalization factor.
@$
satisfies
Letting = E , we obtain which is vanishing if u # v. Then, letting , R(2) an element in I which does not appear in be say S11, we calculate @jL2?, which is vanishing if p v # 0. In this way, we calculate @f! and = by R(3) S5 and R(4)= S I O .
@rd,
+
@rd
)
Now, we consider the case of p = v = 1, where there are three independent bases
136
Problems and Solutions an Group Theory
Following the method used in Problems 22 and 23 of Chapter 3, we introduce W , which is the sum of elements in the class C5, to pick up the linear combinations of the vectors such that the new bases belong to the irreducible representations:
w=
,A
c
5
j=O
(Tj +Tj")
,T'
)
Dr(w) = J l ,
)
= 12)
= 4p1
Q*2
= 4p,
p1
, = G
3,
,H
=0 ,
p = ( A 1)/2 = q 
+ 771,
= (&
+ 1)/2 = 112
 q2.
In order to calculate the matrix form of W in the bases a::), we only need to calculate the coefficients of three terms of E , S and SS in the 1 action of W ,
w@;) { p E  plS1 + . .} /&, =
*
w@\;)5plE ={
wait'
4 2p1s1 4p2s6 4
. . .} 15,
= (p2s1 p2s6
+ . . .} 15.
is
The matrix form of W in the bases
py.
P2
p2
7'1)
The eigenvectors corresponding to the eigenvalues 4p', 3 and 0 respec* tively are
: ( 2 ; I ) >
P
A ( :)) L ( 2 ) . m
1 where

Thus, we obtain the new basis belonging to the representation we are only interested in the terms of E and S1:
par:)}
/2 = E/(2&)  plS1/10
+
*
..
Sl\II:i
= p'E/10
+ . . . = p'9*' 11/&+
....
Solving it, we have D Z ( S 1 ) = p'/&. On the other hand, following the method used in the preceding Problem, we reduce the subduced representations from the irreducible representations D1, D2 and D3 of SO(3) with respect to the subgroup I, respectively.
Threedimensional Rotation Group
137
In fact, the subduced representations from D 1 and D2 with respect to the subgroup I are the irreducible representations, and that from D3 is equivalent to the direct sum of DT2((I) and D G ( I ) . Due to the forms of Dr(To) we have chosen, we are able to define
From the explicit form of d'(P) given in Problem 9, we have
where c = cos 64, s = sin 64 and O4 is an acute angle. Thus, we obtain
c2 = p  l / & ,
s2 = p/&,
cs = c2  s2 = l / &
tan64 = s/c = p = Then,
(&  1 ) 12,
64 = 31.72".
c2  a c s s2  4 c s c2  s2 J2cs s2 a c s c2
p1
a
Jz
1
Jz
p
= DT1(Sl),
D2(Si)
c4
2c3s &2s2
c4
2 3
fics(c2
3c2s2
+ 3c2s2
fic2s2
fics(c2
2cs3
2cs3
s4
+ s4
 2) c4  4 2 2 + s4 fiCs(.2  .2)
fic2s2
 s2)
3c2s2
fics(c2
+ 54

s4
2cs3
c4
+ 3c2s2
2) fic2s2
 2 ~ ~ s
c4
2cs3
2c3s
p  2 2p1
2p1
p2
JG
1
fi
2p
p2
p2
2p
& &
2 p p2 p 2 2p
4 fi
&
p2
&
2p1 2p1 p2
By the unitary similarity transformation X ,
138
Problems and Solutions in Group Theory
we obtain
(0 0 c 0 UOOb* 0 0 0 0 X= 0 1 0 0 0 0 0 0 O O d 0 \bOOa*
0 0 d* OO 0 1 0 0 0 0 0 0 1 0 0 0 c* OO 0
= (X')+,
where 1ul2 + lbI2 = lcI2 ld12 = 1. The row index of X runs from 3 to 3, and the column index runs over (T2,1), (T2,2), (T2,3),( G ,l), ( G , 2 ) , ( G , 3 ) ,and (G,4). We calculate the first column and the third column in the fifth row of the formula X  l D 3 ( S 1 ) X = DT2(S1)@ DG(S1):
+
D;2(S1)a + D;(3)(Sl)b= 0,
D;,(Sl)c+ D?(a)(Sl)d= 0.
From Problems 7 and 9, we have DE,(S,) = d2,(284) exp{   i p r } , and
D 3 S 1 ) = Jloc3s(c2  2s2) = J 2 7 2 5 p , D;(3)(s1) =  f i c 2 s 4 =  J m p ,
~ : 3 ( ~ = )J15c4s2 1
=dWpl,
D;(2)(s1)= Jlocs3(2c2  2)= d@%pl.
The solution is a / b = and c/d = have a = d = &@, b = c = and
m,
1 p
m.Due to normalization, we
,
DT2(S1) = 
Ib
DG(S1) = 
rpJzp7 JlI
Jz 1 Jz p1 Jz p
p 1 1 1
p1
1
p1
pl
1 1 p
l \ p1 p 1
Now, we are going to calculate the directions for the symmetric axes in a regular icosahedron. The azimuthal angles can be determined directly from Fig. 3.1. Here, we only calculate their polar angles. Following the notations used in Problem 18 of Chapter 3, we denote by 81 the polar angle for the fivefold axes except for To, by 82 and 83 the polar angles of two sets of the threefold axes, and by 8 4 and 8 5 the polar angles of two
Threedimensional Rotation Group
139
sets of the twofold axes, respectively. Another set of the twofold axes lies in the X Y plane with the polar angle 7r/2. We have calculate 04 from DT1(S1). The remaining polar angles can be calculated by the common geometry knowledge. Let the length of the edge of the regular icosahedron be 1. Denote by R the radius of the circumcircle, and by r the radius of the incircle. The face of the regular icosahedral is a regular triangle. The distance from the center of the triangle to the vertex is d = From Fig. 3.1 we see that sine4 = (2R)l, t a n & = d / r , and R2 = d2 + r 2 . Thus,
m.
(y)
1/ 2
= 0.9511,
& = 0.7558,
4d3
d 4 t a n & =  =  3  &= 2p2, = i r 3+&
e2 = 37.380,
2 tan O4 d5 1 tane5 = cote4 =pl = = 2, 2 1  tan2 e4 e2 e3 = 2e5 = 7r el, e3 = el  e2 = 79.190 tan O1 tan 0 2 = 3 4& = 2p2. tan03 = 1  tan81 tan02 tanel =
+
+
+
13. Prove that the elements u(n,w) with the same w form a class of the SU(2) group.
Solution. It has been proven that the SU(2) group is homomorphic onto the SO (3) group through a twotoone correspondence:
Denote by 8 and cp the polar angle and azimuthal angle of the direction n, respectively. We define
Thus,
S3 b
=nb,
140
Problems and Solutions an Group Theory
Since u(n,w ) = 1 cos(w/2)  i (5 n)sin(w/2), we have
v'u(n, w)v = 1 cos(w/2)  ia3 sin(w/2) = u(e3,w > .
Namely, all elements u(n,w ) with a given w are conjugate to u(eg,u ) . They belong to the same class. Since Tr u(n,w) = 2cos(w/2), 0 5 w 5 27r, those u(n,w) with different w must not be conjugate to each other. They belong to the different classes. This completes the proof.
4.3
Lie Groups and Lie Theorems
A group G is called a continuous group if its element R can be described by g independent and continuous parameters r p , 1 5 p 5 9, varying in a gdimensional region. g is called the order of a continuous group G, and the region is called the group space of G. It is required that the set of parameters r , is onetoone correspondence onto the group element R at least in the region where the measure is not vanishing. The parameters t , of the product T = RS are the functions t , = fp(rl . . . ,r g ; . . . , s g ) = s1 j,(r; s) of the parameters of the factors R and S. fp(r; s) are called the composition functions. The domain of the definition of f p ( r ; s )is square of the group space, and the domain of function is the group space. The composition functions have to satisfy some conditions such as
*
where F, are the parameters of the inverse Rl, and e, are the parameters of the identity E . e, are usually taken to be zero for convenience. The continuous group is called the Lie group if the composition functions are analytic, 'or at least piecewise continuously differentiable. A Lie group is said to be a mixed Lie group if its group space falls into several disjoint regions. A mixed Lie group contains an invariant Lie subgroup G whose group space is a connected region in which the identity element of the group lies. The set of elements related to the other connected region is the coset of the invariant Lie subgroup. If the group space of a Lie group G is multiplyconnected, there exists another Lie group G' with simplyconnected group space such that G' is homomorphic onto G. G' is called the covering group of G. The group space of SO(3) is doublyconnected, and its covering group is SU(2). The Lie group is said to be compact if its group space is compact. If the group space of G is a closed Euclidean region, G is compact.
Threedimensional Rotation Group
141
The group elements are said to be adjacent if their parameters differ only slightly from one another. An element is said to be infinitesimal if it is adjacent with the identity E. The infinitesimal elements describe the local property of the Lie group. The representation matrix of an infinitesimal element A ( a ) is determined by g generators I ,
*
D(A)= 1 
ix
9
a,Ip,
I, = i 
(4.17)
,=l
I, are linearly independent of one another if the representation D ( G ) is faithful .
The First Lie Theorem: The representation of a Lie group G with a connected group space is completely determined by its generators. Namely, the representation matrix of any element R in G can be solved from the following differential equation and the boundary condition:
*
where S,, ( r ) is independent of the representation. Spy r ) depends upon ( the choice of the group parameters and the composition functions of the Lie group:
The Second Lie Theorem: The generators of any representation of a Lie group satisfy the common commutation relations
I,Iv  I,I, = i
where C,: tation:
*
x c,pr,
e
(4.19)
are called the structure constants, independent of the represen
Usually, they are calculated by the commutation relations of the generators in a known representation. The structure constants depend upon the choice of the group parameters.
142
Problems and Solutions in Group Theory
The Third Lie Theorem: The necessary and sufficient condition for a set of parameters ;C to be the structure constants of a Lie group is the , the parameters satisfy
P
A representation D ( G ) and its generators I , of a Lie group G satisfy
D(R)I,D(R)' =
c
U
IuD;",jR),
(4.21)
where Dadj(G) is called the adjoint representation of the Lie group. The generators in the adjoint representation is related to the structure constants:
The selfrepresentation of SO(3) is the adjoint representation of both the SU(2) group and the SO(3) group:
(4.23)
where PR is the transformation operators for the scalar functions, and OR = QRPR is the transformation operators for the spinor functions of rank s:
La, Sa, and J a are the orbital, the spinor, and the total angular momentum operators, respectively. If the spinor function S ~ ( X ) rank s belongs to the pth row of the of irreducible representation Dj(SO(3)),
Threedimensional Rotation Group
143
j(j
@ i ( z )is the simultaneous eigenfunctions of J 2 and J3 for the eigenvalues 1) and p. Similar formulas hold for the orbital space and the spinor
+
space.
14. Calculate the representation matrices of the generators in an irreducible representation of SU(2) by the second Lie theorem. Solution. This is a fundamental problem both in the matrix mechanics
theory of quantum mechanics and in the representation theory of the Lie groups. The angular momentum operators are the generators in the transformation operators of SU(2). From the second Lie theorem, the generators in an irreducible representation of S U ( 2 ) satisfy the common commutation relations like that of the angular momentum operators:
J2 = J i
+ J3 + JJ+
= J:  5 3
+ J+J,
J* = 5 A i J Z , 1 [ J 2 , Ja] = O .
[J3, Jj] = k J * ,
[J+, J] = 2 J 3 ,
Since J 2 is commutable with all generators Ja, it takes a constant in an irreducible representation due to the Schur theorem. In terms of the Dirac notation, we denote by l j ) the eigenfunction of J3 with the highest eigenvalue j in the representation space:
If the dimension of the representation is finite, there must exist a positive integer n such that
(J)""
Thus,
53
l j ) = 0,
( J  ) @l j ) # 0,
o 5 p 5 n.
Id} = ( J  J3  J  ) l j ) = ( j  1){ J  l j ) } , 5 3 (JY Id = ( j  n) (JY ) , lj
{JJ 2 (J)n
IA = ( j  n ) ( j n  1)(J)"
lj) =j ( j
+ 1)(J)"
lj).
The positive root for n is 2 j . Therefore, for an irreducible representation with a finite dimension, j must be an integer or a halfoddinteger. Let
where, except for A  j = 0, A , is a nonzero constant to be determined. We are going to show by induction that this 2 j 1dimensional space spanned
+
144
Problems and Solutions an Group Theory
by Ip) is closed for the angular momentum operators J,, namely,
where B,+1 is a constant to be determined. It is true when p = j because of Bj+l = 0. If it is true when v 5 p 5 j , we want to show that it is also true for p = v  1 2  j ,
This completes the proof. The above equation gives a recursive formula:
Choosing the suitable factor in the basis, we are able to make A, = B, = I?; to be a positive real number,
15. For any oneorder Lie group with the composition function f ( r ; s ) , please try to find a new parameter r' such that the new composition function is the additive function, f ' ( r ' ; s ' ) = r' + s'. The Lorentz transformation A(v) for the boost along the 2 axis with the relative velocity v is taken in the following form. The set of them forms a oneorder Lie group:
1 0
0 0
A ( v ) = ( O 0 0 iyv/c iyv/c ) , y
;
0
y = (1  v2/c2)
1/2
,
f( W V 2 )
=
1
+ v1v2/c2'
v1 f v 2
Find the new parameter with the additive composition function.
Solution. For any given oneorder Lie group with the composition function
Threedimensional Rotation Group
145
From the first Lie theorem,
where I is the generator of the representation D ( r ) , e is the parameter of the identity, taken to be zero. Solving the differential equation with the boundary condition, we obtain
Define the new parameter
w(r) =
lr
,
S(t)dt.
The new parameter w ( 0 ) of the identity is still zero. Thus,
D ( r ) = exp {  i I w ( r ) } ,
exp {  i I [ w ( r )
= exp {  i I w ( r s ) }
D ( s ) = exp {  i I w ( s ) } ,
+ w ( s ) ] } = D ( r ) D ( s )= D ( r s )
w(rs) = w ( r ) + w ( s ) .
For the multiplication of elements, the new parameter is additive. As an example, for the Lorentz transformation with the relative velocity v along the 2 axis, the composition function for the parameter v is
Defining a new parameter w
w=
1'
S(t)dt = arctanh
= arctanh
(5) ,
we have tanhw = v/c, coshw = (1  v2/c2)1'2 = y, sinhw = yv/c,
146
Problems and Solutions an Group Theory
A(w)=("
0
0
).
0 0 coshw i sinhw 0 0 isinhw coshw
In the multiplication of two Lorentz transformations, the parameter w is additive.
4.4
Irreducible Tensor Operators
The set of ( 2 k + l ) operators L % ( z )k 5 p 5 k , is called the irreducible , tensor operators of rank k if those operators transform according to the irreducible representation D k ( R )in the rotation R of SO(3):
(4.26)
*
For the infinitesimal rotation, we have
(4.27)
Equation (4.26) is equivalent to Eq. (4.27). They are both regarded as the definition of the irreducible tensor operators of rank k . L;(z) are called the irreducible tensor operators of rank k with respect to the orbital space or the spinor space, if replacing OR with PR or QR, respectively. The irreducible tensor operator of rank one is called the vector operator. Since the representation D1 is equivalent to the selfrepresentation R of SO(3) [see Eq. (4.14)], there are two kinds of components for the vector operators: O R L ; ( z ) O ~ l=
A=1
3
(4.28)
O,Va(Z)OR1 =
b= 1
Threedimensional Rotation Group
147
L: =  I / v ( V 1
L; = v 3 ,
Lll = m
( v
+ iV2),
1  iV2),
v1
=
v2 =
v3
i m
( L i l  L:) ,
(LLl
+ L;) ,
= LA.
L;(z) are called the spherical components of the vector operators, and
are called the rectangular components of the vector operators. The examples for the vector operators in common use are the electric dipole operators Z a or Y: (n),the momentum operators p a , the angular momentum operators J a , L a and S a etc.
Va(2)
*
Being a vector operator,
Ja
satisfies
where Ja is the generator for the transformation O R , and R b a is the representation matrix in the adjoint representation of SO(3) [see Eq. (4.23)]. Letting a = 3, we have
OR J 3 ( x ) O R 1 = J n(0, cp),
R = R(cp,0,7 ) .
(4.29)
y is often taken to be zero. If !P$(x)is the function belonging to the pth row of the irreducible representation Dj [see Eq. (4.25)], @$(x) is the
simultaneous eigenfunctions of J 2 and 5 3 for the eigenvalues j ( j 1) and p. Now, from Eq. (4.29), OR!P~(Z) the eigenfunction for J  n ( O , c p ) with is the eigenvalue p. For a spherical symmetrical system, the eigenfunction of the energy can be chosen to belong to a certain row of an irreducible unitary representation:
+
*
The WignerEckart theorem says that if OR is a unitary operator for the inner product of functions, then
where ( @ f l l ! P j ) , called the reduced matrix entry, is independent of p. Namely, the functions belonging to two inequivalent irreducible representations are orthogonal to each other, the functions belonging to different
148
Problems and Solutions in Group Theory
rows of an irreducible unitary representation are orthogonal to each other, and the inner product of two functions belonging to the same row of an irreducible unitary representation is independent of the row. The reduced matrix entry depends upon j as well as the explicit forms of @ and Q. The theorem can be generalized to the matrix entries of the irreducible tensor operators in such basis functions. Because
OR {Lt(x)Q;(x)} = {oRLt(x)o,’}
=
c {G(s)Q!(4)
XU
{oRQ~(x)}
?
{D$(W@p(R)}
L$(z)!Pi (z) is transformed according to the direct product representation, which can be reduced by the unitary similarity transformation C j k :
j+k
( C j y ( D j ( R ) x o k ( R ) )c
j k
=
@ DJ(R).
J=ljkl
The matrix entries of C j k are called the ClebschGordan (CG) coefficients. According t o the phase convention, we make the CG coefficients real and satisfying
‘;:J(j+u)
and
‘L;k)J(pk)
are positive,
 kl
CpuJM 0 only when A = p + u and Ij jk # 4
5 J <j 
+k
= (1)kJp
 (l)jJ+”
() 2 ()J + 1
2J+1
112
1/2
The ClebschGordan (CG) coefficients can be used to combine L: (z)!PL(2) into the functions F& (z) belonging to the irreducible representation:
Threedimensional Rotation Group
149
From the WignerEckart theorem, we have
where the constant c is independent of the subscripts p‘ and Ad. c depends on the indices j ‘ , k and j as well as the explicit forms of a, L and Xk. c is IILk still called the reduced matrix entry, denoted by (d IIXkj). Thus,
1)(2k 1)(2j + 1) matrix entries in the form of There are (2j’ (*$ (x)lL:(x)I!P$(x)). Due to their property in rotation, they are related by the CG coefficients. The WignerEckart theorem simplifies the calculation of (2j‘ + 1)(2k 1) ( 2 j 1) matrix entries to the calculation of only one parameter (d IILkllXkj). If there is one matrix entry with given subscripts p‘, p and p which can be calculated, the remaining matrix entries all are calculable. In most cases, even one matrix entry is hard to be calculated. For example, the explicit forms of the wave functions a:(’) and @ c l ( x ) are unknown. However, Eq. (4.32) gives some relations among the matrix entries, which sometimes can be compared with experiments.
+
+
+
+
16. Directly calculate the CG coefficients for the direct product representation of two irreducible representations of SU(2) in terms of the raising and lowering operators J*: a) D112x D1/2,b) D1/2x D1, c) D1 x D1, d) D1 x D 3 j 2 .
Solution. In the representation space of Dj x D k ,the basis state l j , p ) l k , v) is the eigenfunction of 5 for the eigenvalue p+v. The state with the highest 3 eigenvalue of 5 is l j , j ) l k , k ) . It means that in the reduction of D j x D k 3 there is an irreducible representation Dj+lC. Ij, j ) Ik,k) is also the state with the highest eigenvalue of 5 in the representation space of Dj+lC, 3
Applying the lowering operator 5 to it, we can calculate all remaining basis states in the representation Dj+lC.In the subspace which is orthogonal to the basis states belonging to D j + k , we try to find the state with the highest eigenvalues of 53,which belongs to another irreducible representation D in the reduction of Dj x D k . Then, calculate the basis states in D J by the
150
Problems and Solutions an Group Theory
lowering operator. In the calculation the following formulas are useful:
JIIJ,
M ) = [(J M ) ( J  M
+
+ 1 ) p 2 llJ, M  1),
a) In the direct product representation space, the state with the highest eigenvalue of J3 is ~ 1 / 2 , 1 / 2 ) ~ 1 / 2 , 1 / 2 )which belongs to the irreducible , representation D1.
= p / 2 , 1/2)11/2, 1/2).
The remaining state which is orthogonal to I 1,O) is
110,O) = 21/2 { \ 1 / 2 , 1 / 2 ) p / 2 , 1/2) 
It belongs to the irreducible representation Do. From the phase convention (4.31) for the CG coefficients, we choose the coefficient in the term 11/2,1/2)11/2, 1/2) to be positive. Finally, we obtain the ClebschGordan series for the direct product representation
D112 x D112N D1 @ D o .
b) In the direct product representation space, the state with the highest eigenvalue of J3 is ~ 1 / 2 , 1 / 2 ) ~l1), ,which belongs to the irreducible repre
Threedimensional Rotation Group
151
sentation D312
1 1 3 3/2) = ~~
IW,1/w, I),
I13/2,1/2) = 31/25 I13/2,3/2)
= 3lj2 { J z I 1 / 2 , 1 / 2 ) p , O )
+ 11/2) 1/2)11,
l)} ,
( ( 3 / 2 ,1/2) = 21J113/2,1/2)
= (12)ll2 {211/2,1/2)11, 1)
+ JZ11/2, 1/2)11,0)
+ Jz11/2, 1/2)11,0)}
= 3ll2 { 11/2,1/2)11, 1)
7
+ v q 1 / 2 , 1/2)11,0)},
+ 211/2, 1/2)11,
1))
I13/2, 3/2) = 3112 J I13/2,  1/2)
= 31 {11/2, 1/2)11, 1)
= p / 2 , 1/2)11, 1).
The basis state which is orthogonal to 113/2,1/2) belongs to the irreducible representation From the phase convention for the CG coefficients, we take the coefficient in the term ~ 1 / 2 , 1 / 2 ) ~ 1 , 0 ) be positive: to
111/2,1/2) = 31/2 { 11/2,1/2)p,O)  @/2,
11 /2. 1  1 /2\ = .7_ll1/2.1/2\ 111
I
1/2)11
1
 I
I
1
= 31/2 { Jz11/2) 1/2)11, 1)
 2l1/2,
+ p / 2 , 1/2)
1P)I1,0)}, = 3ll2 { J z ( 1 / 2 ) 1/2)11, 1)  11/2) 1/2)11)0)}.
Finallv. we obtain the ClebschGordan series for the direct product representation
u
, x u
 u
u/u
.
c) In the direct product representation space, the state with the highest eigenvalue of J3 is I1,l) 11,1),which belongs to the irreducible representation D2.
I
152
Problems and Solutions in Group Theory
The remaining basis states can be calculated similarly. They can also be calculated by the symmetry (4.31) of the CG coefficients:
The basis state which is orthogonal to 112,l) belongs to the irreducible representation D1. From the phase convention for the CG coefficients, we take the coefficient in the term I1,l)I1,O) to be positive:
In the direct product representation space, there are three basis states with the zero eigenvalue of J 3 . We have obtained two orthogonal basis states 112,O) and 111,O).The third state Il0,O) can be calculated from the orthogonal condition. But here we prefer to use another method. This method 3 is based on the condition for the state with the highest eigenvalue of 5 in an irreducible representation of SO(3) that it is annihilated by the raising operator J+. Let
110,O) = all, 1)11,1)
+ b l l , O ) l l , 0) + c p , 1)11, 1).
J+ annihilates this state:
The solution is a =  b = c. After normalization we obtain
Il0,O) = 3li2 { I1)1)11, 1)  p , O ) l l , 0)
+ 11,1)11,1)}.
Finally, we obtain the ClebschGordan series for the direct product representation
D 1x D ~  D ~ c % D ~ $ D ~ .
Threedimensional Rotation Group
153
d) In the direct product representation space, the state with the highest eigenvalue of 5 is 11,1)13/2,3/2), which belongs to the irreducible repre3 sentation D512.
The basis state which is orthogonal to 115/2,3/2) belongs to the irreducible representation D312.According to the phase convention of the CG coefficients, we take the coefficient in the term I1,l)13/2,1/2) to be positive.
In the direct product representation space, there are three basis states with eigenvalue 1 / 2 of J3. We have obtained two orthogonal basis states 115/2,1/2) and 113/2,1/2). The third state belongs to the representation D112,and can be calculated by the orthogonal condition or by the annihilation condition of J+. Now, we use the latter method. Let
J+ annihilates this state:
The solution is u&
= b&
= c a . After normalization we obtain
154
Problems and Solutions in Group Theory
111/2,1/2) = 6lI2 { 11,1)13/2, 1/2)  fi11,0)13/2,1/2)
+ 611, 1)p/2,
3/2)}.
The expansions for the remaining basis states can be obtained by the symmetry (4.31) of the CG coefficients:
Finally, we obtain the ClebschGordan series for the direct product representation
17. Prove two sets of formulas for the CG coefficients in terms of the raising and lowering operators J k : a) In the reduction of D1I2 x D j ,
Threedimensional Rotation Group
155
b) In the reduction of D 1x Dj,
{
(j
+ M ) ( j + A4 + 1)
2 j ( 2 j + 1)
Solution. We proof them by induction. a) We only prove the first equation. The second equation can be obtained by orthogonality of the basis states and the phase convention. In the direct product representation space, the state with the highest eigenvalue of 5 is 3 I1/2,1/2) Ij, j ) , which belongs to the irreducible representation Dj+1/2,
It satisfies the first equation with A4 = j 1/2. Suppose that the first equation holds for M = m. We are going to show it also holds for M = m  1.
+
+
156
Problems and Solutions in Group Theory
The first equation is proved. b) In the direct product representation space, the state with the highest eigenvalue of J3 is 11)1)lj, j ) ,which belongs to the irreducible representation
Dj+1
It satisfies the first equation with M = j + l . Suppose that the first equation holds for M = rn. We are going to show it also holds for M = m  1.
Threedimensional Rotation Group
157
The first equation is proved. We substitute M = j in the second equation:
This equation is correct because J+Ilj, j) = 0. Suppose that the second equation holds for M = m. We are going to show it also holds for M = m  1.
The second equation is proved. The third equation can be proved with the same method. However, it can also be proved by orthogonality of the basis states and the phase convention.
18. Calculate the eigenfunctions of the total spinor angular momentum in a threeelectron system. Solution. We first calculate the eigenfunctions of the spinor angular momentum in a twoelectron system. The spin of each electron is 1/2. The spinor state for a twoelectron system is the product of the spinor states of two electrons, lp)lu), where p and u are taken to be f 1 / 2 , denoted by f. The eigenfunction of the total spinor angular momentum in a twoelectron system is denoted by IS12, M12), where 5'12 takes 0 or 1, IM12l 5 ,912. From
158
Problems and Solutions in Group Theory
the result given in Problem 16 we have
One may regard the threeelectron system as a compound system composed of a twoelectron subsystem and a oneelectron subsystem. Its state is the product of the states of two subsystems, IS12, M12)lp). Let IlS, S12, M ) be the state in the threeelectron system with the spin S and A4, calculated in terms of the results given in Problem 16:
+>I>
>
+ 1+>1)1+) + 1)1+>1>
+
+
+ m1 1, 1>1+>
>I>
If the twoelectron subsystem consists of the last two electrons, we calculate the eigenstate IS, S23, M ) in the threeelectron system with the spin S and A4 in terms of the results given in Problem 16:
Threedimensional Rotation Group
159
Comparing two sets of results, we find that the expressions for the states with total spin 3/2 are the same, but the expressions for the states with total spin 1/2 are mixed together. Actually, two sets of solutions are related by the Racah coefficients.
19. The spherical harmonic function YA(n) belongs to the mth row of the representation De of SO(3) so that it is the eigenfunction of the or
bital angular momentum operator L3 for the eigenvalue m. Calculate the eigenfunction for the eigenvalue m of the orbital angular momentum operator L  ii along the direction ii = (el  ea) /fi terms of in combining Y$ (n) linearly.
Solution. The polar angle and the azimuthal angle of the direction Ci = (el  e2)/2 are 8 = n / 2 and cp = n/4, respectively. Letting S = R( n/4,7r/2,0), we have
The eigenfunction of L . a for the eigenvalue m is
160
Problems and Solutions in Group Theory
because
20. Let the function + k ( x ) belong to the mth row of the irreducible repre
sentation D of SO(3). Calculate the eigenfunction for the eigenvalue ‘
b = (fie2 + e 3 ) /2 in terms of combining $ k ( x ) * linearly. Hint: Use the similarity transformation between the representation D j of SO(3) and its complex conjugate representation.
rn of the orbital angular momentum operator L . b along the direction
Solution. The representation Dj (R)*is equivalent to the representation Dj(@ ,
Combine the functions $&(x)* into the function + k ( x ) belonging to the irreducible representation De:
Thus, +L(x)is the eigenfunction of the orbital angular momentum L3 along the 2 axis for the eigenvalue m. The polar angle and azimuthal angle of b are n/3 and 7r/2, respectively. By the method used in Problem 19, we obtain the eigenfunction of the orbital angular momentum L  b along the direction b as follows, where S = R(7r/2,n/3,0):
Threedimensional Rotation Group
161
Another method of calculation is based on the property that L3t,bem(z)* = m$L(z)*. Thus, Pst,bLm(z)* is the eigenfunction of PsL3Pi1 for the eigenvalue m. Letting S = R ( x / 2 , ~ / 3 , 0we obtain the ), eigenfunction of the orbital angular momentum L  b along the direction b for the eigenvalue m:
Two results are the same up to the phase factor. Note that the summing index m' can be replaced with m'.
21. QR is the rotational transformation operator in the spinor space. In the rotation QR, the spinor basis e ( S ) ( p ) belongs to the pth row of the irreducible representation D" so that it is the common eigenfunction of the spinor angular momentum S2 and S3 for the eigenvalues s(s 1) and p, respectively. Based on this property, calculate the eigenfunction of the spinor angular momentum S  2 along the radial direction, where i. is the unit vector in the radial direction. Solution. The operator S . i is a scalar operator in the whole spatial rotation, but a vector operator in the rotation of the spinor space. Letting the polar angle and azimuthal angle of r be 8 and cp, respectively, we have
+
S i. 1Q R S ~ Q R ~ , R = qcp, e, 0).
*
Its eigenfunction for the eigenvalue p is
&Re(')((P) =
C e('))(J)&(R)
x
(s 6 ) QRe(')(p)= pQRe(S)(p).
22. There are three sets of the mutual commutable angular momentum
operators. One set consists of L 2 , L3, S2, and S3. The other set consists of J 2 , 5 3 , L2 and S2, and the third set consists of J 2 , J3, S2 and S  i.. Calculate the common eigenfunctions of three sets of operators, respectively.
Solution. The common eigenfunction of L 2 , L3, S2, and S3 is the product of the spherical harmonic function Yh(n) and the spinor basis e ( ' ) ( p ) . Combining the products Y:(n)e(')(p) by the CG coefficients, we obtain the common eigenfunction of J 2 , J3, L2 and S2, which is called the spherical
162
Problems and Solutions in Group Theory
spinor function
yjtS (n):
where the spinor basis e ( ' ) ( p ) is a spinor of rank s. The above formula is a spinor equation, namely, a (2s 1) x 1 matrix equation. Since the sum runs over the azimuthal quantum numbers, the spherical spinor function Yies(n) is also the eigenfunction of L2 and S 2 :
+
J2Yies(n)= j ( j
+ l)y;ts(fi) ,
(n) ,
J3YP jes(ii) = pYits(n) ,
JkYies (n) = I$pY$l L2Yiyn) =
e(e + 1 ) Y j y n ) ,
s " Y y ( n ) = s(s
+ 1)Yiyn).
By making use of the ClebschGordan coefficients calculated in Problem 17, we obtain the explicit forms of spherical spinor function Yics(n), where s = 1/2 and e = j 1/2:
The common eigenfunction of J2, 5 3 and S2 is a spinor of rank s and belongs to the irreducible representation Dj of SO(3) in the whole spatial rotation,
Let the spherical coordinates of the position vector r be (r,e,cp). T = R(cp,B,y) is a rotation transforming the 2 axis to the radial direction r. Tro = r, where ro = (r,O,O) is the position vector r along the 2 axis.
Three dimensional Ro t at i on Group
163
Replacing r with ro and R with 5"' in the above formula, we have
Write it in the component equation,
Since the lefthand side of the equation is independent of y,its righthand side has to be independent of y,too. Namely, !I!i(ro)g has to be zero except for the case u = (T,
Substituting it into !Pi (r)p,we have
where TO= R(p, 0,O). Remind that the quantity in the square bracket is nothing but the eigenfunction of S  f for the eigenvalue v [see Problem 211. is the eigenfunction of S  f if and only if the sum contains only one term with a fixed u. Namely, & ( r ) is nonvanishing only when u takes a given value. Therefore, the common eigenfunction of J 2 , J3, S2 and S  t respectively for the eigenvalues j ( j + 1))p, s ( s + 1) and v is
Qi(r)
[ Q T o e ( s ) ( 4 qu(TO)*+j,b) ] =
c
P
e ( S ) ( P ) q u ( c p6, OP;,(cp, ,
&O)*$j,(r),
where & ( r ) plays the role of the radial function, or simply the role of the normalization factor.
23. Calculate de(0) I: 2de(S)'}
representation.
, where is the representation { 0De of SO(3) and I: isde(e)third generator in the the matrix of R(e2,O)in
mm
164
Problems and Solutions in Group Theory
Hint: Use the property of the adjoint representation. Solution. The selfrepresentation of SO(3) is just its adjoint representation. From the definition (4.21) for the adjoint representation, we have
1 Since I! is diagonal, and I; = (: ishing diagonal matrix entry, so
+ I!)
/2 does not contain any nonvan
sin2 t~  { ( t m)(t m 1) (t m 4 sin2 e (e2 + t  m 2 }+ m2cos28. 2
+
+ + + + l)(t m ) }+ m2
COS~
e
24. Establish the differential equation satisfied by the matrix entries @,(a, P,y) of the representation Dj of SO(3).
Solution. D:,(a,P,r) is the matrix entry of the rotational operator OR(^,^,^) between the eigenstates of the angular momentum. Using the Dirac notation, we have:
Define
Thus, OR(u,p,y) be expressed by the angular momentum operators J,, can
Threedimensional Rotation Group
165
Ja is a vector operator, satisfying:
Letting s, = s i n a , c, = COSQ etc., we have
Hence,
Noticing the order of J+ and J  , we have
Finally, we obtain
Calculating the matrix entry between two eigenstates of the angular momentum, (j,v ( and l j , p ) , we obtain the differential equation satisfied by
166
Problems and Solutions in Group Theory
4.5
Unitary Representations with Infinite Dimensions
SO(3) group is a compact Lie group. It has irreducible unitary representations Dj of finite dimensions. However, SO(2,l) group is not a compact Lie group. Except for the identical representation, any irreducible representation of SO(2,l) with a finite dimension is not unitary, and the dimension of any irreducible unitary representation of SO(2,l) is infinite. In this section we will study the irreducible unitary representations of SO(2,l) in some detail as an example of the Lie groups which are not compact.
The set of all threedimensional real matrices R satisfying
*
RTJR = J ,
detR = 1,
J = diag(1, 1, o},
(4.33)
in the multiplication rule of matrices, forms the SO(3) group if 0 = 1 and the SO(2,l) group if u = 1. We will denote these two groups by G uniformly. The following inner product is invariant for the group G:
(4.34)
It can be seen from Eq. (4.33) that R& = 1  o (R23 R ; , ) . IR331 5 1 for S 0 ( 3 ) , but IR331 2 1 for SO(2,l). Therefore, SO(3) group is a compact Lie group, while SO(2,l) group is not compact. The group space of SO(2,l) falls into two disjoint regions. The region with R33 2 1 corresponds to the invariant subgroup SO+(2,1). For convenience we will still denote this subgroup S 0 + ( 2 , 1 ) by the same symbol SO(2,l) if no confusion arises. Discuss the infinitesimal elements and the generators of G:
+
Threedimensional Rotation Group
167
R = 1  iaX,
J=RTJR= Jia(XTJ+JX),
1 = d e t R = 1  iaTr X .
Thus,
T X = 0, r
X = JXTJ =
C uaTU.
a= 1
3
Denoted by La three generators, and by Ta their representation matrices in the selfrepresentation:
00 0
.=(
0 Oia iO o ) oo 0
,
0 i 0 T3=(;;;).
They satisfy the following commutation relations
Define the raising and lowering operators
L* = L1 f iL2,
[L3 , L*] = fL*,
[L+ , LI = 2aL3.
(4.35)
The Casimir operators are
L2 = 0 (L? L;)
+
+ L; =
aL+L
+ L3 (L3  1)
= aLL+
+ L3 (L3 + 1 ) )
(4.36)
[L2
, La] = 0,
a = 1, 2, 3.
As we have known, the covering group of SO(3) is SU(2). Similarly, we will show that the covering group of SO(2,l) is SL(2,R)) the set of twodimensional real matrices with determinant +1 where the multiplication rule of elements is the matrix product. The space composed by twodimensional traceless real matrices is a real space with respect to three real bases
71
*
=03,
r2= a l ,
73
= i a2.
(4.37)
Any twodimensional traceless real matrix X maps onto the position vector x = ( q , z 2 , q ) in the real threedimensional space by a onetoone
168
Problems and Solutions in Group Theory
correspondence:
21
= Tr
1 2
(XT~), 2 = Tr (XT~), = "I (XT~). x3 2 2 2
1
1
(4.38)
After a similarity transformation uXu' = X', €SL(2,R), X' is still a u traceless real matrix with the same determinant. Thus, the position vector x corresponding to X' is related to the position vector x corresponding to ' X by a transformation R ~ S 0 ( 2 , 1 ) .
Similar to the proof for the covering group of S0(3), we can show the twotoone correspondence between two matrices fu ~ s L ( 2 , and R ~ S 0 ( 2 , 1 ) . R) This correspondence is invariant in the product of elements. Therefore, SL(2,R) group is the covering group of SO(2,l). The representations for the SO(3) group and the SO(2,l) group including the doublevalued representations are actually the representations of the SU(2) group and the SL(2, R ) group, respectively.
25. Discuss all inequivalent and irreducible unitary representations of the
SO(3) group and the SO(2,l) group. Solution. Choose the basis states IQ, m) in the irreducible representation space where the representation matrices of L2 and L3 are diagonal,
In a unitary representation, the representation matrices of L3 and L2 are both hermitian, so Q and m are both real numbers. Let
Q = j ( j + 1).
There are two solutions for j
(4.40)
Two solutions are related with each other by the following transformation
j++jl.
(4.41)
Threedimensional Rotation Group
169
Recall that the irreducible representation is described by the real parameter Q . Two j's related by the transformation (4.41) describe the same representation. j is real when Q 2 1/4, and j is complex when Q < 1/4. From Eqs. (4.35) and (4.36), we have
Thus, beginning with a given basis state I&, mo), by the raising and lowering operators &, we can obtain a series of basis states IQ, m) where the neighboring m are different by f l :
(4.42)
until AQm or B Q equals to zero. Equation (4.42) is the definition for these ~ basis states. They are all common eigenstates of L2 and L3, satisfying Eq. (4.39). We need to show that the set of basis states is closed for L h , namely
(4.43)
The proofs for two equations are similar. We will prove the first equation in Eq. (4.43) as example. When m > mo, we have
= {aLL+

+ L3 (L3 + 1)) IQ, m  1) OAQm(LIQ, m ) )+ (m2 m ) IQ, m  1).
(4.44)
From Eq. (4.42), AQm is nonvanishing if I&, m ) exists, so LIQ, m) is proportional to IQ, m  1). Thus, Eq. (4.43) is proved. So, these basis states form a complete set, spanning an irreducible representation space of G. Note that the definition (4.42) for the basis state I&, m ) allows a multiplied factor in front of it. Namely, the values of AQm and BQm are undetermined. On the other hand, the basis state IQ, m ) is also the eigenstate of the operators L+L and LL+ for the eigenvalue AQmBQmand AQ(m+llBQ(m+l), respectively. In a unitary representation, the representation matrices for L+ and L are conjugate to each other, so the representation matrices of L+L and LL+ are both positive semidefinite. From Eq. (4.44), when
170
Problems and Solutions in Group Theory
m > mo, we have
Similarly, it is also true for m 5 mo. We may choose the factor in the basis state IQ,rn) such that
According to whether the state chain is truncated, i.e., whether AQ, (BQ,) is vanishing at some m, the irreducible unitary representations of the SO(3) group and the SO(2,l) group are classified into the unbounded spectrum D(Q,mo),the spectrum D s ( j ) bounded below, the spectrum D  ( j ) bounded above, and the bounded spectrum D ( j ) [Adams et al. (1987)I.
m
jm+l=O
I
j+m+l=O
Fig. 4.1 The varied region for the parameters j and m in the irreducible unitary representation of SO(3) and SO(2,l).
Recall that the values of o are different for two groups: 0 = 1 for the SO(3) group and o = 1 for the SO(2,l) group. From Eq. (4.45), the varied regions of the parameters j and m for two groups are different due to different 0 . For S0(3), Q 2 m2  m 2 1/4, and j is real. The possible values of j and m are restricted in the intersection of two regions: the region above or below two crossed lines j m = 0 and j  m 1 = 0, and
+
+
Threedimensional Rotation Group
171
the region above or below two crossed lines j  m = 0 and j m 1 = 0. The intersection is divided into two infinite regions indicated by SO(3) in Fig. 4.1 [Adams et al. (1987)]. Two regions are equivalent because they are related by the transformation (4.41). In the above region of the intersection,  j 5 m 5 j , and j >_ 0, namely, the values of m are finite. Letting the upper boundary of m be M , from
+ +
AQ(M+1) [ ( j =
+ M + l ) ( j jk!)]”2 = 0,
we obtain M = j >_ 0. Here and the remaining part in this Problem we take k to be the nonnegative integer. Letting the lower boundary of m be j  k, from
we obtain j = k / 2 = 0, 1/2, 1, 3/2, .... Therefore, the bounded spectra D ( j ) , where j is a nonnegative half integer and m = j , j  1,  .,  j , are the only unitary representations for the SO(3) group. Obviously, they are the commonly used finite dimensional representations D j (SO(3)). For the SO(2,l) group, j may be real or complex. When j is a complex ip, p > 0, one may have number, j = 1/2
*
+
Q = 1/4  p2,
m = m f k, o
 1/2
< mo 5 1/2.
(4.47)
This is called the principal series of representations D,(P,mo) for the unbounded spectrum. When j is a real number, the possible values of j and m are restricted in the intersection of two regions: the region in the left or right side of two crossed lines j + m = 0 and j  m 1 = 0 and the region in the left or right side of two crossed lines j  m = 0 and j m 1 = 0. The intersection is divided into two infinite areas and a finite area indicated by SO(2,l) in Fig. 4.1. Considering the equivalent transformation (4.41), we obtain a few kinds of representations of SO(2,l). First, if AQ, (BQ,) is always nonvanishing for any m in the representation, there is no truncated in the series of m. In this case, 1/4 5 Q < 0, and it is the representation D s ( Q ,mo) in the supplementary series for the unbounded spectrum :
+
+ +
1/2 5 j
< 0,
m = m fk, o
lmol <  j = 1/2  (&
+ 1/4)’12.
(4.48)
Second, if Q = j = m = 0, this is the identical representation. Third, if there is a lower boundary m for m in the representation, the lower o
172
Problems and Solutions in Group Theory
boundary lies only on the line of A 0 (or the equivalent line CB). It is the representation D + ( j ) for the spectrum bounded below where j < 0, m > 0, and mo = j:
j<O,
m=j+k.
Finally, if there is an upper boundary mo for rn in the representation, the upper boundary lies only on the line DO (or the equivalent line FB). It is the representation D  ( j ) for the spectrum bounded above, where j < 0, m < 0, and mo = j:
j<O,
m=jk.
Therefore, the dimension of any irreducible unitary representation of SO(2,1), except for the identical representation, is infinite.
Chapter 5
SYMMETRY OF CRYSTALS
5.1
Symmetric Operations and Space Groups
The fundamental character of a crystal is the spatial periodic array of the atoms composing the crystal, called the crystal lattice. By the periodic boundary condition, the crystal is invariant in the following translation:
*
aj
where [is called the vector of the crystal lattice. Three fundamental periods of the crystal lattice, which are not coplanar, are taken to be the basis vectors of the crystal lattice, or briefly called the lattice bases. The lattice bases are said to be primitive if any vector of the crystal lattice is an integral combination of the lattice bases. For simplicity, we only use the primitive lattice bases if without special indication.
The multiplication of two translations is defined to be a translation where two translation vectors are added. The set of all translations T ( 4 which leave the crystal invariant forms the Abelian translation group T of the crystal. Usually, in addition to the translation symmetry, a crystal has some other symmetric operations which leave the crystal invariant. A general symmetric operation may be a combinative operation composed of the spatial inversion, the rotation and the translation. Denoted by g(R, 6) a general symmetric operation, where R is a proper or improper rotation, R E0(3), and d is a translation vector, not necessary an integral combination of the lattice bases aj:
g(R,G)r = Rr
*
+ 6.
(54
Moving out the vector of the crystal lattice l f r o m G, we express the general
173
174
Problems and Solutions in Group Theory
symmetric operation as
g(R,a')= T(i+)g(R,, t)
t=
a' = e'+ t
(53)
C
j=l
3
ajtj,
o 5 tj < 1.
When R = E , a'has to be a vector of the crystal lattice zand g(E, = T ( 4 . From the definition (5.2), the multiplication of two symmetric operations satisfies
4
Thus, the inverse g(R,d)l and the identity E are
The set of all symmetric operations g(R,a') for a crystal with the multiplication rule (5.4) forms a group S, called the space group of the crystal. In the multiplication (5.4) of two symmetric operations, the multiplication RR' of two rotations does not matter with whether there are the translations a' and or not. Therefore, the set of the rotational parts R in g(R,a') forms a group G, called the crystallographic point group. For a given crystal with the space group S, it is easy to show by reduction to absurdity that t in the symmetric operation g(R,t) depends upon R uniquely. Generally, R is not an element of S, and G is not a subgroup of S. The space group S is called the symmorphic space group if G is the subgroup of S. In a symmorphic space group, any element can be expressed as g(R,o = T(C)R. From Eq. (5.4) we have
p
g(R,a')T(Gg(R,' )  1 = T(Ri+). a
(5.6)
Namely, the translation group T is the invariant subgroup of the space group S. Due to Eq. (5.3), the coset of T is completely determined by the rotation R. Thus, the quotient group of T with respect to S is the crystallographic point group, G = T / S . From Eq. (5.6), the action of any element R of G on any vector of the crystal lattice {has to be a vector of the crystal lattice 8
Re'= 2.
(5.7)
This is the fundamental constraint for the possible crystallographic point groups, the crystal systems, and the Bravais lattices.
Symmetry of Crystals
175
In the crystal theory, the lattice bases aj are usually chosen as the basis vectors in the real three dimensional space. The merit for this choice is that, due to Eq. (5.7), any matrix entry Rij of R in the bases is an integer:
*
i= 1
The shortcoming is that aj are generally not orthonormal. Define a set of basis vectors bj satisfying
bj is called the basis vector of the reciprocal lattice, or briefly called the reciprocal lattice basis. It is convenient in the crystal theory to express a rotation R in a doublevector form:
1. In the rectangular coordinate frame, write the doublevector forms and the matrix forms of the proper and improper sixfold rotations around the z axis. Solution. An improper sixfold rotation s around the z axis is equal to the 6 corresponding proper one c multiplying by a spatial inversion. Therefore, 6 their doublevector forms and the matrix forms are different only by a sign.
2. Express the directions of the proper and improper rotational axes in the groups T d and Oh by the basis vectors ej of the rectangular coordinate frame. Express the doublevector forms of the generators of the proper rotational axes in O h by ej.
contains four proper threefold axes, three improper fourfold axes, and six improper twofold axes. For Oh, all those axes are both the proper and improper ones.
Oh. T d
Solution. The group T d is a subgroup of the group
176
Problems and Solutions in Group Theory
The directions and the doublevector forms of the generators of the proper threefold axes are Direction : m e e2 {1
+ + e3},
e2
21= e2e1 + e3e2 + ele3,
#
#
Direction : m e  e2  e3}, {1 Direction : Direction :
2 2
D 4
= 281
+
e3e2
 eie3,
J17'5 J17'5
{el {el
+ e3},
R3 = e2e1  e3e2  e1e3,
+ e 2  e3},
Rq = e2e1  e3e2 + 6183.
# #
The directions and the doublevector forms of the generators of the proper fourfold axes are
+ +
Direction : Direction : Direction :
el,
2'1 = elel
#
+
e3e2
 e2e3,
e2,
e3,
T 2
+
+ e2e2 + e1e3, T3 = e2e1  e1e2 + e3e3.
= e3e1
The directions and the doublevector forms of the generators of the proper twofold axes are Direction : m e + e 2 } , (1 Direction : m e  e2}, {1 Direction : Direction :
3 = e2e1 + 8162  e3e3, 1
+
#
3 2
= e2e1  e1e2  e3e3,
J172
#
{e2
 e3},
3 = elel 4
4
 e3e2  ~
8 3 ,
Direction : m e e3}, {1 Direction :
+
$ 5
#
= e3e1  e 2 e 2
+ e1e3,
 e1e3.
@{el  e3) , 9 = e3e1 6
 e2e2
3. Write the doublevector form of a rotation R(n,w ) around the direction n through the angle w by the unit vector i and the unit vector n.
Solution. Let m be a unit vector orthogonal to n. Then, n x m is a unit vector orthogonal to both n and m . Applying R ( n , w ) to those three unit
Symmetry of Crystals
177
vectors, we have
t
In terms of the project operators nn and orthogonal to n, respectively, we have
+
1'
nn onto n and the plane
where the following formulas are used:
x (n x m) = m.
5.2
Symmetric Elements
point is called the symmetric center of a symmetric operation g(R,t) if the point is invariant in g(R,t ) . A straight line is called the symmetric straight line of a symmetric operation g(R,t) if the straight line is invariant in g(R,t ) . A plane is called the symmetric plane of a symmetric operation g(R,t) if the plane is invariant in g(R,t).
*A
A symmetric operation g(R,t) is called closed if its power may be equal to the identity, g(R,t)" = E , otherwise, it is called an open operation. There are only two kinds of open operations. One is g ( C N , t), N # 1, where CN is a proper Nfold rotation, and the nonvanishing component tll o f t along the direction n of the rotational axis is a multiple of the smallest vector of the crystal lattice in the direction n divided by N. This axis is called the screw axis, which is parallel to n through the point ro, where ro satisfies
( B  CN)ro = t  tll = tl.
tll # 0 is the gliding vector. A screw axis is a symmetric straight line of ~ ( C N . The other is g(S2,t ) , where S is an improper twofold rotation t) , 2 (reflection), and the component t l of t along the reflection plane is half of the smallest vector of the crystal lattice in the direction. This reflection
*
178
Problems and Solutions a Group Theory n
plane is called the gliding plane, which is orthogonal to the rotational axis n of the improper rotation S2 through the point ro, satisfying
( E  S 2 ) r o = t  t l = tll. t l is the gliding vector on the plane. A gliding plane is a symmetric plane of g(S2, t ) . The closed transformation has a symmetric center ro, satisfying
( E  R)ro = t.
When R is a proper rotation CN ( N # l),t has to be orthogonal to the rotational axis. The straight line which is through the point ro and parallel to n is a symmetric line. Each point on the symmetric line is the symmetric center for the closed operation. When R is the reflection 5’2, t has to be parallel to the rotational axis, i.e. orthogonal to the reflection plane. The plane which is through the point ro and orthogonal to the direction n of the rotational axis is a symmetric plane. Each point on the symmetric plane is the symmetric center for the closed operation. When R is an improper rotation S N ,N # 2, the closed transformation only has a symmetric center.
A symbol for a space group of a crystal in the international notations contains all information on the symmetry of the crystal. In the international notations, the first character indicates the Bravais lattice, f denotes the two order inversion group Ci,and the digit N or N denotes the proper or improper axis. Except for the rhombohedra1 system, the axis denoted by a digit without prime indicates the principal axis, along which the lattice basis a3 directs, the axis denoted by the digit 2 with a prime indicates a twofold axis orthogonal to the principal axis, along which the lattice basis a directs, and the axis denoted by a digit N > 2 with a prime or denoted 1 by the digit 2 with double primes indicates the rotational axis along other direction. In analysis of the symmetry of a crystal from its international symbol, one has to know the rule for choosing the lattice bases aj in the given crystal system. Whether aj are primitive or not depends upon the Bravais lattice. Denote by aj the length of aj, and by a3 the angle between a1 and a2, and so on. Let al, a2 and a3 construct the righthanded system. The rule for choosing the lattice bases aj is as follows. (1). Triclinic system contains P1 and Pi. The lattice bases aj are taken to be primitive. There is no restriction on the lengths and the angles of the lattice bases. (2). Monoclinic system contains P2, PZ, P f 2, A2, A?, and A f 2.
*
Symmetry of Crystals
179
Define the shortest vector of the crystal lattice along the twofold axis to be a3, and take two noncollinear smallest vectors of crystal lattice in the plane orthogonal to a3 to be a1 and a2, respectively. Two noncollinear vectors of the crystal lattice are called the smallest in a plane if any vector of the crystal lattice in the plane is their integral combination. The restriction for the lattice bases is a1 = a2 = n / 2 . ( 3 ) . Orthorhombic system contains P22', P2?, P f 22', C 2 2 / , C22', A22', C f 22', I22', I22', I f 22') F22', F25', F f 22'. Define the shortest vectors of the crystal lattice along three twofold axes to be a1 , a2 and a3, respectively. For CaV (23')) a 3 is along the proper twofold axis. The restriction for the lattice bases is a1 = a 2 = a 3 = ~ / 2 . ( 4 ) . Tetragonal system contains P 4 , Pa, P f 4 , P42', P42', P32', P32', P f 42', 14, 14, I f 4 , I42', I42', I32', I32', and I f 42'. Define the shortest vector of the crystal lattice along the fourfold axis to be a 3 , and take one shortest vectors of the crystal lattice in the plane orthogonal to a 3 to be al. Define a 2 = C4al. The restriction for the lattice bases is a1 = u2 and a1 = a 2 = a 3 = n / 2 . For the point group D2d, there are two different symbols 32' and 42", according to whether a1 directs along proper or improper twofold axis. (5). Cubic system contains P3'22', P3'22', P3'42', P3'32', P3'42', 13/22', I3'22', I3'42', I3'$2', I3'42', F3'22') F3'22', F3'42', F3'32', and F3'42'. Define three shortest vectors of the crystal lattice along three orthogonal twofold axes [for T (3'22') and T h (3'22')l or fourfold axes [for 0 (3'42'), T d (3'32') and O h (3'42')] to be al, a2 and a3, respectively. The restriction for the lattice bases is a1 = a2 = a3 and a1 = a2 = a 3 = n / 2 . ( 6 ) . Hexagonal system contains P3, P3, P32', P32', P32', P32', P32', P32", F 6 , P6, P f 6 , P62', P62', P62', P62' and P f 62'. Define the shortest vector of the crystal lattice along a threefold axis or a sixfold axis to be a3, and take one shortest vectors of the crystal lattice in the plane orthogonal to a 3 to be al. Define a2 = C3a1. The restriction for the lattice bases is a1 = u2, a1 = a2 = n / 2 and a 3 = 2 n / 3 . For D3, C3v and Dsd, there are two kinds of symbols 32' or 32', 32' or 32') and 32' or 32', where 2' (or 2') means that a1 directs along a twofold axis, and 2' (or 5') means that a1 directs along the angular bisector of two twofold axes. For D6 (62'))c 6 v (62') and D6h ( f 6 2 ' ) , a1 directs along a twofold axis. For D 3 h , there are two kinds of symbols 62' and 62", according to whether al directs along a proper or improper twofold axis. (7). Rhombohedra1 system contains R 3 , RS, R32', R32' and R32'. Three lattice bases are distributed symmetrically around the threefold
180
Problems and Solutions in Group Theory
axis with the same acute angle. The sum of three lattice bases is equal to the shortest vector of the crystal lattice along the threefold axis. The restriction for the lattice bases is a1 = a2 = a3 and a = a = a 3 . For D3 1 2 (32‘), CsV (32’) and D 3 d (32’),the map of any lattice basis onto the plane orthogonal to the threefold axis directs along the angular bisection of two twofold axes. For different Bravais lattices, in addition to the integral combinations t?, the vectors of the crystal lattice are allowed to be some fractional combinations f:
P type of Bravais lattice : R type of Bravais lattice
A type of Bravais lattice
: :
f = 0, f = 0, f = (a2 + as) /2,
f = (a1 + a2) /2,
B type of Bravais lattice : f = (a1 + as) /2,
C type of Bravais lattice :
I type of Bravais lattice
:
(5.10)
F type of Bravais lattice :
+ a2 + as) /2, f = ( a 2 + as) /2, (a3 + al) /2,
f = (a1
and
(a1
+ a2) /2.
+e2)
4. Let R be a rotation around the direction n = (el
/ A through
2r/3. Find the symmetric straight line for g(R,t) where (a) t = e3, (b) t = el + e 3 , If the symmetric straight line is a screw line, find its gliding vector. Simplify g(R,t) by moving the origin to the symmetric straight line for checking your result. Solution. Write the doublevector form of R = R(n,2~/3) n = with (el e 2 ) / \ / 2 in the rectangular coordinate frame:
+
(a) t = t l = e3.
Symmetry of Crystals
181
g(R,t ) has a symmetric straight line through the point ro and parallel to n, where ro satisfies g(R,t)ro = Rro + t = ro.
Write the equation in the rectangular coordinate frame:
Taking 702 = 0, we have 701 = and ~ 0 = 1/2. Each point on the 3 symmetric straight line is the symmetric center, Moving the origin to the point ro, we transform g(R,t) to g'
g' = T(ro)g(R, t)T(ro) = g(R, ro
+ t + Rro) = R.
(b) t = el
+
e3
g ( R , t) has a screw straight line which is through the point ro and parallel to n with the gliding vector tll, where ro satisfies
g(R,t l ) r o = Rro + t l = ro.
Write the equation in the rectangular coordinate frame:
and 703 = (1 m ) / 2 . Moving Taking 732 = 0, we have r01 = 1/2+ the origin to the point ro, we transform g(R,t ) to g'
5 . Let R be a reflection with respect to the zy plane. Find the symmetric plane of g ( R , t ) where (a) t = e3, (b) t = el +e3. If the symmetric plane is a gliding plane, find its gliding vector. Simplify g(R,t) by moving the origin to the symmetric plane for checking your result.
182
Problems and Solutions an Group Theory
Solution. R = aR(e3, n), where a is the spatial inversion. The doublevector form of R in the rectangular coordinate frame is
(a) t = tll = e3. g(R, t) has a symmetric plane through the point ro and orthogonal to the z axis, where ro satisfies
g(R, t)ro = Rro
+ t = ro.
Letting ro be on the z axis, we have ro = e3/2. Each point on the symmetric plane is the symmetric center. Moving the origin to the point ro, we transform g(R, t ) to g’
g’ = T(ro)g(R, t)T(ro) = g(R, t
+ ( R E)ro) = R.
(b) t = el e3. tll = e3 and t l = el. g(R, t) has a gliding plane which is through the point ro and orthogonal to the z axis with the gliding vector t l = el, where ro satisfies
+
Letting ro be on the z axis, we have ro = e3/2. Moving the origin to the point ro, we transform g(R,t) to g‘
6. Analyze the symmetry property of the crystal with the following space 111) ] , 2 group: No. 52 [D”,, P f 2( f $0)2‘( f f f)], No. 161 [civ, ’ ( 5 R32 and No. 199 [T5,1 3 ’ 2 ( i O ; ) 2 ’ ( i i O ) ] .Point (a) The general form of the symmetric operation; (b) The relations of directions and lengths among three lattice bases; (c) The symmetric straight line, symmetric plane and the gliding vector of the generator in each cyclic subgroup of the space group, if they exist; (d) The equivalent point to an arbitrary point r = alxl a222 a323 in the crystal cell.
+
+
Solution. Let us analyze the property of the space groups, respectively. (1) The space group D;h = P f 21/2,1/2,02~~2,,/2,,/2 belongs to P type of Bravais lattice of the tetragonal system. Its crystallographic point group is D2h, which contains three orthogonal twofold axes, both proper and
Symmetry of Crystals
183
improper. Three primitive lattice bases are along three twofold axes, respectively. a = a = a3 = 7r/2 but the lengths of the lattice bases may 1 2 not be equal to each other. The doublevector forms of the generators are
The arbitrary element in the space group is expressed as
where S denotes the spatial inversion, and n1, n2 and 7x3 are taken to 1 be 0 or 1, respectively. The origin is the symmetric center of the spatial inversion. g(C2,q) with q = q l = (a1 a 2 ) / 2 has a symmetric straight line through the point ro and parallel to a3. Each point on the symmetric straight line is the symmetric center. Let the components of ro along the . 3 lattice bases be r01, 7 3 2 , and ~ 0 3 Take ~ 0 = 0. Thus,
+
We have ro1 = ro2 = 1/4. From another viewpoint, since the spatial 1 inversion S is a symmetric operation, there is also an improper twofold axis around the direction a3. Slg(C2, q) has a gliding plane, which is through the origin and orthogonal to a3 with the gliding vector q. 9Wi, P) with P = P1 + P1, I
has a screw axis which is through the point rb and parallel to a1 with the gliding vector P I [ .The component of rb along the direction a1 can be taken to be zero, and the components along a2 and a 3 are denoted by rb2 and rb3, satisfying:
Thus, r& = rb3 = 1/4. Similarly, there is also an improper twofold axis along the direction a l . SIg(Ci, p) has the gliding plane which is through the point p11/2 = a1/4 and orthogonal to a1 with the gliding vector PI.
184
Problems and Solutions a Group Theory n
From an arbitrary point r = a l x l a 2 2 2 a 3 2 3 in the crystal cell, we can obtain eight equivalent points by the symmetric operations. Due to the spatial inversion, the equivalent points are paired by different signs:
+
+
(2) The space group Cgw= R3?i/2,1/2,1/2 belongs to the R type of Bravais lattice in the Rhombohedra1 system. The crystallographic point group CsW contains a proper threefold axis, which is the principal axis, and three improper twofold axes, distributed symmetrically on the plane orthogonal to the principal axis. Recall that those improper twofold rotations are reflections with respect to the planes containing the principal axis. Three lattice bases have the same length and distributed around the principal axis symmetrically :
The doublevector forms for the generators are
The general element in the space group is
where m is taken to be 0, 1 or 2, n is 0 or 1. The proper threefold axis is the principal axis. The origin is taken on the principal axis. Each point on the principal axis is the symmetric center for the threefold rotation. g(Sa, (a1 a2 a3)/2) with p = p l = (a1 a2 a3)/2 has a gliding plane which is through the origin and orthogonal to the direction a1  a2 with the gliding vector PI. From an arbitrary point r = a l x l a 2 2 2 a 3 2 3 in the crystal cell, we
+ +
+ +
+
+
Symmetry of Crystals
185
obtain six equivalent points by the symmetric operations:
(3) The space group T5 = 1 3 ' 2 1 ~ ~ , 0 , 1 ~ 2 2 ~ belongs2to the I type of ~2,1~ ,0 Bravais lattice in the cubic system. The crystallographic point group T contains three orthogonal proper twofold axes and four proper threefold axes distributed symmetrically. Three lattice bases with the same length are along three twofold axes, respectively. The lattice bases are not primitive. The following fractional combination is allowed to be a vector of the crystal lattice:
f = (a1
+ + as) /2.
a2
The doublevector forms of generators are
The general element of the space group is
where 121,122 and 123 are taken to be 0 or 1,M is 0, 1 or 2. The origin is on the threefold axis. Ci is a pure rotation. g(C2, q) with q = 91 q i 1
+
9 = a3/2, 1 1
ql = a l p ,
has a screw axis which is through the point ro and parallel to the direction a3 with the gliding vector 91 The components of ro along the lattice bases 1. are denoted by ~ 0 1 702 and ~ 0 3 . We may choose r03 = 0. 701 and r02 , satisfy :
186
Problems and Solutions in Group Theory
Thus, POI = 1/4 and r02 = TO3 = 0. g(Ci,p) with p = p11 p i ,
+
has a screw axis which is through the point rb and parallel t o the direction a with the gliding vector p11. 1 The components of rb along the lattice bases are denoted by r&, 7b2 and rb3. We may choose rbl = 0. 7b2 and rb3 satisfy:
O = (Ci  E ) rb
+ p i = {2rb2 + 1/2} a2  2rb3a3.
Thus, rb2 = 1/4 and rbl = rb3 = 0. From an arbitrary point r = a121 + a222 + a 3 2 3 in the crystal cell we obtain 24 equivalent points by the symmetric operations, where half points can be calculated by a translation f . The remaining 12 points are:
5.3
International Notations for Space Groups
Attaching as the subscripts the suitable translation vectors to the international symbol for the symmorphic space group, we obtain the international symbol for the general space group. The general element of a space
*
Symmetry of Crystals
187
group can be expressed as
(5.11)
where n, n1 and n2 all are integers, L = f is a vector of crystal lattice, f depends upon the Bravais lattice, and the translation vectors t, p and q are the fractional combinations of the lattice bases aj:
t+
The translation vectors t, p and q should satisfy the group property, i.e., the multiplication of two elements of the space group has to be expressed in the same form (5.11). First, due to the property of a cyclic subgroup, the translation vector t in g(R,t) has to satisfy
t = O
{n (n . t)} = tll = mall/N
when R = C1 = E , when R = C N , N when R = S2, when R = S N , N
# 1,
# 2.
{t  i (n  t)} = t l = m a 1 / 2 i
(5.12)
\ I
t no restriction
Second, there are constraints resulted from the multiplication of R, R1 and R2. For example, due to R1 = RRIR, we have
g(R1, p + L) = g(R, t)g(R1, p)g(R,t) = S(RRlR,t
+ RP + RRlt)
( R  E)p = L  ( E + RR1)t.
(5.13) At last, we have to avoid the equivalent symbols related to the different choice of the origin. Let the vector from the origin 0 to the new origin 0' be ro. If the symmetric operation is denoted by g(R,t) for the origin 0, it is changed to g' in the new origin 0'
g' = T(ro)g(R,t)T(ro) = g(R, t
+ ( R E)ro).
(5.14)
The different choice of the origin does not change the space group, but changes its symbol. Those two symbols are called the equivalent symbols. We have to remove the equivalent symbols for the same space group.
188
Problems and Solutions in Group Theory
Usually, we choose the origin such that the translation vector t in Eq. (5.11) becomes as simple as possible. In other words, we choose the origin such that the part of t , which is not restricted in Eq. (5.12), vanishing. t can be more simplified if the remaining t is a component of L. Then, we may choose the origin again to simplify the translation vectors p and q under the condition that t keeps invariant. This is the general method to determine the symbol of a space group. Since the choice of the origin first depends upon the simplification of t in g(R,t), The space groups are classified as type A ( R = S N , N # a), type B ( R = C N )and type C ( R = ,572). There are only two space groups in type C: PZ;,, and AZgoo.
*
7. Calculate the 12 space groups related to the crystallographic point group
D2d
Solution. The point group D 2 d contains an improper fourfold axis (the principal axis), two proper twofold axes and two improper twofold axes. Four twofold axes are distributed symmetrically on the plane orthogonal to the principal axis. The lattice basis a3 is along the principal axis. According to whether the lattice basis a is along a proper or an improper twofold 1 axis, the international symbol of D 2 d is denoted by 32’ or 32”, namely, denoted by the product of two subgroups S4C4 or S4Cg. The space groups belong t o the tetragonal system, where there are two Bravais lattices P and I . The restriction for the lattice bases are a1 = a 2 = a 3 , and a1 = u2. The doublevector form of the fourfold rotation is
#
g 4
= a2bl
+ a l b 2  a3b3.
The space groups with the crystallographic point group D 2 d belong to type A. Choosing the origin on the principal axis, we express the general element in the space group as
where m is taken 0, 1, 2 or 3, n is 0 or 1. C in the formula may be changed L to C;, depending on the discussed space groups. We may translate the origin by ro, ro = alrO1 a2702 a3703, satisfying
+
+
Thus, ro may be taken the following values, in addition to L (a1 + a 2 ) / 2 , a3/2 Bravais lattice P or I ro = Bravais lattice I . (2al as) /4, (2a2 as) /4,
{
+
+
Symmetry of Crystals
189
In the following we study two kinds of space groups with the crystallographic point groups 32' and ;?2", respectively. (1) For the crystallographic point group 32', a1 is along the direction of a proper twofold axis. The doublevector form of the twofold rotation is
Since S4C;S4 = Ci, we have
For the P type of Bravais lattice, we have solutions pl = p2 = 0 or 1/2, and p3 = 0 or 1/2. For the I type of Bravais lattice, because f = (a1 a2 + a s ) /2 is a vector of crystal lattice, the solution p = f is ruled out, and two solutions p = (a1 a 2 ) /2 and p = a3/2 become equivalent to each other. In addition, there are new solutions p = (2al + as) /4 and p = (2a2 as) /4. In all those solutions, the parallel component p11is equal t o multiple of half vector of crystal lattice along the rotational axis. Moving the origin by ro, p may be changed by:
+
+
+
For the P type of Bravais lattice, p is not simplified. But for the I type of Bravais lattice, the solution p = a3/2 is ruled out, and two solutions p = (2al as) /4 and p = (2a2 a 3 ) /4 become equivalent. Therefore, we have the following space groups:
+
+
2 2 2
2 4
which are listed in the table of space groups with Nos. 111, 112, 113, 114, 121 and 122. (2) For the crystallographic point group 32", a is along the direction of 1 an improper twofold axis. The doublevector form of the twofold rotation is
190
Problems and Solutions in Group Theory
Since S4C;S4 = C;, we have
Although this equation is the same as the former case (l), solution may the be different from the former case because of the different lattice bases. For the P type of Bravais lattice, we have the same solutions as the former case: pl = p2 = 0 or 1/2, and p3 = 0 or 1/2. For the I type of Bravais lattice, we only have solutions p = a 3 / 2 or 0. The solutions p = (2al + a s ) /4 and p = (2a2 + a s ) /4 are ruled out because their parallel component p1 = (a1  a2) /4 is less than half of the vector of crystal lattice along i the rotational axis. Moving the origin by ro, p may be changed by:
Both for the P type and the I type of Bravais lattice, p is not simplified. Therefore, we have the following space groups: P42",
P42bl,$, p42; f r o ,
pq2'i L L ,
2 2 2
142",
142i0$,
which are listed in the table of space groups with Nos. 115, 116, 117, 118, 119 and 120.
8. Please calculate the 9 space groups related to the crystallographic point group D2.
Solution. The international symbol of the crystallographic point group 0 2 is 22'. It contains three orthogonal proper twofold axes and can be expressed as the product of two subgroups: C2Ci. The space groups with the crystallographic point group 0 2 belong to the orthorhombic system with P , A , F and I types of Bravais lattices. Three lattice bases are orthogonal to each other, but have different lengths. The doublevector forms for twofold rotations are
.
6 2
= albl  a2b2
+ a3b3,
C = albl  a2ba  a3b3. h
+ +
The space groups with the crystallographic point group D2 belong to type B. Choosing the origin on one of the twofold axes, we express the general element in the space group as
S y m m e t r y of Crystals
191
where both rn and n are taken to be 0 or 1. t = ta3 with t = 0 or 1/2. For the A type, F type and I type of Bravais lattices, L contains the component a3/2 so that t can be removed by a translation ro of the origin
Under the condition that the simplified t keeps invariant, the origin can be made a further translation ro
ro =
{
~ 1 ' ~ a2/2, )
r03a3
a1 + a21 /4,
Bravais lattices P, A , F and I Bravais lattice F,
where ~ 0 is arbitrary. 3 Since C2C~C2 Ci and C2Cit = t, we have =
Letting p = alp1
+ a2p2 + a3p3, we have:
2 For P , A , F and I types of Bravais lattices, we obtain that pl and p are respectively taken to be 0 or 1/2, but p is arbitrary. There is an additional 3 solution for the F type of Bravais lattice: p = (a1 a2) /4. However, the latter is ruled out by the constraint that p11has to be a multiple of half vector of crystal lattice along the rotational axis. In translating the origin by ro, p is made the following transformation:
+
Thus, p is removed. Now, we study the space groups for different Bravais 3 lattices. For the P type of Bravais lattice, t , pl and pz are allowed to be 0 or 1/2 independently. However, some space groups may be equivalent to each other. Obviously, two space groups with t = pl = p = 0 and 1/2 are 2 inequivalent. They are denoted by P22' and P200+2/,,,,listed in the table 2 2 of space groups with Nos. 16 and 19. If one of t , pl and p is 1/2 and the 2 remaining are vanishing, we obtain three equivalent space groups because three twofold axes are distributed symmetrically. In other words, three solutions t = 1/2 (pl = p = 0), pl = 1/2 ( t = p = 0) and p = 1/2 2 2 2 ( t = pl = 0) are equivalent because one can be obtained from the other by interchanging the lattice bases. Those three equivalent space groups are
192
Problems and Solutions an Group Theory
denoted by P22b4,, listed in the table of space groups with No. 17. If two of t , pl and p2 are 1 / 2 and the third one is vanishing, we also obtain three equivalent space groups. In other words, the solution with t = p = 1 / 2 can a be obtained from the solution t = pl = 1 / 2 by changing a1 to the direction of a2. For the solution with pl = p2 = 1/2 and t = 0, two twofold rotations are expressed as
After moving the origin by ro = a2/4, they become
Then, interchanging a1 and a3, the solution becomes that with t = p2 = 1 / 2 and pl = 0. Those three equivalent space groups are denoted by P22',,,, 2 2 listed in the table of space groups with No. 18. For the A type of Bravais lattice, t = 0. Since f = (a2 + a s ) / 2 is a vector of crystal lattice, the solution with p2 = 1 / 2 can be removed. Thus, there are only two space groups for the A type of Bravais lattice A22' and A22iO0,listed in the table of space groups with Nos. 21 and 20. For the F type of Bravais lattice, t = 0. Since f = (a1 +a3) /2 and (a2 as) / 2 are the vectors of crystal lattice, the solutions with pl = 1 / 2 and (or) p2 = 1 / 2 can be removed. Thus, there is only one space group for the F type of Bravais lattice, F22', listed in the table of space groups with No. 22. For the I type of Bravais lattice, t = 0. Since f = (a1 + a 2 + a s ) /2 is a vector of crystal lattice, the solution with pl = p2 = 1/2 can be removed. The solution with pl = 0 and p2 = 1/2 can be obtained from that with pl = 1/2 and p2 = 0 by changing a1 to the direction of a2. Thus, there are only two space groups 122' and 122; for the 1 type of 200 Bravais lattice, listed in the table of the space groups with Nos. 23 and 24. Usually, the space group with No. 24 in the table of the space groups are denoted by 1200t26t0, which is equivalent to 122iO0. One will convince himself by subtracting f from p, removing p3 through moving the origin, and interchanging a1 and a 3 . In summary, there are nine inequivalent space groups related to the crystallographic point group D2, listed in the table of space groups with NOS. 1624.
+
Chapter 6
PERMUTATION GROUPS
6.1
Multiplication of Permutations
A rearrangement of n objects is called a permutation. Denote a permutation R which transforms the object in the position j into the position rj by a 2 x n matrix
*
R=
(rtr:::::).
In the matrix for R the order of columns does not matter, but the corresponding relation between two digits in each column is essential. The multiplication rule for two permutations is defined as successive applications of the two transformations.
There are n! permutations for the system of n objects. The set of those n! permutations forms a group, called the permutation group S n of n objects, where the identity and the inverse of R are
If a permutation S keeps ( n  l ) objects invariant and changes the remaining l?objects in order, S is called a cycle with length l , described by a onerow matrix:
193
194
Problems and Solutions in Group Theory
In the onerow matrix for a cycle, the order of digits is essential, while the transformation of digits in sequence is permitted. The order of a cycle S with length l is l , namely, l is the smallest power with S = E . A cycle ' with length 2 is called a transposition. Any permutation can be decomposed into a product of cycles which contain no common object. The product of two cycles without any common object is commutable. Up to the order, the decomposition of a permutation into a product of cycles without any common object is unique. The set of lengths of those cycles is called the cycle structure of the permutation. A cycle is equal to a product of two cycles by cutting it in any digit and repeating the digit:
*
(a

a
*
bcd
f) = ( a ... b c ) ( c d .  . f).
Conversely, two cycles which contain one common object may be connected to be one cycle through Eq. (6.1). The product of two cycles which contain more common objects can be simplified as a product of several cycles without any common object by using Eq. (6.1) repeatedly. For example,
a2+1 . . . a j d)(d bl . . . b, c b,+l . . . b,) ai c ) ( c ai+1 . . . aj d)(d bl . . . b, c)(c b,+l . . . b,) a c)(ai+l . . . aj d)(d C ) ( C d)(d bl ... b,)(c b,+l ... b s ) i ai c ) ( c b,+l . . . b,)(Ui+l . . . a j d)(d bl . . . b,) = (a1 a c b,+l ... bs)(ai+l ... aj d bl ... b,). i (6.2) A cycle with length l may be decomposed into a product of ( l  1) transposition by Eq. (6.1). Any permutation may be decomposed into the product of transpositions by different ways, where the number of transpositions in the product is not unique, but it is determined whether the number is even or odd. A permutation is said to be even (or odd) if its decomposition contains even (or odd) number of transpositions. The permutation parity S(R) of an even permutation is 1, and that of an odd permutation is 1. The permutation parity of a cycle with length l is ( l  1).
. . a
(a1 . . . a c 2 = (a1 . . . = (a1 . . . = (a1 . . .
When a permutation S is moved from the leftside of a permutation R to its rightside, the digits (objects) in R are made the permutation S. Conversely, when a permutation S is moved from the rightside of R to its leftside, the digits in R are made the inverse permutation Sl. In other words, if the permutation R is a cycle (or the product of cycles) containing some digits rj, and the permutation S transforms rj into t j , then, T =
*
Permutation Groups
195
SRSl is the same cycle as R except that the containing digits rj are changed to t j , respectively. This property in the product of permutations is called the interchange rule. From the rule, two permutations with the same cycle structure are conjugate to each other. The class in a permutation group is described by the cycle structure of the permutations in the class.
Denote by Pa = (a a 1) the transposition of two neighbored objects a and ( a + 1). It is easy to know from the interchange rule in the product of permutations that Pa satisfy
*
+
1. Simplify the following permutations into the product of cycles without any common object:
(1) : (1 2)(2 3)(1 2), (2) : (1 2 3)(1 3 4)(3 2 l ) ,
(3) : (1 2 3 4)1, (5)
:
(4) : (1 2 4 5)(4 3 2 6),
(1 2 3)(4 2 6)(3 4 5 6).
Solution: The first two problems can be solved by the interchange rule, and the last three problems can be solved by Eq. (6.1). (1) : (1 31, (4) : (5 1 2 6)(4 3), (2) : (2 1 4 ) ,
(3) : (4 3 2 l ) ,
(5) : (1 2 6)(4 5).
2. Decompose an arbitrary transposition ( b d) as a product of the trans
positions P a of the neighbored objects. Solution. Without loss of generality, we suppose b
> d and obtain
3. Prove that the order of a cycle R with length l is
! namely, Re = E . ,
Solution. Let R = (a1 a2 . .  a [ ) and b be an object different from any am, 1 5 m 5 l . Then, R transforms am into am+l, and into a l . Thus, Re" transforms am into a t , R transforms into a l , and Rml transforms a1 into am Altogether, Re transforms an arbitrary am into itself, while keeping b invariant, namely, Re is the identity E .
196
Problems and Solutions in Group Theory
4. Prove that PI = (1 2) and W = (1 2 3
. ..
n) are two generators of the
permutation group S,.
Solution. From Problem 2, any permutation can be decomposed as a product of the transpositions P, of the neighbored objects. From the interchange rule, WP, = P,+1W. From Problem 3 we have W' = W n  l , so P, = WPu_1W"1 = WalPlW,a+l.
5 . There are 52 pieces of playing cards in a set of poker. The order of cards is changed in each shuffle, namely, the cards are made a permutation. If the shuffle is "strictly" done in the following rule: first separate the cards into two parts in equal number, then pick up one card from each part in order. After the strict shuffle the first and the last cards do not change their positions, while the remaining cards are rearranged. Try to find this permutation, to decompose it into a product of cycles without any common object, to write the cycle structure of the permutation, and to explain a t least how many times of the strict shuffles will make the order of cards into its original one.
Solution. After the strict shuffle, the cards in the first part are arranged into the odd positions, while the cards in the second part are arranged into the even positions. In the mathematical language, after a shuffle, the nth card is changed to the (2n  1)th position, while the (26 n)th card is changed to the (2n)th position, where 1 5 n 5 26. By this rule, we want to decompose the permutation corresponding to the strict shuffle into the product of cycles without any common object. First, this permutation contains two cycles (1) and (52) with length 1. Second, according to the rule, the shuffle transforms 2 into 3, 3 into 5, 5 into 9, 9 into 17, 17 into 33, 33 into 14, 14 into 27, and 27 into 2, so we obtain a cycle with length 8
+
(2 3 5 9 17 33 14 27).
We arbitrarily choose a card in the remaining cards, say the fourth card. The shuffle transforms 4 into 7, 7 into 13, 13 into 25, 25 into 49, 49 into 46, 46 into 40, 40 into 28, and 28 into 4, so we obtain another cycle with length 8
(4 7 13 25 49 46 40 28).
Similarly, in the remaining cards, we choose the sixth card. The shuffle transforms 6 into 11, 11 into 21, 21 into 41, 41 into 30, 30 into 8, 8 into 15,
Permutation Groups
197
15 into 29, and 29 into 6, so we obtain a cycle with length 8 (6 11 21 41 30 8 15 29).
In the remaining cards, we choose the tenth card. The shuffle transforms 10 into 19, 19 into 37, 37 into 22, 22 into 43, 43 into 34, 34 into 16, 16 into 31, and 31 into 10, so we obtain a cycle with length 8
(10 19 37 22 43 34 16 31).
In the remaining cards, we choose the twelfth card, the shuffle transforms 12 into 23, 23 into 45, 45 into 38, 38 into 24, 24 into 47, 47 into 42, 42 into 32, and 32 into 12, so we obtain a cycle with length 8
(12 23 45 38 24 47 42 32).
In the remaining cards, we choose the 18th card, the shuffle transforms 18 into 35, and 35 into 18. This time we obtain a cycle (18 35) with length 2. In the remaining cards, we choose the 20th card, the shuffle transforms 20 into 39, 39 into 26, 26 into 51, 51 into 50, 50 into 48, 48 into 44, 44 into 36, and 36 into 20, so we obtained a cycle with length 8
(20 39 26 51 50 48 44 36).
Now, 52 cards are exhausted, and the permutation corresponding to the shuffle is decomposed into the product of two cycles with length 1, one cycle with length 2 and six cycles with length 8. All those cycles do not contain any common digit. The cycle structure of this permutation is (12, 2, @). The least common multiple of the cycle lengths is 8, namely, after 8 times of the strict shuffles, the order of playing cards turns into its original one. In fact, the shuffle is not so strict. Since many accidental cases occur, there is no worry about.
6.2
Young Patterns, Young Tableaux and Young Operators
A class in the permutation group S, is described by the cycle structure = ( e l , &, . . .) of the permutations in the class. Since l!j = n, ( l )is a partition number of n. For a finite group, the number of the inequivalent irreducible representations is equal to the number of the classes. Therefore, the irreducible representation of S, can also be described by a partition
(e)
*
Cj
198
Problems and Solutions in Group Theory
number of n, denoted by [A] = [ A 1 , A 2 ,  .  ] , A1 2 A 2 2 .   , C.A. = n. 3 3 Usually, when some k‘j or A j are duplicated, one may denote them by an exponent, for example, (23,32) = ( 2 , 2 , 2 , 3 , 3 ) . From a partition number [A] of n, we define a Young pattern which consists of n boxes lined up on the top and on the left, where the j t h row contains A j boxes. The Young pattern is also denoted by [A]. Remind that in a Young pattern, the number of boxes in the upper row is not less than that in the lower row, and the number of boxes in the left column is not less than that in the right column. A Young pattern [A] is said to be larger than a Young pattern [A’] if X i = A:, 1 5 i < j, and X i > A>. There is no analytic formula for the number of different Young patterns with n boxes. However, one can list all different Young patterns with n boxes from the largest to the smallest. Filling n digits 1, 2, . . ., n arbitrarily into a given Young pattern [A] with n boxes, we obtain a Young tableau. A Young tableau is said to be standard if the digit in each column of the tableau increases downwards and the digit in each row increases from left to right. For two standard Young tableaux with the same Young pattern, if the digits filled in the first (i  1) rows and in the first ( j  1) columns of the i t h row are the same as each other, respectively, but the digits filled in the box at the j t h column of the ith row are different, the Young tableau with the larger digit in this box is said to be larger than the other. This increasing order for the standard Young tableaux is the socalled dictionary order. We can enumerate the standard Young tableaux from the smallest to the largest by an integer p. The number of the standard Young tableaux with a given Young pattern [A] may be calculated by the hook rule,
*
*
where hij is the hook number of the box in the j t h column of the ith row of the given Young pattern, which is equal to the number of boxes at its right in the ith row, plus the number of boxes below it in the j t h column, and plus one. Two Young patterns related by a transpose are called the associated Young patterns. The corresponding standard Young tableaux for two associated Young patterns are related by a transpose such that the larger Young tableau for one Young pattern becomes the smaller for the associated Young pattern, but the numbers of the standard Young tableaux for two associated Young patterns are the same.
Permutation Groups
199
A permutation of the digits in the same row of a given Young tableau is called the horizontal permutation P of the Young tableau, and a permutation of the digits in the same column is called its vertical permutation Q. Generally, a horizontal permutation P may be a product of the horizontal permutations Pj, where Pj is a permutation of digits in the j t h row. Similarly, Q may be a product of Q k , where Q k is a permutation of digits in the Icth column. The sum of all horizontal permutations of a given Young tableau is called its horizontal operator P = C P = C P’, and the sum of all vertical permutations multiplied by their permutation parities is called its vertical operator Q = CS(Q)Q = C d ( Q k ) Q k . The Young operator Y corresponding to the Young tableau is equal to the product of the horizontal operator and the vertical operator, Y = PQ. A Young operator is said to be standard if its Young tableau is standard. Since for a given Young operator Y , its Young tableau and its Young pattern have been determined, we often say the Young pattern Y and the Young tableau Y for convenience. P and Q are said to be the horizontal permutation and the vertical permutation of the Young operator Y (or the Young tableau Y ) , respectively, and PQ is said to be a permutation belonging to the Young operator Y (or the Young tableau Y ) . Except for the identity, no permutation can be both the horizontal permutation and the vertical permutation belonging to the same Young tableau.
*
nj
nk
The Young operator Y can be expressed as a combination of its permutations, Y = CG(Q)PQ,and satisfies
*
[,c ,
x
,=1
(a, h,)] Y = 0,
Y [E p=l
(c,
4
=0 7
(6.5)
where the last two formulas are called the Fock conditions. The sum in the first Fock condition runs over all X objects a, in a given row, while b, is any object in another row with less boxes, A’ 5 A. Similarly, the sum in the second Fock condition runs over all r objects c, in a given column, while d, is any object in another column with less boxes, r‘ 5 r. If there exist two digits filled in the same row of a Young tableau y and in the same column of another Young tableau Y’, then Y’Y = 0. Two Young operators Y and Y‘ with different Young patterns are orthogonal to each other, YY’ = Y‘Y = 0. For a given Young pattern, if a standard Young tableau Y , is larger than another standard Young tableau y,, then
200
Problems and Solutions in Group Theory
YpY, = 0.
If a permutation R ES, transforms one Young tableau y into another Young tableau y’ with the same Young pattern [A] for the permutation group S,, the corresponding Young operators satisfy y‘ = RyRl. R can be calculated in the following way: Put the digits filled in the Young tableau y to the first row of the tworow matrix for R, and the digits in the Young tableau y’ to the second row in the same order, namely, in each column of the tworow matrix for R, the upper digit and the lower digit are filled in the same box of two Young tableaux, respectively. The necessary and sufficient condition for y’y # 0 is that R belongs to the Young tableau y , namely, R is a product of the horizontal permutation P and the vertical permutation Q of the Young tableau Y . Note that P‘ = RPRl and Q’ = RQR’ are respectively the horizontal one and the vertical one of the Young tableau y‘, and R = PQ = P‘Q’ = Q‘P. The square of a Young operator y corresponding to the Young pattern [A] for S, is
*
where d [ x(S,) is the number of the standard Young tableaux for the Young ~ pattern, given in Eq. (6.4).
6. Write all Young patterns of the permutation groups the largest to the smallest, respectively.
s5, s 6
and
s7
from
Solution. For the permutation group S g , its Young patterns are [5], [4,1], [3,21, [3, i23,[z2, 11, P, i31,p51. For the permutation group s6, its Young patterns are [6], [5,1], [4,2], [4,121, ~ 2 1 [3,2,11,[3,131, 1231, ~ 2 i21,12,141, ~161. , , For the permutation group S7, its Young patterns are [7], [6,1], [5,2], [5, i21, [4,31, [4,2,11, [4, i31, [32,11, 13,221, [3,2,121, [3,141, [23,11, [22,131, P, 151, ~ 7 3 .
7. Calculate the number n(C) of elements in each class (C) of the permutation groups S4, S5, s and 6
S7,
respectively.
Solution. The number n(C) of elements in the class (C) for S4: n(4) = 6, n(3,l) = 8, n(22) = 3, n(2, 12) = 6, n(14) = 1. The number n(C) of elements in the class (C) for S5: n(5) = 24, n ( 4 , l ) = ) ) ) 30,n(3,2) = 2 0 , ~ ( 3 , 1 2= 2 0 , 4 2 2 , q = 1 5 , ~ ( 2 , 1 3= i 0 , ~ ( 1 5 = 1. The number n(C)of elements in the class (C) for SS: n(6)= 120, n ( 5 , l ) = 144, n(4,2) = 90, n(4, 12) = 90, 71(3~) 40, n(3,2,1) = 120, n(3, 13) = 40, =
Permutation Groups
20 1
4 2 3 ) = 1 5 , ~ ( 2 2 , 1 2= 4 5 , 4 2 , 1 4 ) = 15, n(i6) = 1. )
The number n ( l ) of elements in the class ( l ) for S7: n(7) = 720, n(6,l) = 840, n(5,2) = 504, n(5, 12)= 504, n(4,3) = 420, n(4,2,1) = 630, ~ n(4, 13)= 210, n(32,1) = 280, ~ i ( 3 , 2 = )210, n(3,2, 12) = 420, n(3, 14)= 7 0 , 4 2 3 , i ) = 105,n(22,13) = 1 0 5 , ~ ( 2 , 1 5= 2 1 , 4 1 7 ) = 1. )
8 . Calculate the number dlA](S,) of the standard Young tableaux for each Young pattern [A] of the permutation groups S,, 5 5 n 5 9.
Solution. We calculate the number d[A](S,) of the standard Young tableaux for the Young pattern [A] by the hook rule (6.4). Remind that the associate Young patterns [A] and [ ]the same number of the standard i have Young tableaux.
202
Problems and Solutions in Group Theory
9. Write the Young operators corresponding to the following Young tableaux:
(1) :
(2) :
F pl l (3) :
Solution.
(1) : [E (1 2) (1 3) (2 3) (1 2 3) (3 2 l ) ] [E  (1 4)] = E (1 2) (1 3) (2 3) (1 2 3) (3 2 1) (1 4)  (2 1 4 )  (3 1 4 )  (2 3)(1 4)  (2 3 1 4 )  (3 2 1 4 ) .
+ + + + + + + + +
+ +
(2) [E+ (1 211 [E + (3 4)1 [E  (1 3)1 [E  (2 4)i = E (1 2) (3 4) (1 2)(3 4)  (1 3)  (2 1 3)  (4 3 1)  (2 1 4 3)  (2 4)  (1 2 4)  (3 4 2)  (1 2 3 4) (1 3)(2 4) (1 3 2 4) (3 1 4 2) (3 2)(1 4).
+
+
+ + (3) : E + (1 2) + (1 3) + (1 4) + (2 3) + (2 4) + (3 4) + (1 2)(3 4) +(1 3)(2 4) + (1 4)(2 3) + (1 2 3) + (1 3 2) + (1 2 4) + (1 4 2) +(1 3 4) + (1 4 3) + (2 3 4) + (2 4 3) + (1 2 3 4) + (1 2 4 3) +(13 4 2) + (1 3 2 4) + (1 4 2 3) + (1 4 3 2)  (1 5)  (2 1 5)
+
+
(3 (3 (3 (2 (3
1 5)  (4 1 5)  (2 3)(1 5)  (2 4)(1 5)  (3 4)(1 5) 4)(2 1 5)  (2 4)(3 1 5)  (2 3)(4 1 5)  (2 3 1 5) 2 1 5)  (2 4 1 5)  (4 2 1 5)  (3 4 1 5)  (4 3 1 5) 3 4)(1 5)  (2 4 3)(1 5)  (2 3 4 1 5)  (2 4 3 1 5) 4 2 1 5)  (3 2 4 1 5)  (4 2 3 1 5)  (4 3 2 1 5).
10. Write five standard Young tableaux Y p corresponding to the Young pattern [3,2] of S5 from the smallest to the largest, and calculate the permutations R,, transforming the Young tableau y , into Young tableau &.
Solution. The standard Young tableaux pattern [3,2] of S5 are as follows:
Y , corresponding to the Young Y4 Y5
Yl
Y2
Y3
Permutation Groups
203
11. Calculate the permutation R12 transforming the following Young tableau & to Y1, and check the formulas 7'1R12 = R127'2, QlR12 = R12Q2 and YlR12 = R12Y2, where
Solution.
71 1 'R2
R12
=
1243 ( 1 2 3 4 ) =(34)'
+ (1 2) + (1 3) + (2 3) + (1 2 3) + (3 2 l)](3 4) = (3 4) + (1 2)(3 4) + (1 3 4) + (2 3 4) + (1 2 3 4) + (2 1 3 4), R127'2 = (3 4) [E + (1 2) + (1 4) + (2 4) + (1 2 4) + (4 2 l)] = (3 4) + (1 2)(3 4) + (3 4 1) + (3 4 2) + (3 4 1 2) + (3 4 2 l ) ,
= [E
QlRl2 R12Q2
= [ E  (1 4)] (3 4) = (3 4)  (1 4 3),
= (3 4) [ E  (1 3)] = (3 4)
 (4 3 1).
Thus,
12. It can be seen that there is no pair of digits filled in the same row of the Young tableau y and in the same column of the Young tableau y'. Calculate the permutation R transforming the Young tableau y into the Young tableau Y', and express R as PQ belonging to the Young tableau y , and as P'Q' belonging to the Young tableau Y':
The Young tableau Y
The Young tableau
y'
204
Problems and Solutions in Group Theory
Solution.
R = ( 1 2 43743556 7 8 9 = ( 4 3 5 6 8 9 7 ) 12 96 )
= (7 4)(4 3 5 6 8 9) = (7 4)(3 5)(5 6 8 9 4) = (7 4)(3 5)(5 6 8 9)(9 4) = (7 4)(3 5)(9 5)(5 6 8)(9 4)
= (7 4)(3 5 9)(6 8)(8 5)(9 4) = PQ,
P = (7 4)(3 5 9)(6 8),
Q = (8 5)(9 4),
R = (4 3 5 6 8 9 7) = (4 3)(5 6 8 9 7 3) = (4 3)(5 6)(8 9 7 3 6) = (4 3)(5 6)(8 9)(6 9 7 3) = (4 3)(8 9)(5 6)(7 6 9)(7 3) = (4 3)(8 9)(5 6 7)(6 9)(7 3) = P'Q',
P' = (4 3)(5 6 7)(8 9) = RPRl,
Q' = (9 6)(7 3) = RQRl.
13. Try to expand all nonstandard Young tableaux y of the Young pattern [2,1] in terms of the standard Young tableaux Y,, Y = C , t,y,, where t, is the vector in the group space of S3.
Solution. There are six Young tableaux corresponding to the Young pattern [2,13, where two Young tableaux & and & are standard, and the remaining are nonstandard.
The symmetry and the Fock conditions of a Young operator lead to
Permutation Groups
205
6.3 Primitive Idempotents in the Group Algebra
A vector e in the group algebra L of a finite group G is called an idempotent if e2 = e E C . n vectors e, in C are called a set of the mutually orthogonal idempotents if e,ev = S,,e,. Le, = L, and e,L = R, are respectively called the left ideal and the right ideal of C generated by the , idempotent e,. Two left (or right) ideals generated by two orthogonal idempotents contain no common vector. For a complete set of the arbitrarily chosen bases IC, a left ideal in L, = Le,, the matrices D ( S ) of the group elements S in the bases, Sz, = C, I c T D T ~ ( Sform a representation of the group G, which is called the ), representation corresponding to the left ideal L,. Two left ideals, or two idempotents generating those two left ideals, are said to be equivalent if the representations corresponding to two left ideals are equivalent. A left ideal C is said to be minimum and its idempotent is said to be primitive , if the representation corresponding to the left ideal is irreducible. The necessary and sufficient condition for an idempotent e, to be primitive is that e,te, = Ate, holds for any vectors t in the group algebra L,where At is a constant depending on t , which is allowed to be zero. The necessary and sufficient condition for two primitive idempotents e, and e, to be equivalent is that there exists at least one group element R such that e,Re, # 0. These conclusions are also suitable for the right ideals, except for the calculation of the representation matrix of S in a right ideal R, by z,S = D,,(S)z,.
The Young operator is proportional to the primitive idempotent of the permutation group. Two Young operators corresponding to different Young patterns are orthogonal to each other. The representations generated by them are inequivalent and irreducible. Thus, an irreducible representation for a permutation group is described by a Young pattern. When n 5 4, those standard Young operators corresponding to the same Young pattern for the permutation group S, are orthogonal to one another. This conclusion is not generally true. When n 2 5 , for a given Young pattern [A], we define 4x1 (S,) mutually orthogonal primitive idempotents in the following way. For simplicity, the index [A] and S, are omitted. Denote by R,, the permutation transforming the standard Young tableau y , into the standard Young tableau y,. If y,yv # 0, then R,, = P;’)Q?) = PL”)Qf),where P;’) and Q?) are respectively the horizontal and the vertical permutations belonging to the Young tableau y,, while Pi”’ and are respectively the horizontal and the vertical
*
*
c,
Qt)
206
Problems and Solutions in Group Theory
permutations belonging to the Young tableau y p . Let
d primitive idempotents ep are mutually orthogonal. In another way, let
p=
1
d primitive idempotents e; are mutually orthogonal. The identity E of the permutation group may respectively expand in terms of two sets of the primitive idempotents
Note that n ! / d l x ] is nothing but the product of the hook numbers all boxes in the Young pattern [A].
14. Expand explicitly the identity of the operators.
Sq
hij
for
group in terms of the Young
A B C Solution. E =  +  +  where 24 12 8 ’
=2{E
+ (1 2)(3 4) + (1 3)(2 4) + (1 4)(2 3) + (1 2 3) + (1 3 2) + (1 2 4) + (1 4 2) + (1 3 4) + (I 4 3) + (2 3 4) + (2 4 3)) ,
Permutation Groups
207
=E
+ (1 2) + (3 4) + (1 2)(3 4)  (1 3)  (2 1 3)
1)  (2 1 4 3)  (2 4)  (1 2 4)  (3 4 2)  (1 2 3 4)
 (4 3
+ (1 3)(2 4) + (1 3 2 4) + (3 1 4 2) + (1 4)(3 2) + E + (1 3) + (2 4) + (1 3)(2 4)  (1 2)  (3 1 2)
 (4 2 1)  (3 1 4 2)  (3 4)  (1 3 4)  (2 4 3)  (1 3 2 4)
+ (1 2)(3 4) + (1 2 3 4) + (2 14 3) + (1 4)(2 3)
= 2 {E
+ (1 2)(3 4) + (1 3)(2 4) + (1 4)(3 2))  (1 2 3)  (1 3 2)
 (1 2 4)  (1 4 2)  (1 3 4)  (1 4 3)  (2 3 4)  (2 4 3),
y ~ + y F
4
= 2 {E + (1 2 3) + (1 3 2)  (2 3)(1 4))  (2 1 4)  (4 1 2)  (3 1 4)  (4 1 3)  (2 3 1 4) + (4 1 2 3)  (3 2 1 4) + (4 1 3 2) + 2 { E + (1 2 4) + (1 4 2)  (2 4)(1 3))  (2 1 3)  (3 1 2)  (4 1 3)  (3 1 4)  (2 4 1 3) + (3 1 2 4)  (4 2 1 3) + (3 1 4 2) + 2 {E + (1 4 3) + (1 3 4)  (4 3)(1 2))  (4 1 2)  (2 1 4)  (3 1 2)  (2 1 3)  (4 3 1 2) + (2 1 4 3)  (3 4 1 2) + (2 1 3 4) = 6E  2(2 3)(1 4)  2(2 4)(1 3)  2(4 3)(1 2).
15. Calculate the orthogonal primitive idempotents for the Young pattern
[2,2, I] of the permutation group
Solution. There are five standard Young tableaux tern [2,2,1] of S 5 :
After checking, the product of each pair of the standard Young operators is vanishing except for &y5 # 0. The permutation transforming the Young tableau ys into the Young tableau Yl is R15 = (4 2 3 5) = (2 5) (5 4)(2 3). h o r n Eq. (6.7) we obtain y1 = E  (2 5), and the remaining ycLare E.
+
S5.
y p for the Young patY 5
Yl
Y2
p FF q ll
Y3 Y 4
208
Problems and Solutions in Group Theory
Thus, the orthogonal primitive idempotents are
From Eq. (6.8) we may decompose R15 as R15 = (3 4) (4 2)(3 5), and obtain another set of the orthogonal primitive idempotents
16. Calculate the orthogonal primitive idempotents for the Young pattern [4,2] of the permutation group s6.
Solution. There are nine standard Young tableaux yp for the Young pattern [4,2] of s6, listed from the smallest to the largest:
1234 1235 1236 1245 1246 56 46 45 36 35 1256 1345 1346 1356 34 26 25 24
After checking, only the products of three pairs of the Young operators are nonvanishing:
From Eq. (6.7) we obtain the orthogonal primitive idempotents as
1 el = Y1 80 1 e3 = Y3 80
[ E  (2 5)(3 6 4)]
e2
=
 9 2
[E  (2 4)(3 5)] ,
ep =
1 80 1
[ E  (2 4)(3 6 511
4 9 5 9 .
From Eq. (6.8) we may decompose the above R,, as R18 = (4 3 2)(5 6) (6 2), R29 = (5 3 2)(4 6) (6 2), R39 = (3 2)(4 5) (5 2), and obtain another set of orthogonal primitive idempotents
Permutation Groups
209
17. Calculate the orthogonal primitive idempotents for the Young pattern [3,2,1] of the permutation group SS.
Solution. There are 16 standard Young tableaux Y , for the Young pattern [3,2,1] of s6, listed from the smallest to the largest: 123 123 124 124 125 125 126 126 45 46 35 36 34 36 34 35 6 5 6 5 6 4 5 4
134 134 135 135 136 136 145 146 25 26 24 26 24 25 26 25 6 5 6 4 5 4 3 3
After checking, there are 8 pairs of the Young operators whose products are nonvanishing:
R1,ll R1,12
= (2 4 5 3) = (2 4)(5 3) (3 4) = (4 5)(3 2) (2 5),
= (2 4 6 5 3) = (5 3)(6 2) (2 4)(6 3) = (3 2)(5 4) (4 6)(5 2),
= (2 4 6 3) = (2 4)(6 3) (3 4) = (4 6)(3 2) (2 6), = (2 4 5 6 3) = (6 3)(5 2) (2 4)(5 3) = (3 2)(6 4) (4 5)(6 2),
R2,13 R2,14
= (2 3 6 5 4) = (5 4)(6 2) (2 3)(6 4) = (4 2)(5 3) (3 6)(5 2),
R4,16
= (2 3 5 6 4) = (6 4)(5 2) (2 3)(5 4) = (4 2)(6 3) (3 5)(6 2),
R5,15 R7,16
= (2 3 6 4) = (6 2) (2 3)(6 4) = (4 3) (3 6)(4 2),
= (2 3 5 4) = (5 2) (2 3)(5 4) = (4 3) (3 5)(4 2).
Thus, we obtain the orthogonal primitive idempotents from Eq. (6.7):
el =  1 [ E  (2 4)(3 5)  (2 6)(3 5)] , 7
1 45
210
Problems and Solutions in Group Theory
Another set of the orthogonal primitive idempotents is obtain from Eq. (6.8)
18. Calculate the orthogonal primitive idempotents for the Young patterns [3,3] and [4,1,1] of the permutation group s6, respectively.
Solution. There are five standard Young tableaux tern [3,3] of s 6 :
Y , for the Young pat
Yl
Y2
Y3
Y4
y5
123 124 125 134 135 456 356 346 256 246
The digit 6 can only be filled in the last box of the second row, so the standard Young tableaux for the Young pattern [3,3] of s are the same 6 as those for the Young pattern [3,2] of S 5 , respectively, except for adding the last box filled with 6. The calculation for the orthogonal primitive idempotents is also the same. After checking, the product of each pair of the standard Young operators is vanishing except for y1y5 # 0. The permutation transforming the Young tableau y5 into the Young tableau Y1 is R15 = (2 4 5 3) = (2 4)(5 3) (3 4) = (4 5)(3 2) (2 5). From Eqs. (6.7) and (6.8) we obtain two sets of the orthogonal primitive idempotents as
There are ten standard Young tableaux yp for the Young pattern [4,1,1] of s 6 , listed from the smallest to the largest:
Permutation Groups
211
1234 1235 1236 1245 1246 4 4 3 3 5 6 6 5 6 5 1256 1345 1346 1356 1456 3 2 2 2 2 4 6 5 4 3
After checking, each pair of the Young operators are orthogonal. Y , / 7 2 are ten orthogonal primitive idempotents for the Young pattern [4,1,1] of S6
19. Give an example to demonstrate that the primitive idempotent generating a minimum left ideal is not unique.
Solution. Take the S3 group as an example. Let Y1 be the standard Young operator corresponding to the Young tableau & generates the minimal left ideal L1 = L y l . Let & be the nonstandard Young operator corresponding to Young tableau F l . It is easy to see that
w!
J$ = (1 3 ) & ( l 3) = (1 3)&. Being a group algebra, L = L(l 3). Therefore, the left ideals generated by & and Y2 are the same, L1 =
LCy, = LCy2. & / 3 and &/3 are both the primitive idempotents generating the same minimal left ideal L1.
6.4
Irreducible Representations and Characters
In this section we will discuss the irreducible representation of S, denoted by a given Young pattern [A]. Sometimes, we neglect the index [A] for simplicity. Let the permutation R,, transform the standard Young tableau y, into the standard Young tableau y,, and d be the number of the standard Young tableaux corresponding to the Young pattern [A]. Define the standard basis as
*
For a given v , the d bases b,, span the left ideal L,, while for a given p , the d bases b,, span the right ideal R,. The representation matrix of the group element S in the standard bases take the same form both for the left
212
Problems and Solutions in Group Theory
ideal and for the right ideal:
Of course, another set of the standard bases can also be used:
For definiteness, we will only discuss the former set of bases. The representation matrix of the group element S in the standard bases can be calculated by the tabular method. Denote by Y,(S) the Young tableau transformed from the standard Young tableau y, by the permutation S. Let y p = C S k T k , and denote by Ypk the Young tableau transformed from the standard Young tableau yp by the permutation T;'. Define the coefficient ALk(S) by comparing two Young tableaux Y,, and Yv(S). If there is a pair of digits filled in the same row of the Young tableau Y,(S) and in the same column of the Young tableau Y,k, then ALk(S)= 0. Otherwise, ALk(S)is defined to be the permutation parity of the vertical permutation R of the Young tableau Y p k which transforms the Young tableau yPk into the Young tableau y' such that the digits in each row of the Young tableau y' are the same as those in the same row of the Young tableau Y,(S). The representation matrix entry D p v ( S )of the permutation S in the standard bases b,, is equal to C k6kALk(S), which can be calculated by the tabular method. In fact, one may draw a table, where each row is designated by the algebraic sum of the Young tableaux, Ed, {the Young tableau Y p k } , and each column is designated by the Young tableau y,(S). A$) can be calculated by comparing the Young tableau Ypk in the row and the Young tableau yv(S) the column. The irreducible representation D ( S ) in the in standard bases is generally not unitary. In the orthogonal basis ap,the representation matrix D(P,) of a transposition P, for two neighbored objects, P, = ( a a l),is the direct sum of several one or twodimensional submatrices, which can be calculated as follows. If a and a + 1 are filled in the same row or the same column of the standard Young tableau Yp, we have a onedimensional submatrix D,,(P,) = 1 or 1, respectively. Otherwise, we have a twodimensional submatrix. Without loss of generality, we assume that a is filled in the upper row of the standard Young tableau yp than ( a l), and the standard Young tableau ypais obtained from the standard Young tableau y,
*
*
+
+
Permutation Groups
213
by interchanging a and a 1. Since a and a 1 are filled in the different rows and in the different columns of the standard Young tableau y,, the Young tableau ypamust be standard and larger than the standard Young tableau y,. Now, the twodimensional submatrix is
+
+
(6.12)
where the digit a is filled in the c,(a)th column of the r,(a)th row of the standard Young tableau y,, and similar for the digit a 1. m,(a) is called the content of a in the Young tableau y,. If one goes from the box filled with a in the Young tableau y, downwards or leftwards and at last reaches the box filled with a + 1, the number of steps is equal to m.
+
The merit of the representation D ( S ) of Sn in the standard bases is that the bases are explicitly obtained, but the representation is not unitary. The merit of the representation B ( P a )of Sn in the orthogonal bases is that the representation is real orthogonal, but its bases have to be calculated from the standard bases. The similarity transformation matrix X from the standard bases to the orthogonal bases is an upper triangle matrix: D(Pa)X= XD(Pa), X,, = 0, when p
*
> v.
(6.13)
If only one left ideal L, is considered, the orthogonal basis
4pvis
(6.14)
P
P
If both left and rightideals are considered, we have to calculate the or” thogonal bases @,
(6.15)
P P
There is a graphic method for calculating the character of a class ( l ) in the irreducible representation [A]. Rearrange the partition number l j in an arbitrary order, but it is convenient to arrange l j from the smallest to
*
214
Problems and Solutions an Group Theory
the largest. According to the following rule, we fill n digits into the Young pattern [A] in order: first el digits 1, then & digits 2 and so on:
a) After each set of e j digits j is filled, the boxes which have been filled must constitute a subYoung pattern, namely the boxes are lined up on the top and on the left, where the number of boxes on the upper row is not less than that on the lower row, and the number of boxes at the left column is not less than that at the right column. In each row, there is no unfilled box embedding between two filled boxes. b) The boxes filled with the same digit are connected such that from the lowest and the leftist box one can go rightwards or upwards, without going leftwards or downwards, through all boxes filled with the same digit.
If all digits are filled into the Young pattern according to the rule, we say there is one regular application. The filling parity for the digit j is defined to be 1 if the number of rows of the boxes filled with j is odd, and to be 1 if that is even. The filling parity of a regular application is defined to be the product of the filling parities for all digits. The character $'I {(C)} of the class (1) in the representation [A] is equal to the sum of the filling parities of all regular applications. If the digits cannot be all filled in the Young pattern satisfying the above rule, we say there is no regular application and obtain x['] { ( 1 ) } = 0. For the class (1") composed of only the identity, each standard Young tableau corresponds to one regular application, so its character is just the dimension of the representation. Usually, the character of the identity is calculated by the hook rule instead of the graphic method. Two Young patterns related by a transpose are called the associated Young patterns. Each regular application of the class (C) in one Young pattern is a transpose of a regular application in its associated Young pattern. Since the sum of the number of rows and the number of columns of the boxes filled with j in a regular application is equal to Cj + 1. Note that (l)'j+' is just the permutation parity of the cycle with length e j . Therefore, the product of two filling parities of the digit j in the corresponding regular applications for two associated Young patterns is equal to the permutation parity of the cycle, and the characters of the class (C) in two representations denoted by two associated Young patterns are the same if the permutation parity of the elements in the class is even, and different by a sign if that is odd. The Young pattern of the identical representation is [n]and its
*
Permutation Groups
215
associated Young pattern [l”] denotes the antisymmetric representation, where the representation matrix of S is equal to the permutation parity 6(S). A representation denoted by [A] is equivalent to the direct product of the antisymmetric representation [I”] and the representation denoted by the associated Young pattern [XI.
20. Calculate the ClebschGordan coefficients for the reduction of the selfproduct of the irreducible representation [2,1] of the S3 group in the standard bases and in the orthogonal bases, respectively.
Solution. There are three inequivalent irreducible representations for S3. [3] is the symmetric (identical) representation. [13]is the antisymmetric representation. Take the transpositions PI and P of the neighboring ob2 jects as the generators of S3. Their representation matrices are known:
[2,1] is the mixed representation in which the representation matrices of the generators can be calculated by the tabular method. There are two orthogonal standard Young operators for the Young tableau [2,1]. The listed rows in the table are designated by the standard Young tableaux YcL, in the first column of the table. Apply Pa, a = 1 or 2, to the standard The Young tableau y,, one obtains the Young tableau yv(Pu). columns in the table are designated by the Young tableau Y,(P,), listed in the first row of the table. The coefficients AL(Pu) is determined by comparing the Young tableaux indicating the pth row and the vth column. AL(Pa) is nothing but the representation matrix entry of Pa, a = 1 or 2.
The representation matrices for the generators in the mixed representation [Z, 1 are 1
216
Problems and Solutions in Group Theory
The reduction for the selfproduct of the representation [2,1] is
Denote by X the similarity transformation matrix for reduction
( o  1 1 1 ) 1 l 0 l 0 0 11 0 0 0 1 0 0 0 1
x ~ x (1o00l 0 ) o
0 0 11 0 0 01 1 0 0 0
'
(:;
;)x=x(:d: 0 0
1 0
;).
Due to P = E , the eigenvalues for two matrices are both f l . The eigenz vectors for the eigenvalue 1 of two representation matrices before transformation respectively are
The transpose of the common eigenvector is X r = ( 2 1 1 2 ) . The eigenvectors for the eigenvalue 1 respectively are
The transpose of the common eigenvector is XT = (0 1  1 0). XI and X, are respectively the first and the second columns of X. Denote by X3 and X the third and the fourth columns of X. F'rom the first formula of the 4 similarity transformation, X3 is an eigenvector with the eigenvalue 1 of the representation matrix before transformation, XT = ( a b b 2b). Substituting X3 into the second formula, we calculate the third column of the matrix, and obtain XT = (2b b b u ) . Substituting them into the first formula, we
Permutation Groups
217
calculate the fourth column of matrix, and obtain
a
 2b
a  2b
Thus, a = b. Taking a = b = 1, we have 2 0  1 2
x=(11 1 1 ) 11
2 0 21

The reader is suggested to be familiar with this method. If not, he can calculate the matrix X by a direct way
Substituting it into the formulas of the similarity transformation, he will obtain the same result. It is worthy to mention that the columnmatrices in X are not orthogonal to one another because the representation before transformation is not unitary. Denoting by Ip)Iv) the bases before transformation, and by I I [3]), I I [ 13]) and I [2,13, p ) the bases after transformation, where p , u and p are respectively taken 1 and 2, we have
I
For the representations [3] and [13], the representation matrices in the orthogonal bases are the same as those in the standard bases. For the representation [2,1], the representation matrices of the generators in the orthogonal bases are
218
Problems and Solutions in Group Theory
Denote by 2 the real orthogonal similarity transformation matrix in the reduction of the selfproduct of the representation [2,1]:
1" (
4
(1 1 :).=.(.
1 0 0 0
1 0 0 0
0
0
10
14 4 3 3
1 f i
d) 3
1
a 3
f i
fi z =  z
; (;
0
0 l O1 00 ) ) 0
0 0 01 x 2
: :)
1 6 & 1

0
Due to P: = E , the eigenvalues of two matrices are both f l . The eigenvectors for the eigenvalue 1 of two representation matrices before transformation respectively are
T The normalized common eigenvector is Z = (1 0 0 1)/& The eigenvectors for the eigenvalue 1 respectively are
0 0
(i).(si;.:($ (
The normalized common eigenvector is 2 = (0 1  1 O)/fi. ; 2 and 2 2 1 are the first and the second columns of 2, respectively. From the condition of real orthogonal matrix, one may write the matrix 2 as
.=.i"0" 'i !
fi
0  1 0 1 ' 1 0 a0
Substituting it into the formulas for the similarity transformation related to Pz, we obtain a = 1 from the first row of the fourth column. Denoting again by 1p)Iv) the bases before transformation and by II[3]), 11[13]) and
Permutation Groups
219
11[2,13,p ) the bases after transformation, where p, v and p are respectively taken 1 and 2, we have
21. Calculate the standard bases for the irreducible representation [3,1] in the group space of Sq, and calculate the representation matrix of the transposition Pa for the neighbored objects in the standard bases by the tabular method. Solution. There are three standard Young tableaux Y1, &, and Y3 for the Young pattern [3,1]of the S3 group
The corresponding Young operators yF are orthogonal to one another. The permutations between three standard Young tableaux respectively are
Removing the factor 1/8, we obtain the standard bases
where each column belongs to one left ideal, each row belongs to one right
ideal, and
Y1 = E + (1 2) + (1 3) + (2 3) + (I 2 3) + (3 2 1)  (1 4)
 (2 1 4)
Y2
 (3 1 4)  (2 3)(1 4)  (2 3 1 4)  (3 2 1 4))
Y3
= (3 4)Y1(3 4))
= (2 4)%(2 4).
The representation matrices for P a are calculated by the tabular method:
220
Problems and Solutions in Group Theory
RW. [ 3 ~ 1 123 4 124 3 134 2
Pl 213 214 234
1
0
0
p2 132 134 124 4 2 3
1
0
p3
124 123 143 3 4 2 0
1
1
1
0
0
0
1
1
0
0
0
1
0
0
1
0
1
0
0
0
1
Thus, the representation matrices of the transpositions Pa for the neighbored objects in the representation [3,1] are:
22. Calculate the real orthogonal representation matrix of the transposi
tions Pa for the neighbored objects in the irreducible representation [3,1] of the S4 group in the orthogonal bases by the formula (6.12). Calculate the similarity transformation matrix between two representations in the standard bases (Problem 21) and in the orthogonal bases and write explicitly the orthogonal bases < ,p , in the group space of Sq.
Solution. The representation matrices of Pa in the orthogonal bases of the irreducible representation [3,1] of S4 can be calculated by Eq. (6.12))
D(P1) =
(:;:)
001
)
a(&)=
(::iz
&2)
)
0 J 3 / 2 1/2
Denote by X the similarity transformation matrix transforming the representation D ( P a ) in the standard bases into the representation a ( P a )in the orthogonal bases. X is an upper triangular matrix with three column matrices X I , X and X3: 2
Permutation Groups
221
where the coefficient J8 is added for simplicity. In calculation we neglect the rows where all entries are vanishing. From
we obtain a1 = 1 and
a2
= 3. From
D(P2)X2=X2+X3, 2 1 2
(;)
0
=
(1) q
3
+
we obtain bl = b2 = & and b3 = 2&. Finally, the similarity transformation matrix X and its inverse matrix are
Jg1
fi
According to Eqs. (6.14) and (6.15), we first calculate the bases +clv in a left ideal by X , then calculate the nine orthogonal bases @,,([3,1]) by Xl. In the results the normalization factors and 2& are used.
di@
411
= J ' l W 1 b = = =
d m { E+ (1 2) + (1 3) + (2 3) + (1 2 3) + (3 2 1) (1 4)
 (2 1 4 )  (3 1 4 )  (2 3)(1 4)  (2 3 1 4 )  (3 2 1 4 ) ) ,
421
d m { E+ 3(3 4)) hi d m { E+ (1 2)  2(1 3) + (2 3) + (1 2 3)  2(3 2 1)  (1 4)  (2 1 4) + 2(3 1 4 )  (2 3)(1 4)  (2 3 1 4) + 2(3 2 1 4) + 3(3 4)+ 3(3 4)(1 2) + 3(4 3 2) + 3(4 3 1 2)
t3 h i ) =
~/m{hi
 3(3 4 1)  3(3 4 2 1)  3(3 2 4 1)  3(3 1)(4 2 ) ) ,
431
+ b2i 4 2b31) = d m { E4 (3 4) 2(4 2)) hi = d W { E (1 2) + (2 3)  (1 2 3)  (1 4) + (2 1 4 )  (2 3)(1 4) + (2 3 1 4) + (3 4)  (3 4)(1 2) + (4 3 2)  (4 3 1 2)  (3 4 1) + (3 4 2 1)  (3 2 4 1) + (3 1)(4 2) + 2(4 2) + 2(4 2 3)  2(2 4 1)  2(2 3 4 l)},
= dm{bii
222
Problems and Solutions in Group Theory
412
= d b = d m ( 3 4)b22 mi 2 = d n ( ( 3 4)
+ (3 4 2 1)  (4 3 1)  (4 3 2 1)  (4 1)  (4 2 3 1)
 (4 1x3 2) 
+ (3 4)(1 2) + (3 4 1) + (3 4 2) + (3 4 1 2)
(4 2 1)) 7 4)
422
+ 3E} b22 = J m ( ( 3 4) + (3 4)(1 2)  2(3 4 1) + (3 4 2) + (3 4 1 2)  2(3 4 2 1)  (4 3 1)  (4 3 2 1) + 2(4 1)  (4 2 3 1)  (4 1)(3 2) + 2(4 2 1) + 3E + 3(1 2) + 3(2 4) + 3(1 2 4)
= dW(bi2 3322) = J 9 6 ( ( 3
+
 3(1 3)  3(2 1 3)  3(2 4)(1 3)  3(2 4 1 3)},
432
d m ( b i 2 + b22 + 2b32) = Jm{ (3 4) + E + 2(2 3)) b22 = d D ( ( 3 4)  (3 4)(1 2) + (3 4 2)  (3 4 1 2 )  (4 3 1) + (4 3 2 1)  (4 2 3 1) + (4 1)(3 2) + E  (1 2) + (2 4)  (1 2 4)  (1 3) + (2 1 3)  (2 4)(1 3) + (2 4 1 3) + 2(2 3) + 2(3 2 4)  2(2 3 1)  2(2 4 3 1)) ,
= = dnbl3 = d w ( 2 4)b33 = d m ( ( 2 4)
413
+ (2 4 3 1)  (4 2 1)  (4 2 3 1)  (4 1)  (4 3 2 1)
 (4 1 "
$23
+ (2 4)(1 3) + (2 4 1) + (2 4 3) + (2 4 1 3 )
3)  (4 3 1)) 7
= J1/96{b13 3b23) = J m ( ( 2 4) 3(2 3)) b33 = d W { ( 2 4)  2(2 4)(1 3) + (2 4 1) + (2 4 3)  2(2 4 1 3 )
+
+
+ 3(2 3 4 1) 3(3 2 1)  3(3 1)  3(3 4 2 1)  3(3 4 I ) ) , 433 = d { 1 + b23 + 2333) = d m { ( 2 4) + (2 3) + 2 E ) b33 mb 3 = d i 7 5 { ( 2 4)  (2 4 1) + (2 4 3)  (2 4 3 1)  (4 2 1) + (4 1)  (4 3 2 1) + (4 3 1)+ (2 3)  (2 3 1) + (2 3 4)
 (2
+ (2 4 3 1)  (4 2 1) + 2(4 2 3 1)  (4 1)  (4 3 2 1) + 2(4 1)(2 3)  (4 3 1) + 3(2 3) + 3(2 3 1) + 3(2 3 4)
3 4 1)  (3 2 1) + (3 1)  (3 4 2 1) + (3 4 1)
+ 2 E + 2(3 4)  2(1 2)  2(3 4)(1 2)},
Permutation Groups
223
a11([3,1]) = &@ { 3 h  $12
 $13) = {3bll  (3 4)b22  (2 4)b33) = Ji752(3E + 3(1 2) 3(1 3) + 3(2 3) 3(1 2 3) 3(3 2 1)  (1 4)  (2 1 4 )  (3 1 4 )  (2 3)(1 4)  (2 3 1 4 )  (3 2 1 4 )  (3 4)  (3 4)(1 2)  (3 4 1)  (3 4 2)  (3 4 1 2)  (3 4 2 1)  (2 4)  (2 4)(1 3)  (2 4 1)  (2 4 3)  (2 4 1 3)  (2 4 3 1)) ,
Jm
+
+
+
a 1([3,1]) 2
(3421  $22  $23) = = (1/24) (3 [E 3(3 4)] b i i  [(3 4) 3E] b22  [(2 4) 3(2 3)] b33) = (1/6) {(1 4)  (2 1 4)  (2 3)(1 4)  (2 3 1 4)  (3 4 1)  (3 4 2 1)  (3 2 4 1)  (3 1)(4 2)  (2 4)  (3 4 1 2)  (1 2 4)  (2 3 4) + 2(3 1 4) + 2(3 2 1 4) + 2(3 4)
+
+
+
+ 2(3 4)(1 2) + 2(4 3 2) + 2(4 3 1 2)} ,
%i([3,1]) = &@{3$3i  $32  $33) = [E (3 4) 2(4 2)] b i i
dW{3 +
4)
+
 [(3 4) + E + 2(2
3)] b22
 “2
=
dm{(l + (2 1 4 )  (2 3)(1 4) + (2 3 1 4 )  (3 4 1) 4) + (3 4 2 1)  (3 2 4 1) + (3 1)(4 2) + (4 2) + (4 2 3)

+ (2 3) + 2EI b33)
(2 4 1)  (2 3 4 1)),
(2412  613) = (1/6) (2(3 4)b22  (2 4)b33} @12([3,1]) = = (1/6) {2(3 4) 2(3 4)(1 2) 2(3 4 1) 2(3 4 2)
+ + + + 2(3 4 1 2) + 2(3 4 2 1)  (4 3 1)  (4 3 2 1)  (4 1)
 (4
2 3 1)  (4 1)(3 2)  (4 2 1)  (2 4)  (2 4)(1 3) (2422  $23}
 (2 4 1)  (2 4 3)  (2 4 1 3)  (2 4 3 I ) ) ,
a 2 ([3, 2
I]) =
= J m ( 2 [(3 4) 3E] b22  [(2 4) 3(2 3)] b33) = J m { 2 ( 3 4) 2(3 4)(1 2)  (3 4 1)  (3 4 2)  (3 4 1 2)  (3 4 2 1)  (4 3 1)  (4 3 2 1) + 5(4 1)  4(4 2 3 1)  4(4 1)(3 2) + 5(4 2 1) + 6 E + 6(1 2) + 5(2 4) + 5(1 2 4)  3(1 3)  3(2 1 3)  4(2 4)(1 3)
 4(2
+ +
+
4 1 3)  (2 4 3)  (2 4 3 1)  3(2 3)  3(2 3 1)),
224
Problems and Solutions in Group Theory
@32([3,1]) = (2432  433) = d m { 2 [(3 4) E 2(2 3)] b22
+ +
 [(2 4) + (2 3) 4 2E] b33) = J i p { ( 3 4 2)  ( 3 4 1 2)
+ (2 4)  (1 2 4) + (4 2 1)  (4 1) + (3 4 2 1)  (3 4 1)  2(4 2 3 1) + 2(4 1)(3 2)  2(2 4)(1 3) + 2(2 4 1 3)  3(4 3 1)  3(1 3) + 3(2 1 3) + 3(4 3 2 1)  3(2 3 1)  3(2 4 3 1) + 3(2 3) + 3(3 2 4)},
@i3([3,1]) = $13 = d m ( 2 4)b33 = d@{(2 4) (2 4)(1 3) (2 4 1) (2 4 3) (2 4 1 3 ) (2 4 3 1)  (4 2 1)  (4 2 3 1)  (4 1)  (4 3 2 1)
+
+
+
+
+

(4 1)(2 3)  (4 3 1)) I
@23([3,1])= $23 = d W ( ( 2 4) + 3(2 3)) b33 = d P ( ( 2 4) (2 4 1) (2 4 3) (2 4 3 1)  (4 2 1)
+
+
+
 (4
1)  (4 3 2 1)  2(2 4)(1 3)  2(2 4 1 3) + 2(4 2 3 1)
+ 2(4 1)(2 3)  (4 3 1) + 3(2 3) + 3(2 3 1) + 3(2 3 4) + 3(2 3 4 1)  3(3 2 1)  3(3 1)  3(3 4 2 1)  3(3 4 I)}, @33([3,13)= $33 = d m { (2 4) + (2 3) + 2 E ) b33
= Ji7E{(2

4)  (2 4 1) + (2 4 3)  (2 4 3 1)  (4 2 1)
+ (4 1)  (4 3 2 1) + (4 3 1) + (2 3)  (2 3 1) + (2 3 4)
(2 3 4 1)  (3 2 1)
+ (3 1)  (3 4 2 1)+ (3 4 1)
+ 2 E + 2(3 4)  2(1 2)  2(3 4)(1 2)).
23. Calculate the representation matrices of the transpositions Pa for the neighbored objects both in the standard bases and in the orthogonal bases of the irreducible representation [2,1, I] of the S4 group. Calculate explicitly the orthogonal bases aPv the irreducible representaof tion [2,1,1] in the group space of S4.
Solution. There are three standard Young tableaux sponding to the Young pattern [2,1, I] in S3:
yl, y and ys 2
corre
Yl
:
Y2 :
Y3 :
Permutation Groups
225
The corresponding Young operators yClare orthogonal to one another. The permutations between three standard Young tableaux respectively are R 1 2 = (2 3), R 1 3 = (2 3 4) and R 2 3 = (3 4), so the standard bases are
The representation matrices of the transpositions Pa of the neighbored objects in the representation [2,1,1] can be calculated by the tabular method:
2 1 3 4
Rep. [2,1,1]
1 2 3 4 13 2 4 1 4 2 3
Pl 2 3 1 4
p 2
2 4 1 3 1 3 2 4 0 1 1 2 3 4 1 0 1 4 3 2 0 0 1 2 4 3
p3
1 4 2 3 0 0
1 3 2 4
1  1 1
0 1 0
1
0
0
1
0
0  1
0
0

1
0
1
0
Namely,
11 1
D(P1) = ( 0  1 0 0 0 1
),
0 1 0
D(P2)
= (1 0 0
0 01
),
D(P3)
=
(
10 0 0 0 1). 0 1 0
The representation matrices of Pa in the orthogonal bases of the irreducible representation [2,1,1] is calculated by the formula (6.12):
P 3 ) = D(
(d
0
0 1/3
0 a/3) .
1/3
4/3
From D(P,)X = XD(P,), the similarity transformation matrix X and its inverse matrix can be calculated to be
226
Problems and Solutions in Group Theory
The orthogonal bases with the normalization factor by Eqs. (6.1314):
are calculated
drn{(l + + + + + + (3 1 2 4)  (4 2) + (4 2 3)  (2 4 1) + (4 1 2 3) + 2 E  2(3 4) + 2(1 2)  2(1 2)(3 4)}, a 1([2,1,1]) = d m 2 + 4b21  b12  2b22 + + 2b23) = d m ( 2 [E+ 2(2 3)] b l l  [(2 3) + 2E] b22
{&l b13
(2bll  b12 b13) ([2, l]) = = d W ( 2 b i i  (2 3)b22  (4 2)b33) = 3)  (1 4) (1 3 4) (4 3 1)  (2 1 3 )  (2 1 4 ) (2 1 3 4) (2 1 4 3)  (2 3) (3 2 4)  (3 1)
dm
+
d57G{(3 4)  (3 2 1 4 ) + (3 1 2 4) + (3 1 4 ) + (2 4) 2  (1 2 4)  (1 4) + (4 2 1)  2(2 3)(1 4) + 2(2 3 1 4)  2(3 2 4 1) + 2(3 1)(2 4)  3(1 3) + 3(1 3 4) + 3(2 1 3)  3(2 1 3 4) + 3(2 3)  3(2 3 1) 3(2 3 4) + 3(2 3 4 1)) , a 1([2,1,1]) = d 3 m {  2 h i + 2b21 + 6b31 + b12  b22  3b32  b13 + b23 + 3b33) = J1/192{2 [E + (2 3)  3(2 4)] b l l  [(2 3) + E + 3(3 4)] b22
=
+ [(4
2) + 2(3 411 b33)
+ [(4 2) + (3 4) + 3E]b33)
=
+ (2 3 14)  (2 4) + (2 4)(1 3) + (2 4 1)
+ (2 4 3)  (2 4 13)  (2 4 3 l)},
dm{(l  (4 3 1)  (2 1 4 ) + (2 1 4 3)  (2 3)(1 4) 4)
@12([2,1,1])= d n ( 3 b i 2  b i 3 ) = d m { 3 ( 2 3)b22 + (4 2)b33} = J57G{(4 2)  (4 2 1)  (4 2 3) + (4 2 1 3 ) + (2 4 1)
 (4 1)  (4 1 2 3) + (4 1 3 )  2(2 3)(1 4) + 2(3 2 4 1)  2(2 3 1 4) + 2(3 1)(2 4) 3(2 3 1)  3(3 1)  3(3 1 2 4) + 3(3 1 4) + 3(2 3)  3(3 2 1)  3(3 2 4) + 3(3 2 1 4)},
+
Permutation Groups
227
+ + (4 1 2 3)  (4 1 3) + (3 1 2 4)  (3 1 4)  2(3 4) + 2(3 4)(1 2) + 3(2 3)  3(3 2 1)  3(2 3 1) + 3(3 1)  4(2 3)(1 4) + 4(3 2 4 1) + 4(2 3 1 4)  4(3 1)(2 4)  5(1 4)  5(2 4) + 5(1 2 4) + 5(4 2 1)+ 6 E  6(1 2)), $)32([2,1,1])= (1/24) {3bi2 + 3b22 + 9332 + b13  b23  3333)
= (1/24) (3 [(2 3)
= (1/6) ((2 3)(1 4)
@22([2,1,1]) = dm{3b12 6322  b13  2b23) =J m 3 3) 2E] b22  [(4 2) 2(3 4)] b33) ( [(2 = d i p z {  ( 3 2 4) (3 2 1 4 )  (4 2 3) (4 2 1 3 )
+
+
+
+
+ (3 2 4)  (3 2 4 1)  (3 2 1 4)  (2 3 1 4)  (3 1 2 4) + (3 1)(2 4) + (3 1 4)  (1 4)  (2 4) + (1 2 4) + (4 2 1)+ 2(3 4)  2(3 4)(1 2)  2(3 4 1)  2(3 4 2) + 2(3 4 1 2) + 2(3 4 2 l)} , @13([2,1713) = d h =  d m ( 4 2)b33 m3 = d i p 3 {  ( 4 2) + (4 2 1) + (4 2)(1 3) + (4 2 3)  (4 2 3 1)  (4 2 1 3)  (2 4 1) + (4 1) + (2 4 1 3) + (4 1 2 3)
 (4 1)(2 3)  (4 1 3))
9
+ E + 3(3 4)]b22  [(4 2) + (3 4) + 3E]b33)
+ + (4 1)(2 3) + (4 1 3) + 2(3 4)  2(3 4)(1 2)  2(4 3 1)  2(4 3 2) + 2(4 3 1 2) + 2(4 3 2 l)}, @33([2,1,13) = d m {  h 3 + b23 + 3b33)
@23([2,1,1])= (1/6) (b13 + 2b23} = (1/6) {(4 2) + 2(3 4)) b33 = (1/6) {(4 2) (4 2 1) + (4 2)(1 3) + (4 2 3)  (4 2 3 1)  (4 2 1 3) (2 4 1)  (4 1)  (2 4 1 3)  (4 1 2 3)
+
= d W ( ( 4 2) (3 4) 3E) b33 = d r n { ( 4 2)  (4 2 1)  (4 2)(1 3)  (4 2 3) + (4 2 3 1) + (4 2 1 3 )  (2 4 1) + (4 1) + (2 4 1 3 ) + (4 1 2 3)  (4 1)(2 3)  (4 1 3) (3 4)  (3 4)(1 2)  (4 3 1) (432)+(4312)+(4321) 3E  3(1 2)  3(1 3)  3(2 3) 3(1 2 3) 3(3 2 1 ) ) .
+
+
+
+
+
+
228
Problems and Solutions in Group Theory
24. Calculate the standard bases and the orthogonal bases in the group
space of S4 for the irreducible representation [2,2], and calculate the representation matrices of the transpositions Pa for the neighbored objects in these two sets of bases. Sohtion. There are two standard Young tableaux Y and Y2 correspond1 ing to the Young pattern [2,2] of S4
Two Young operators Y1 and Y2 are orthogonal to each other. The permutation relating two standard Young tableaux are R12 = R21 = (2 3). Therefore, removing the normalization factor 1/12, we obtain the standard bases
bll
= Y1 = E
+ (1 2) + (3 4) + (1 2)(3 4)  (1 3)  (2 1 3)
 (4 3 1)  (2 1 4 3)  (2 4)  (1 2 4)  (3 4 2)  (1 2 3 4) (1 3)(2 4) (1 3 2 4) (3 1 4 2) (3 2)(1 4))
+
b2i
+
+
+
= (2 3)yi
b12
= (2 3)&,
b22
= y = (2 3)&(2 3). 2
The representation matrices of the transposition Pa of the neighbored objects in the standard bases for the representation [2,2] are calculated by the tabular met hod:
21 34 1 0 23 14 1 1 13 24 0 1 12 34 1 0 1 0 12 43 14 23 1 1
Rep1 3 1 2
[2,21
2 4 3 4
On the other hand, the representation matrices of Pa in the orthogonal bases of the irreducible representation [2,2] are calculated by Eq. (6.12)
We calculate the similarity transformation matrix X and its inverse matrix from D(P,)X = x D ( P ~ ) :
Permutation Groups
229
Thus, from Eqs. (6.1314), we obtain the orthogonal bases with the normalization factor
dw:
+
@ii([2,2])= d m ( 2 b i i 312) = d ~ ( 2 h i (2 3)b22) =J m ( 2 E 2(1 2) 2(3 4) 2(1 2)(3 4) 2(1 3)(2 4)
+
+
+
+ 2(1 3 2 4) + 2(3 1 4 2) + 2(3 2)(1 4)  (1 3)  (2 1 3)
 (4 3 1)  (2 1 4 3)  (2 4)  (1 2 4)  (3 4 2)  (1 2 3 4)  (2 3)  (2 3 1)  (3 2 4)  (3 1 2 4)  (2 1 3 4)  (3 4 1)  (2 1 4 )  (1 4 ) ) ,
@21 ([2,2]) =
d m (2311 44b2i  312  2322) =d m (2 [E+ 2(2 3)]311  [(2 3) + 2E] b22) = (1/4) ((1 3) + (2 1 3) + (4 3 1)  (2 1 4 3)  (2 4) + (1 2 4) + (3 4 2)  (1 2 3 4) + (2 3) + (2 1 3 4)  (2 3 1)
 (2 1 4)  (3 2 4)  (3 4 1) + (3 1 2 4) + (1 4)},
@12([2,2]) = ( 1 / 4 ) b = (1/4)(2 3 ) h = (1/4) ((2 3) (2 3 1) (3 2 4) + (3 1 2 4)  (3 2 1)  (3 1)  (3 2 1 4)  (3 1 4 )  (2 3 4)  (2 3 4 1)  (2 4)
+
+

(2 4 1)
+ (2 1 3 4) + (3 4 1) + (2 1 4) + (1 4)) ,
+ + + + +
@22([2,2])= d77@(312 2322) = J r n ( ( 2 3) + 2 E ) b22 = J i p i ( ( 2 3)  (2 3 1)  (3 2 4) + (3 1 2 4)  (3 2 1) (3 1) + (3 2 1 4)  (3 1 4)  (2 3 4) + (2 3 4 1) + (2 4)  (2 4 1) + (2 1 3 4)  (3 4 1)  (2 1 4) (1 4) 2 E 2(1 3)(2 4)  2(1 2)
 2(3 1 4 2)
 2(3
4)  2(1 3 2 4) + 2(1 2)(3 4)
+ 2(2 3)(1 4 ) ) .
25. Calculate the symmetric bases of the oscillatory wave function of a
molecule with the symmetry of a regular tetrahedron, for example, the methane CH4. Solution. The molecule CH4 has the symmetry of a regular tetrahedron. The carbon atom is located at the center of the regular tetrahedron denoted by 0, while four hydrogen atoms lie in four vertices of the tetrahedron denoted by Aj, 1 5 j 5 4, respectively. The stretching oscillators of the CH bonds OA, are described by the state lala2a3a4) with four oscillatory
230
Problems and Solutions in Group Theory
quantum numbers. Due to the symmetry, the frequency CJ of four oscillators are the same, and the energy for the state is E = Cj=l ( a j 2)tiw. The symmetry group for a regular tetrahedron is the T d group, which is isomorphic onto the permutation group S4. The correspondence between the elements of two groups can be established as follows. If an element R of T d transform four vertices A j of the tetrahedron respectively into the positions of Arj, R corresponds to the following permutation:
+
R(l
f l f 2 73 74 4
).
Especially, from Fig. 6.1 we see that three improper twofold rotations (el  e3) and (el  e 2 ) /fi coraround the directions (el e2) respond to the transpositions PI, P2, and P3 of the neighbored objects, respectively. Since two groups are isomorphic, we will not distinguish the elements in the T group and in the permutation group Sq. d
+
/a,
/a,
A3
d Fig. 6.1 Diagram of a molecule with the T symmetry.
We have obtained the orthogonal bases Q P v ( [ X ] ) for the inequivalent irreducible representations of S4 in Problems 2124, in addition to the bases for two onedimensional representations
@([4])=
d m { E+ (1 2) + (1 3) + (1 4) + (2 3) + (2 4) + (3 4)
t (1 2)(3 4) t (1 3)(2 4)
+ (1 4 2) + (1 3 4) + (1 4 3) + (2 3 4) + (2 4 3) + (1 2 3 4)
+ (1 4)(2 3) + (1 2 3) + (13 2) + (12 4)
Permutation Groups
231
+ (1 2 4 3) + (1 3 4 2) + (1 3 2 4) + (1 4 2 3) + (1 4 3 2)}, <p([i4]) = ,/W{E 2)  (1 3)  (1 4)  (2 3)  (2 4)  (3 4)  (1 + (1 2)(3 4) + (1 3)(2 4) + (1 4)(2 3) + (1 2 3) + (1 3 2) + (1 2 4) + (1 4 2) + (1 3 4) + (1 4 3) + (2 3 4) + (2 4 3)  (1 2 3 4)
 (1 2 4 3)  (1 3 4 2)  (1 3 2 4)  (1 4 2 3)  (1 4 3 2)}.
Applying those operators to the oscillatory state la1u ~ u ~ Q we obtain the ), symmetric bases of the oscillatory wave function of the molecule with T d symmetry :
R@,v([AI)la1.2a3a4)
=
c
P
~X,'(R)a,,([x])lalaza3a4).
is The application of an element R of S4 to the oscillation state ~ a ~ a 2 a 3 a 4 ) defined as follows. For the permutation R,
R=
we have
( 1 2 3 4 ) = ("
T1 72 73 T4
1
Recall that the application of a permutation R to the state only changes the order of four oscillatory quantum numbers a j , but does not change their values, In fact, there exist 24 linearly independent symmetric bases belonging to the inequivalent irreducible representations if four oscillatory quantum numbers aj all are different. The number of the linearly independent bases will decrease if some oscillatory quantum numbers aj are equal to each other. For example, when two and only two quantum numbers coincide with each other, there exist 12 linearly independent symmetric bases belonging to five irreducible representations ([4] @ [2,2] @ 2 [3,1]@ [2,1,1]).
26. Calculate the representation matrices of the generators (1 2) and (1 2 3 4 5 ) of the S5 group in the irreducible representation [2,2, I]
by the tabular method. Solution. From Problem 15, the orthogonal standard Young operators for the Young pattern [2,2,1] of S5 are J+ [E (2 5)] and y p ,2 5 A, 5 5. ! Now we will calculate the representation matrices of the elements (1 2) and (1 2 3 4 5) in this representation [2,2,1] by the tabular method.
232
Problems and Solutions in Group Theory
From the tables, the representation matrices for the generators (1 2) and (1 2 3 4 5) in the representation [2,2,1] respectively are
[g fl
1 0  1 0
; [; ;
1
111
0 0
,
p1;
.
0 0 0 01
0 0
1 0 0
27. Respectively calculate the real orthogonal representation matrices of the transpositions Pa of the neighbored objects in two equivalent irreducible representations [23] 21 [16] x [3,3] of SS, and find the similarity transformation matrix X between them. Solution. The standard Young tableaux for the Young pattern [23] and the Young pattern [3,3] are listed from the smallest to the largest, respectively
Permutation Groups
233
12 12 13 13 14 34 35 24 25 25 56 46 56 46 36
123 124 125 134 135 456 356 346 256 246
Obviously, each standard Young tableau in [23] is the transpose of the corresponding one in 3,3], but the larger in [23] becomes the smaller in [3,3]. According to Eq. 6.12), the representation matrices of the transpositions P, for the neighbored objects in the representations p3]and [16] x [3,3] respectively are
t
representat ion [23]
representation [16] x [3,3]
PI :
[;H ;Al i ) ,
1 0 0
0
0
0 0
1
0
01
P2+
0 1
o & o
0
; a; o; ) ,
0 0
0
0 0
O A
0 0 0 0
+[; a
 2 0 0 1 0 0
1
0
0
0a 0 1 0
0
1
1
2
a 0
3 0 0 0  3 0
P3:
g [o0
0 3 0 4 ) > 0 1 0
0 4 1
0 0
i( i
,
d 1
 d o
i
0
0 0 0 0
i3:3:) 0 0 3
P 5
:
[;;;!l ['P,H i i] ;)
0 0 1 0 0 0
0 0 0 0  1 0
0 010 0 0 0 1
Each matrix in [16] x [3,3] can be obtained from the corresponding one in [23] by a similarity transformation X = Y Z . Y is a diagonal matrix diag{  1, 1, 1,  1 , l ) . Since the nondiagonal matrix entries of P, in the representation [23] appear only in 12, 13, 24, 34 and 45 rowcolumns, Y changes the signs of all nondiagonal matrix entries of P, in [23]. The
234
Problems and Solutions in Group Theory
matrix entries in 2 are all vanishing except for those on the minor diagonal line which are 1. 2 transpose the matrix in [23] with respect to the minor diagonal line. Namely,
0 0 0 01 0 0 0 1 0
X1[23]X = [16] x [3,3],
X=
This method is suitable for other pair of the representations with the associated Young patterns.
28. Calculate the character of each class of the permutation group
S5
in
the representation [2,2,1] by the graphic method.
Solution. The characters of each class of are calculated in the following table.
S5
in the representation [2,2,1]
class Regular application Application parity
(I5> neglected
P3J>
14 24 3 1 1
(1J2> 13 23 2 1 1
(12,3> 12 33 3 1 1
(2,3) 11 22 2 1 1
(L4) 12 22 2 1 1
(5)
no
,
[(e>l
5
0
29. Calculate the characters of each class of the permutation group
$ 3
in the representations [3,2,1], [3,3], and [23] by the graphic method, respectively.
Solution. We calculate the characters in three representations of SS by the graphic method in the following table, where we also list the permitted regular application. The regular application of the identity (1") is neglected. The Young pattern [23]is the one associated with the Young pattern [3,3]. The regular applications for [Z3] are neglected because they can be obtained from those for [3,3] by transpose. The Young pattern [3,2,1] is a selfassociated Young pattern, so the characters of the classes composed of the odd permutations are vanishing.
Permutation Groups
Characters in the ii reducible representations of s6 Regular application neglected 1 2 3 1 2 4 1 3 4 1 2 5 135 455 355 255 345 245 124 133 134 334 244 234 113 122 123 2 2 3 133 1 2 3 123 144 123 124 134 44 24 444 344 244 4 3 122 133 122 33 2 3 333 3 2 111 1 2 2 111 1 1 2 22 12 222 122
2 1
235
I
(3113) (3,2,1)
I
2
0
l
l
1
1
(4,2)
1
0
1
1
123 333 112 222 122 22 2
30. Calculate and fill in the character tables of S3, Sq, Sg, SS and S7 by the graphic method. Solution. Here we will not give explicitly the regular applications, but only list the calculation results in the table. The character of each class in the antisymmetry representation [In] whether the permutation in shows the class is even or odd. n ( l ) denotes the number of elements in the class ( l ) .Instead of showing the pair of associated representations, only one of them is shown for simplicity.
236
Problems and Solutions in Group Theory
(3.12) (3,2j (4,l) (5)
\ I
20 20 30 24
I
1 1 1 1
1 1 0 1
1 1 1 1 0
2 0 0 0 1
Permutation Groups
237
6.5
The Inner and Outer Products of Representations
The direct product of two irreducible representations of a permutation group S, is called the inner product of two representations for S,, which is reducible in general. There is no special graphic rule for the reduction of the inner product of representations for the permutation group. The CG series and the CG coefficients for S, can be calculated by the character method just like the usual finite group:
*
Since the irreducible representations of S, are real, the multiplicity [cJ]) is totally symmetric with respect to three representations. This symmetry leads to some common property for the CG series of S,. Letting [A] and [?;I be two associated Young patterns, we have
.([A],
[TI,
(6.17)
In the reduction of the inner product of two irreducible representations of S,, there is one identical representation [n] if and only if two representations are equivalent, and there is one antisymmetric representation [ln] and if only if the Young patterns of two irreducible representations are associated. We are going to apply the concepts of the subduced representation and the induced representation, discussed in $3.3, to the permutation group. The subgroup H = S, x S, of the permutation group G = Sn+, is the direct product of two subgroups S, and S,, where S, consists of the permutations among the first n objects in the (n m) objects and S, consists of the permutations among the last r n objects. The order of G is g = (n+rn)!. The order of the subgroup H is h = n!m!and its index is n = g / h . Denote by R j H , 2 5 j 5 n, the left cosets of H in G where the element Rj is a permutation transforming the first n objects to n different positions in all (n rn) objects. Rj is not unique, but we assume that Rj has been chosen. For convenience, we denote the subgroup H by R1H with R1 = E .
*
+
+
238
Problems and Solutions in Group Theory
Let L be the group algebra of G. Denote by e[W] primitive idemthe potent in L,and by Le[W] minimal left ideal generated by &", where the the Young pattern [w] with ( n m) boxes describes the irreducible representation of G. Let P" c LC be the group algebra of the subgroup H . Denote by e[A]e[p] primitive idempotent in Lnm, by Ln"e[A]e[p] the and the minimal left ideal generated by e[A]e[p] , corresponding to the irreducible representation of H described by the direct product of two Young patterns [A] x [ p ] ,where the numbers of the boxes in two Young patterns [A] and [p] are n and m, respectively.
+
A subduced representation from an irreducible representation [w] G of with respect to the subgroup H is generally reducible and can be reduced with respect to the irreducible representations [A] x [p] of H ,
(6.18) From the viewpoint of the group algebra, Eq. (6.18) shows that there are ayp linear independent vectors e[A]e[pltae[w] , which map the left ideal Lce[W] onto the minimal left ideals for H , respectively,
*
Equation (6.19) gives the linear expansion of the new basis in the minimal left ideal in terms of the bases in L e [ W ] The coefficients in the expansion . constitute the columnmatrices in the similarity transformation matrix X. The primitive idempotent e[']e[p]in Lnm is also an idempotent in L, but generally not primitive. The left ideal Ce[A]e[p] generated by e[A]e[p] in C corresponds to a reducible representation of G, which is the induced representation from the representation [A] x [p] of H with respect to G. For the permutation group, the induced representation is called the outer product of two representations, denoted by [A] 8 [ p ] . This representation can be reduced with respect to the irreducible representations [u] of G,
*
From the viewpoint of the group algebra, Eq. (6.20) shows that there are b'fD linear independent vectors e [ w ] t ~ e [ x ] which, map the left ideal e[~]
Permutation Groups
239
Le[x]e[p] onto the minimal left ideals for G, respectively,
(6.21) Equation (6.21) gives the linear expansion of the new basis in the minimal left ideal in terms of the bases in Le[’]e[p]. The coefficients in the expansion constitute the columnmatrices in the similarity transformation matrix Y . Due to the symmetry between left and rightideals, the numbers of the linear independent vectors e[Xle[plt,e[W] e[w]t~e[x]e[pl equal, a& = and are b!p. This is an example of the Frobenius theorem (see Chapter 3). There is a graphic rule, called the LittlewoodRichardson rule, for calculating the multiplicity b z p in the reduction of the outer product of two irreducible representations for the permutation group, although it can be calculated by the character method as usual. The calculation method is as follows. Arbitrarily choose one of the Young patterns in an outer product [A] @ [ p ] ,say [ p ] . Usually, choose the Young pattern with the less boxes. Assign each box in the Young pattern [p] with its row number, namely assign the box in the j t h row with j . Now, attach the boxes assigned with 1 to another Young pattern [A] from its right or from its bottom, and then attach the boxes assigned with 2, and so on, in all possible way subject to the following rule. After attaching the boxes assigned with each digit, say j , the resultant diagram satisfies the following conditions:
1) The diagram constitutes a Young pattern, namely, the diagram is lined up on the top and on the left without unconnected boxes in each row such that the number of boxes in the upper row of the diagram is not less than that in the lower row, and the number of boxes in the left column is not less than that in the right column. 2) No two j appear in the same column of the diagram. 3) We read the attached boxes from right to left in the first row, then in the second row, etc., such that in each step of the reading process, the number of boxes assigned with the smaller digit must not be less than that assigned with the larger digit.
*
In each way subject to the above rule, the resultant diagram [w] after attaching all boxes assigned with digits denotes an irreducible representation [w] of Sn+m which appears in the reduction of the outer product [A] @ [ p ] .
240
Problems and Solutions an Group Theory
If the same Young pattern [w] is obtained in different ways, the number of the ways is the multiplicity of the irreducible representation [w] in the reduction. Due to the Fkobenius theorem, the LittlewoodRichardson rule can also be used for calculating the reduction of the subduced representation from the irreducible representation [w] of Sn+, with respect to the subgroup S, x S,. In the calculation, all the Young patterns [A] with n boxes and all the Young patterns [p] with r boxes are taken to constitute the irren ducible representation [w] of Sn+m with the multiplicity b;p, if [w] appears in the reduction of the outer product [A] 8 [p]according to the LittlewoodRichardson rule with the same multiplicity.
31. Calculate the ClebschGordan series of the inner product of each pair of the irreducible representations for the permutation groups S3, S4, S5, s and S7 by the character method. 6 Solution. We can calculate the CG series for the reduction of the inner product of two representations by Eq. (6.16). We only list some results for the CG series, and neglect the remaining CG series which can be obtained from them by Eq. (6.17).
*
For the S3 group: For the S4 group:
[2, I] x [’&I]N [3]@ [I3] [2, I]. @
[3,1] x [3,1]= [4] @ [3,1]@ [221@ [2, 121, [3,1] x [Z2]
p”3
= [3,11@ [2, 121,
[41CB [221 @ [141.
p21
For the S5 group:
[4,1] x [4,13 = [5] @ [4,13 @ [3,21 @ [3,121, [4, 1 x [3,2] = [4,1] @ [3,2] @ [3,121 P2,11, 1 @ [4, I]
[ 3 , 1 2 1 = [4, 1 @ [3,21 @ ~ 3121 CB [22,11@[2, 131, 3 ,
[3,21 [3,21 = [51 p, 11@ [3,21 (3,121@ [22,11 @ [2,131, 3 3 p, 21 [3,121 p, 1 @ p, 21 2[3,121@ p,1 @ p, 131, [ 3 , q [3,121 [5]@ [4, 11 @ 2[3,21@[ 3 , q @ 2[22,11 @ [2,131@~ 5 1 .
=
For the
s 6
group:
Pe rrnu t at ion Group
241
For the
S7
group:
242
Problems and Solutions in Group Theory
Permutation Groups
243
32. Calculate the ClebschGordan coefficients for the reduction of the selfproduct of the irreducible representation [3,2] of S g . Solution. This is a typical example for calculating the reduction of a reducible representation with a higher dimension for a finite group. Usually in the reduction of a reducible representation, one may try to find one or a few commutable elements in the group such that their eigenvalues can designate the bases for each irreducible representation without degeneration. This method has been used in the calculation for the ClebschGordan coefficients of the groups T, 0 and I in Problems 2527 of Chapter 3. For the permutation group S,, when n > 5, the dimension of the irreducible representation is quite high. It is hard to find the eigenvalues of some commutable elements to designate the bases of each irreducible representation without degeneration. For example, one can designate the bases by the eigenvalues of two commutable elements (1 2) and (3 4 5), where the number of the different pairs of the eigenvalues is six. However, there is a 16dimensional representation [3,2,1] in Sg. In the present Problem we are going to introduce another method for calculating the CG coefficients,
244
Problems and Solutions in Group Theory
which will be suitable for the representations with higher dimension. From Problem 31 we have
We take the orthogonal bases where the representation matrix of each transposition Pa of two neighbored objects is known. The basis state before reduction is denoted by Ip)lv),where p and Y both run from 1 to 5, respectively. Denote by I ][A], p ) the basis state after reduction, where p runs from 1 to the dimension of [A]. There is a characteristic in this example that two representations in the direct product are the same. The expansion of the new basis I ][A], p ) can be chosen to be symmetric or antisymmetric in the permutation of two states in the product. For simplicity, we use the short notation for the basis state with the given symmetry in the permutation: I P , v ) S = IP)IY) k IY)IP) and I P , V ) A = IP)Iv)  Iv)IP) When P = v, only symmetric state exists, and we just use the basis state 1p)lp) without the subscript S. The method is outlined as follows. For a given representation, i I the representation matrix of P is diagonal. The eigenvalue of P is used to classify the basis states. The basis state with the eigenvalue 1 of P is i called the Atype basis, which has the following expansion
The basis state with the eigenvalue —1 of P1 is ccalled the Btype basis, which has the following expansion
The action of the transposition Pa on the basis state is
where the representation matrix of Pa in [A] is known. The representation are matrices of Pa in the representation [3,2]
Permutation Groups
245
1 0 0 0 0
PI:
2
1 0 10 0  1 0 0 0 o 1
),
p2:;[:
0
0 0 0 0  1 1Jsoo 0 P32[ 0 0 0 0 0 1 , 1 3
 1 o f i o 0  1 0 4 4 0 1 0 2 0 ' 0 04 00 0 1 0
Js
0 0
ooool
P42
0 0 3 0 0 0 0  3
0 0  1 4 0 0 & 1 0 0 2 0 0 0 la too 0 4 1
For each irreducible representation [A], we first choose a basis state which is the common eigenstate of as many transpositions P a as possible, and write its expansion according to its eigenvalue of PI. Then, in terms of the representation matrices of P and the orthonormal condition we can a determine the coefficients in the expansion. From the Schur theorem, two basis states belonging to two inequivalent irreducible representations or belonging to the different rows of a unitary irreducible representation are orthogonal. If the multiplicity of [A] in the reduction is one, the expansion must have a definite symmetry in the permutation. If it is larger than one, we can choose the expansions having definite symmetry. From the first basis state we are able to calculate the remaining basis states by Eq. ( c ) , which have the same symmetry in the permutation as that the first basis state has. The orthonormal condition and the symmetry in permutation can be used for check. First, we calculate the basis state in the representation [5], which is the identical representation with dimension one. The basis state II[5]) is given in Eq. (a). All representation matrices of Pa in [5] are 1. According to Eqs. (c) and (d), we have from the action of P 2
246
Problems and Solutions in Group Theory
Comparing the coefficient of each term, especially the coefficients of the terms which do not appear in 11[5]), we obtain a2 = a3 = a5 = a = 0, 6 a4 = ale, a7 = all, a8 = a12 and a9 = ~ 1 3 namely ,
Applying
P3
to 11[5]),we have
Applying
P4
to 11[5]),we have
Thus, a1 = ag. After normalization we obtain
Second, for the fourdimensional representation [4,1], we obtain the representation matrices of Pa from Eq. (12)
Permutation Groups
247
P : 1
O 0 O O ) ) 1 0 0 0 0 1 0 0 0 0 1 3 0 0 0 p 3 ' 3 O Jgf i 1O 0 ) 1 0 l
p 2 : 1 ( o02 O 2 0
2
(
0 0  1 &
7
.:(aj
oofi
1 a
:) ; ;:).
1
0
0
0 0
0 3
0
0 1
The solutionm is
The following calculation is similar. We only list the method and the results. Applying P3 to [4,1], 2} gives
Thus,
248
Problems and Solutions in Group Theory
Applying P to 1J[4,1],3),we have 2
Thus,
Note that 11[4,1],4) is a Btype basis state. Third, we calculate the basis states in the representation [3,2] where the representation matrices of Pa were given in Eq. (d). 11[3,2],1) is an Atype basis state. Applying PI and P2 to 11[3,2],1) one by one leads to a formula similar to Eq. (e):
Applying
P4
to 11[3,2], 1)) we have
a1
+ 2a4 + 2a9 = 0,
a9
2al
+ 4a4  6 ~ = 0. 9
Thus, a1 = 2a4 and
= 0. The normalization condition leads to
Applying
P 3
to 11[3,2],1),we have
Thus,
Permutation Groups
249
Thus,
Thus,
Thus,
Fourth, we calculated the basis strates in the sixdimensional representation [3,1], where the representation matrices of Pa are
100 0 0 0 010 0 0 0 001 0 0 0 0001 0 0 0 0 0 0 1 0 0 0 0 0 0 1
2 0 0 01 0
P2 : 
0
0
&0 i
PI :
1 2
0 0 1
ofio 1 o o f i o
0 0 0 0  4 0 0 0  1 6
0 0 0 0 0 1 0 0  1
0
0
0
P3 :
3
1
6 0
0 0
0 0 0 0 3 0 0 0 0 0  1 6
1 0 0 0 0 30 0
0
0
1
'
0
0
*
'
p4:
4
1
0 0 0 0 0 a 0l O
P2
0 o 0
 1 6 0 a 1 0 0 O 0 04 1 O
11[3,12], 1) is an Atype basis. Applying PI and leads to a formula similar to Eq. (e):
to 11[3, 12], one by one 1)
250
Problems and Solutions an Group Theory
Applying
P4
to 11[3,13 1) changes its sign ’,
a8)/2. The orthogonal condition Thus, a1 = 0 and a4 = a9 = &(a7 between 11[3,12],1) and 11[22,1],1) leads to a4 = a9 = 0. Then, a7 = a8. The basis state 11[3,12], is antisymmetric in the permutation. After 1) normalization we have
+
Applying
P 3
to 11[3,12], we have 1))
Thus,
Applying
P4
to “[3,13 2), we have ’,
Thus,
Applying
P 2
to 11[3, 12], we have 2))
Thus,
11[3,1’1, 4) is a Btype basis state. Applying
P 4
to 11[3, 11 4), we have ’,
Permutation Groups
25 1
Thus,
Applying
P 3
to 11[3, 12],5), we have
Thus,
Fifth, we calculate the basis states in the fivedimensional representation
[22, 1 , where the representation matrices of Paare 1
1o&io
00 0 00 0 3 0 0  3
0
1
0 0 1 0 0 0 0
o&io
0 0 0 0 0
1
0
1Ao
0  2 0 0
0 0
0  1 6
The basis state 11[22, 1],5) is the common eigenstate of P, P 2 and P 4 . We I begin the calculation with this state. 11[22, 13,5) is a Btype basis with the form giver
p 2
252
Problems and Solutions an Group Theory
The solution is bl = b2 = b3 = b4 = 0, b5 =  b 6 , b l l = b12, namely
b7
= blo, b8 =  b g and
Applying
P4
to 11[2” 1],5), we have
The solution is b5 = bll = f i ( b 7  b8)/2. The orthogonal condition between 11[22,1],5) and 11[3,12], leads to b5 = b l l = 0 and b7 = b g . After 6) normalization we obtain the symmetric expansion
The solution is
Thus,
Permutation Groups
253
Thus,
Thus,
IP2, = (2&)' 11,1)
{ IW4  13)13)  1414) + 15)15) + & l V ) s }
 a 1 o 0 d l 0 0 0 0  2 0 0 0 02 4 0 0
0 4 0

Finally, we calculate the basis states in the fourdimensional representation [2, 13], where the representation matrices of Pa are
p1:
(
1 0 0 0 010 0 0 0  1 0 0 0 01 3 0 0
,
p 2 : 1 (
2
)
'
P 3 4 ;
\ O
2:
0
'0) ,
p4:;(
01 4 5
a 1
').
03
0 o
The basis state )/[2,13], is the common eigenstate of P I , P2 and P3. We 4) start the calculation with this state. 11[2, 13], is a Btype basis. Applying 4) P 2 to 11[2, 13],4), we obtain a similar formula to Eq. (g):
Applying
P 3
to 11[2, 13],4), we have
Thus, b5 = b7 = b8 = 0. After normalization we obtain the antisymmetric expansion
Applying P to I1[2, 13], we have 4 4),
254
Problems and Solutions in Group Theory
The solution is
Thus,
Thus,
33. Calculate the reduction of the following outer products of the representations in the permutation groups by the LittlewoodRichardson rule: (1) [3,2,118 ~31, (2) [3,218 ~ 211, (3) P, 1 8 [4, z31. , 1 Solution. (1)
The dimensions of the representations on both sides are
x
9! 6! 3!
16 x 1 = 1344 = 105
+ 162 + 120 + 189 + 168 + 216 + 216 + 168.
Permutation Groups
255
The dimensions of the representations on both sides are
8!
~5~2=560=28+64+14+2~70+56+90+42+56+70.
5! 3!
(3)
256
Problems and Solutions an Group Theory
The dimensions of the representations on both sides are
13! x 2 x 300 = 171600 = 12012 + 9009 + 12870 11583 3 2 x 21450 3! lo! 5005 10296 8580 12870 15015 8580 17160 5720.
+
+
+
+
+
+
+ +
+
34. Calculate the reduction of the subduced representations from the fol
lowing irreducible representations of
s 6x33: 3
S6
with respect to the subgroup
Solution. (1)
[4,2]
2
[3] x [3] @ [3] x [2,1] @ [2,1] x [3] @ [2,1] x [2,1].
The dimensions of the representations on both sides are
9=1x1+1x2+2x1+2x2. (2)
[22,121 fi [2, 1 x [2, 1 CD [2, 11 x 3 1
[i31 CB [PI x
[2, 1 @ [i31 x 1
p31.
The dimensions of the representations on both sides are
9 = 2 x 2 + 2 x 1+1x 2 + 1 x 1.
Permutation Groups
257
(3)
The dimensions of the representations on both sides are
5 = 1x 1
+ 2 x 2.
35. Calculate the similarity transformation matrix in the reduction of the
subduced representation from the irreducible representation [3,3] of SS with respect to the subgroup S3BS3.
Solution. The reduction of the subduced representation from [3,3] of SS with respect to the subgroup S36& was calculated in Problem 34:
The standard Young tableaux for the Young pattern [3,3] were given in Problem 18. Denote by Y1 the Young operator for the standard Young tableau
The standard bases in the representation space C of [3,3] are
From the Fock condition (6.5) we have
{ E + (1 4) { E + (1 5) { E (1 6) {E (3 4)
+ +
+ (2 4) + (3 4)) Yl = 0, + (2 5) + (3 5)) Yi = 0, + (2 6 ) + (3 6)) Yi = 0, + (3 5) + (3 6)) Yl = 0.
The idempotent of the onedimensional representation [3] x [3] of the subgroup S 3 8 S 3 is
258
Problems and Solutions an Group Theory
From Eq. (6.19), the basis state for the representation [3] x [3] is
The idempotent of the fourdimensional representation [2,1] x [2,1] of the subgroup S38S3 is
Since y3y1 = 0, from Eq. (6.19), we choose the basis state for the representation [2,1] x [2,1] as
'$1
= Y3(3 4)Yl = (3 4) {E (1 2)  (1 4)  (2 1 4)) {E
+
= (3 4) {2E  (1 4)  (4 2)) { E
= {2(3 4)
+ 2(4 3 5)  4(4 3 6)  (3 4 1)  (4 1 3 5) + 2(4 1 3 6)  (3 4 2)  (4 2 3 5) + 2(4 2 3 6)} Y1 = 2b2 + 2b3  4(3 6)bl  (1 4)bl  (1 5)bl + 2(6 l ) b l  b4  b5 + 2(6 2)bi = 2b2 + 2b3  b4  b5 + [bl + b2 + b4] + [bl + b3 + b5]  2bl + 6 [bi + b2 + b3] = 6b1 + 9b2 + 9b3.
+ (3 5)  2(3 6)} Y1
+ (3 5)  (3 6)  (5 3 6)) YI
Then, we have
Thus, we obtain the similarity transformation matrix X:
36 0
6 9
3 0
6 0
3 0
0
The generators of
S38S3
0
P4,
9
and
P . 5
are P I , P 2 ,
Their representation
Permutation Groups
259
matrices in [3,3] can be calculated by the tabular method:
1 0 0  1 1 0 1 0  1 0
1 0 0 0 0 0 0 0 1 0
0 0 0 0 1 0 0 0 1 0
0 0 0 11 0 0 0 01
Their representation matrices in [3] x [3] @ [2,1] x [2,1] are
1 0 0 0 0 0 1 0  1 0  ( ~ l ) =0 l @ ( ~  ~ ) ~ ( ~ ~ ) = ~ o o ~ 1 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 , 0 1 0 0 0 0 0 1 0 0 1 0 0 0 0 0 1 1 00  0 ) , 1 ( P 4 ) = 1 @ ~ ~ ; ) x ( ~ 0 ~ ) ( 0 0 D  1 0 0 0 1 o 1
j,
1 0 0 0 0
0 0 0 0 0 1 1 0 0 0 0 0
0 01 0 0 0 0 0 0 . 0 1 1 0
1 1
260
Problems and Solutions an Group Theory
It is easy to check that X satisfies the similarity transformation
36. Calculate the representation matrices of the generators of S4 in the induced representation from twodimensional irreducible representation [2,1] of S 3 with respect to Sq.
Solution. For simplicity, let A and B respectively be the representation matrices of the generators of S 3 in the twodimensional representation [2, I]:
Denoting by we have
&,
p = 1 and 2 , the state bases in the representation space,
2
2 u=l
u=l
This twodimensional space is invariant for the S 3 group, but not for Sq. The S 3 group is a subgroup of S 4 with index 4 . Extend the twodimensional space:
The eightdimensional space is invariant for S4. In fact,
Permutation Groups
26 1
Therefore, the representation matrices of the generators (1 2) and (1 2 3 4) of S in the induced representation are: q
D(12)=
(
A 0 0 0 A O), O A O O
D(1234)=
(
O O O A
0 B O 0
0 O B O 0 0 ABOO O A O
)
*
It is a reducible representation of Sq, which can be reduced into the direct sum of the representations [3,1],[2,2] and 12, 12].
37. Calculate the similarity transformation matrix in the reduction of the
outer product representation [2, I] 8 (21 of the permutation group.
Solution. The outer product representation [2, I] @ [2] is the induced representation from [2,1] x [2] of the subgroup Ss xS2 with respect to the group S5. Its reduction can be calculated by the LittlewoodRichardson rule:
[2,1] 6 [2] = [4,13 @ [3,2] 43 [3, 12] p2,1 3 @ 1 120 ~ 2 ~ 1 = 2 0 = 4 + 5 + 6 + 5 .
6x2
The idempotent of the twodimensional representation [2,1] x [2] of the subgroup S3@S2 is
The standard bases in the representation space of [2,13 x [2] are
From the Fock condition (6.5) we have
and the useful formula for the later calculation,
[E + (1 2)] [E  (1 3)] bl = [2E  (1 3)
 (1
2)(1 3)] bl = 3 h 
262
Problems and Solutions in Group Theory
Extending the twodimensional space, we obtain the bases
b3
b5
= ( 1 4)bl, = (2 4 ) h ,
= ( 1 4)bz = ( 1 4 ) ( 2 3)bl, be = ( 2 4)bz = (4 2 3)bl,
b4
The representation matrices of the generators P1 = (1 2) and W = (12345) of S5 in the out product representation [2,1] []2 for the bases are
Through a direct calculation we have
After the similarity transformation X, the out product representation [2, I] 8 [2] becomes a direct sum of four irreducible representations:
Permutation Groups
263
where R = Pi and W . By the tabular method we obtain
0[4qp1) =
(
1 0 0  1
0 0 1  1 0 0 0  1
’) ,
0[4111(W) =
(’
 1 1 0 0
 1 0 0 1O )  1 0 0 0
’
111  1 0 0 0 10  1 0 0 0 10
10 0 1 0 0 1 0 1 0
111 0 0 0 1 0 01 1 0 0 101 0 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0
i
,
[ 1 0  0 01 11) ol 1 D[22Jl(q)= 0 0 1 0 0 0 0 0  1 0 0 0 0 01
D[22Jl(w)=
[;m;mcl)
11 1 1 0 111 0 0 0 0 1 0 0
.
Now, we calculate the similarity transformation matrix X by Eq. (6.21). The idempotent of the fourdimensional representation [4,1] of S5 is
From Eq. (6.21) we obtain the expansions of the basis states T+@”’ in the extending space:
$~”] =
of [4,1]
C
+
20
bvXv3
= Y2Y1
= [ E T t l 4)
+ (2 4) + (1 5) + (2 5) + (1 4)(2 5)]
+
b5 b15]
x [E (1 211 [E + (4 511 [E  (1 3 1 bi 1 = 6 [ E (1 4) (2 4) (1 5) (2 5) (1 4)(2 5)] bi
+ + + + = 6 [bi + + + b g + b i i +
b3
,
264
Problems and Solutions in Group Theory
The idempotent of the fivedimensional representation [3,2] of S5 is
From Eq. (6.21) we obtain the expansions of the basis states in the extending space:
$F’21of [3,2]
Permutation Groups
20
265
’2I !$
=
C buXu7= (4 5)$F’21
u=l
= 3 [bl
+ b9 + b l l + 2b4  2b5  2b6 + b7 + 2b16  b17
I”!+
=
C b,X,g
20
= (2 3)(4 5)$~!’~]
The idempotent of the sixdimensional representation [3,12] of
S5
is
From Eq. (6.21) we obtain the expansions of the basis states $ J ~ ’ ~ ’ ’[3,12] of in the extending space:
20
266
Problems and Solutions in Group Theory
$,F912]=
20
bvXu13 = ( 2 3)q!Jf3”21
The idempotent of the fivedimensional representation [22,1 of 1
S5
is
Y5
=
From Eq. (6.21) we obtain the expansions of the basis states +F2’l1 [22, 1 of 1 in the extending space:
Permutation Groups
26 7
$f2’11=
20
bvXv19 = ( 2 3)(4 5)llfz2’11
This page intentionally left blank
Chapter 7
LIE GROUPS AND LIE ALGEBRAS
7.1
Classification of Semisimple Lie Algebras
An element R in a Lie group G can be described by a set of independent real parameters ( T I ,r2,. . .). Those parameters vary continuously in a given Euclidean region, called the group space. The number g of the parameters, which is equal to the dimension of the group space, is called the order of the group G. In this chapter we only discuss the Lie group whose group space is simplyconnected. A Lie group is said to be compact if its group space is a closed region in a Euclidean space. A Lie group is said to be simple if it does not contain any nontrivial invariant Lie subgroup. A Lie group is said to be semisimple if it does not contain any Abelian invariant Lie subgroup including the whole group. Any Lie group of order one is a simple Lie group, but not a semisimple Lie group. A simple Lie group whose order is more than one must be a semisimple Lie group.
*
*
The generators I A in a representation D ( R ) of a Lie group G are defined
as
A
The generators I A in any representation D ( R ) of G satisfy the common commutation relation:
[ I A , 1131
=iC
D
cA,”ID,
(7.2)
where the real parameters CA,”, independent of the representation, are called the structure constants of the Lie group.
269
270
Problems and Solutions in Group Theory
The real linear space spanned by the bases ( 4 1 ~is)called the real Lie algebra if the product of two vectors X and Y in the space is defined by the Lie product:
*
If the linear space is complex, the algebra is called the complex Lie algebra, or briefly, the Lie algebra. A real Lie algebra of a compact Lie group is compact. A Lie algebra of a simple or a semisimple Lie group is simple or semisimple, respectively. In fact, a simple Lie algebra does not contain any nontrivial ideal, and a semisimple Lie algebra does not contain any Abelian ideal, including the whole algebra. Therefore, a onedimensional Lie algebra is simple, but not semisimple. In the rest of this chapter, if without special indication, “a simple Lie algebra with the dimension larger than one” will be simply called as “a simple Lie algebra” for simplicity. It is proved that a semisimple Lie algebra can be decomposed into the direct sum of some simple Lie algebras. The Killing form QAB of a Lie group or its Lie algebra is defined from its structure constants cA::
*
is a real symmetric matrix of dimension 9 . When the parameters of a given Lie group are changed, the generators transform like a vector, and the structure constants as well as the Killing form transform like a tensor:
QAB
The Cartan Criteria says that a Lie algebra is semisimple if and only if its Killing form is nonsingular, and a real semisimple Lie algebra is compact if and only if its Killing form is negative definite.
*
Lie Groups and Lie Algebras
271
If { H i , 1 5 j 5 ! is the largest set of the mutually commutable gener} ators in a simple Lie algebra L, the set constitutes the Cartan subalgebra and e is called the rank of the Lie algebra L. It is proved that the group parameters in a simple Lie algebra L can be chosen such that the generators H j and the remaining generators E, satisfy
*
[ H j , E a ] = @a,
( N,,pE,+p
when a'
+p
is a root,
(7.6)
lo
and the Killing form becomes
9jk =  s j k ,
the remaining cases,
Sap = S(CY)p,
gj,
= gaj = 0.
(7.7)
The real vector d in the !dimensional space is called the root in L, and the space is called the root space. The space spanned by the roots in a simple Lie algebra of rank l is &dimensional. The roots kG appear in pair. Except for the pair of roots, any two roots are not collinear. There is a onetoone correspondence between the generator E, and the root a'. Hj and E, are the CartanWeyl bases for the generators in the simple Lie algebra L. In a semisimple Lie algebra, we can constitute a root chain from a root a' by adding or subtracting another root several times. Without loss of generality, if d+np, qag 5 n 5 pap, are the roots, and a'+ (pap and a'  (qap 1)pare not the roots, where pap and qap are both nonnegative integer, it is proved that
*
p
+
+ 1)p
When d =
A root is called positive if its first nonvanishing component is positive, otherwise it is negative. Sometimes, Hj is called the generator corresponding to the zero root, which is l degeneracy. A positive root is called a simple root if it cannot be expressed as the nonnegative integral combination of other positive roots. Therefore, any positive root is equal to a nonnegative integral combination of the simple roots, and any negative root is equal to
*
p,the root chain consists of a', 0 and 6.
272
Problems and Solutions an Group Theory
a nonpositive integral combination of the simple roots. The difference of two simple roots is not a root. From Eq. (7.8), the simple roots are linearly independent. Thus, the number of the simple roots in a trank simple Lie algebra is equal to t . The angle between two simple roots has to be equal to 5.n/6, 3n/4, 2n/3 or n/2, and correspondingly, the ratio of the length squares of the two simple roots is 3, 2, 1, or no restriction, respectively.
There are four series of the classical simple Lie algebras (At, Be, Ce and De) and five exceptional simple Lie algebras ( G 2 , Fq, Eg, E7 and Eg), where the Lie group of At is SU(t + l),the Lie group of Be is S 0 ( 2 t + 1), the Lie group of Ce is Sp(2t), and the Lie group of D is SO(2l). e The Dynkin diagram for a simple Lie algebra is drawn by the following rule. There are one or two different lengths among the simple roots in any simple Lie algebra. Denote each longer simple root by a white circle, and denote each shorter simple root, if it exists, by a black circle. Two circles denoting two simple roots are connected by a single link, a double link, or a triple link depending upon their angle to be 2n/3, 3n/4 or 57r/6, respectively. Two circles are not connected if two simple roots are orthogonal and the ratio of their lengths is not restricted. The Dynkin diagrams for the simple Lie algebras are listed in Fig. 7.1.
G2
*
Ap, t
2 1 1
F4
2
3
c2e1
e
E 6
Bp,
t2
3
OGO1 2
3
e2e1
c
L2.L
o==e
1 2 1 2 3 4 5
ce,e 2 2
Dp, .f 2 4
E7 tt++~==~ 1
1 2
3
e2e1
e
&
2 3 4 5 6 1 2 3 4 5 6 7
o<'~
1
2
3
e3c2
e
& 3
Fig. 7.1 The Dynkin diagrams of the simple Lie algebras.
The simple Lie algebra can also be described by the Cartan matrix. The Cartan matrix for a simple Lie algebra with t simple roots r, is an &dimensional matrix A:
*
Lie Groups and Lie Algebras
273
The diagonal element of A is always equal to 2, and its nondiagonal element can be equal to 0, 1, 2 or 3. In comparison with the Dynkin diagram, if two simple roots r, and r, are disconnected, A,, = A,, = 0, and if they are connected by a single, double or triple link, and the length of r, is not less than the length of r,, A,, = 1 and A,, = 1, 2 or 3, respectively. The Dynkin diagram or the Cartan matrix gives all the information of the simple Lie algebra, such as the angles and the lengths of the simple roots as well as all roots. The roots can be calculated recursively as follows. Any positive root can be expressed as a sum of the simple roots:
a'=
C C,r,,
e
C 2 0, ,
(7.10)
C, Cvis called the level of the root d. Any simple root is a root of level one. If all positive roots with the level less than n have been found, we want to judge whether a' r, is a root where 6 is a root of level (n  1). C alculat e e e q  p = r (a'/.,) = A&,, if a'= Cur,. (7.11)
+
u=l
u=l
Since q is known, one can calculate p from Eq. (7.11) to determine whether a' rP is a root or not.
+
1. Prove the number pap qap 1 of roots in a root chain a'+ nB, qap n 5 pap, in a simple Lie algebra is less than five.
+ +
5
Solution. Prove this problem by reduction to absurdity. If the length of the root chain is larger than four, one can redefine the root a' such that the following five vectors are all nonzero roots:
a'  2&
a'
6,
a',
a'+
/?,
a'+ 28.
However, the following four vectors are not roots:
(a' f 2B) + a' = 2(a' f
B),
(a' f 28)  a' = *2B.
Thus,
o = r[(a'f2 f i ) / a ' ] =
di1(a'* 2B) a' = 2 f 2d;l
Adding two equations leads to contradiction. This completes the proof. S: ie qap +pap 1 5 4, both qap and pap are less than four. Thus, n I r ( w > IL 3.
r
p . a' .
)
+
274
Problems and Solutions in Group Theory
2. Prove that the difference of two simple roots is not a root, and there are
l simple roots in a simple Lie algebra of rank l . Solution. We first prove the difference of two simple roots rcl and r, not to be a root by reduction to absurdity. Suppose rp  r, = a'. If a' is a positive root, rp = r, + a' is the sum of two positive roos. If a' is a negative root, r, = r P + (a') is the sum of two positive roos. It contradicts with the definition of a simple root. Thus,
The angle of two simple roots is not an acute angle. It is obvious that the number of the simple roots is not less than l?, because any root can be expressed as a linear combination of the simple roots. If the simple roots are linearly independent to each other, the number of the simple roots must be l . Now, we prove it by reduction to absurdity. If there is a linear relation among simple roots, we separate the terms with the negative coefficients from those with the positive ones:
P
v
where cp and d, are all positive, and the values of p and u are different. Since a # 0,
0 < a2 = Cc,d,r,
This completes the proof.
'r,
5 0.
3. Prove that the angle between two simple roots has to be equal to 57r/6, 37r/4, 2 ~ / 3 7r/2, and correspondingly, the ratio of the length squares or of the two simple roots is 3, 2, 1, or no restriction, respectively. Solution. Calculate the square of cosine of the angle between two simple roots:
Since rP and r, is not collinear, 1 cos2 81 < 1, and the integer can only be equal to 3, 2, 1 or 0. Namely, the angle 9 can be 57r/6, 37r/4, 27r/3 or 7r/2. Without loss of generality, let the length of rp be not less than that of r,,
Lie Groups and Lie Algebras
275
Due to the restriction for their product, r'(r,/rp) has to be equal to 1 or 0, and I'(r,/r,) is equal to 3, 2, 1 or 0. The ratio of those two number is
or depending on the angle of two simple roots to be 5 ~ / 6 , 3 n / 4 , 2 ~ / 3 ~ / 2 .
4. Calculate the Cartan matrix of the Lie algebra E6.
Solution. The Cartan matrix of E6 is a sixdimensional matrix with the diagonal elements to be 2. According to the enumeration for the simple roots in its Dynkin diagram, the rowcolumn numbers of the positions of the nondiagonal elements, which are 1, are 12, 23, 34, 36 and 45, or vice versa,
A = I j l 1 0l o 0 0 1 2 1 0 0  1 2 0 0 0  1 0 0  1 0
0 0 ) 0 1 1 0 2 0 0 2
'
5 . Draw the Dynkin diagram of a simple Lie algebra where its Cartan
matrix is as follows, and indicate the enumeration for the simple roots.
Solution. The Dynkin diagram consists of four circles. Since A32 = 2, the circle designated by 2 is white corresponding to a longer simple root, and the circle 3 is black corresponding to a shorter simple root. The circles 2 and 3 are connected by a double link. Since A12 = A21 = A34 = A43 = 1, the circles 1 and 2 are connected by a single link, so be the circles 3 and 4. Thus, the circle 1 is white and the circle 4 is black.
r\
1
2
3
4
276
Problems and Solutions an Group Theory
6. Calculate all positive roots in the C2 Lie algebra.
Solution. There are two simple roots in
C2
where r2 is the longer simple root, and r1 is the shorter. Their angle is
3n/4. Usually, the length square of the longer root is taken to be 2. The
Cartan matrix of
C2
is
2 2 A=(12).
First, we calculate pl in the root chain rl have
+ plr2. From Eq. (7.11)) we +
p = I'(rl/r2) = A21 = 1. 1 Thus, rl r2 = (el e2) is a root, but rl 2r2 is not a root. Then, we calculate p in the root chain r2 p2r1. Since 2
+
+
+
in addition to r1 r2, we know 211 r2 = fie1 is a root. 3rl 1'2 is not a root. rl + r2 is the only root of level two, and 211 + r2 is the only root of level three. Since (2rl 212) is twice of rl r2, it is obviously not a root. Therefore, the C2 Lie algebra contains four positive roots and four negative roots. The rank of C2 is 2, and its order is ten. This result can be generalized to the Ce Lie algebra. Ce contains 2C2 roots: f f i e j , & m ( e j ek) and m ( e j  ek). The rank of Ce is C and its order is C(2C+ 1).
+
+
+
+
+
+
7. Calculate all positive roots in the B3 Lie algebra.
Solution. There are three simple roots in B3:
rl = el  e2, r2 = e  e3, 2 r3 = e3,
where rl and r2 are the longer roots with the length square 2, and r3 is the shorter root. The angle between rl and r2 is 2 ~ / 3 The angle between r2 . and r3 is 3n/4. rl is orthogonal to r3. The Cartan matrix of B3 is
Lie Groups and Lie Algebras
277
r2
+ p3r3. From Eq. (7.11)) we have +
We first calculate pl, p2 and p3 in the root chains
rl
+ plr2, r1 + p2r3,
r2
Thus, there are two roots of level two: rl + r 2 and r 2 +r3. We also know that 2r3 is a root of level three, and rl 2 1 2 and r2 331'3 are not the roots. Then, we calculate p4, p5, p and p7 in the root chains (rl r 2 ) p4r1, 6 (ri r2) p5r3, (r2 13) p6rl and (12 r3) p7r2. Since
+
+
+ +
+ +
+ +
+
+
in addition to r 2 2r3, there is another root of level three: rl r 2 + 13. Besides, rl r2 2r3 is a root of level four. Adding r 2 or r3 to the root r2 21'3 is not a root. w e are going to calculate p8 in the root chain (rl r 2 13) ~ 8 1  2 . Since rl + r3 is not a root, we have
+ + + + + + +
+
Thus, rl + r 2 +2r3 is the only root of level four. Adding r3 to it is not a root. We want to calculate pg and plo in the root chains (rl r 2 2r3) pgrl and (rl r 2 2r3) ~ 1 0 1  2 . Since
+ +
+
+ +
+
we obtain the only root of level five: rl 2r2 213. Besides, rl 312 213 is not a root. 2rl 2r2 2r3 is twice of a known root, so it is not a root. Now, we are going to calculate pll in the root chain (rl 2 1 2 213) +pllr3. Since
+
+
+ +
+ +
+ +
there is no root of level six. In summary, the B3 Lie algebra contains three roots of level one (three simple roots: rl = el  e 2 , r 2 = e2  e3 and r3 = e3), two roots of level two (rl + r 2 = el e3 and r 2 +r3 = e 2 ) , two roots of level three (rl + r 2 + r 3 = el and r 2 2r3 = e2 e3), one root of level four (rl r 2 213 = el e3), and one root of level five (rl 2r2 + 21'3 = el + e 2 ) . The rank of B3 is 3, and its order is 21. This result can be generalized to the Bc Lie algebra. Bc contains 212 roots: f e j , f ( e j + e k ) and (ej  e k ) . The rank of Bc is l , and its order is l ( 2 l + 1).
+
+
+
+ +
+
278
Problems and Solutions in Group Theory
8. Calculate all positive roots in the
D4
Lie algebra.
D4:
Solution. There are four simple roots in
rl = el  e2,
r2
= e 2  e3,
r3 = e3  e 4 ,
r 4
= e3
+
e4,
where the length squares of four simple roots are all 2. The angle between r 2 and any other simple root ra, a # 2, is 27r/3, and the other pair of simple roots is orthogonal to each other. The Cartan matrix of D4 is
Lie Groups and Lie Algebras
279
r3
+ 14) + p20r1 and ( r 2 + r3 + r 4 ) + p21r2. Since
there is only one root of level four rl r 2 r3 r 4 = el e3. Adding rl, r3 or 14 to the root of level four is not a root. We calculate p 2 2 in the root chain (rl 12 r3 14) ~ 2 2 1  2 . Since
+ + +
+
+ + + +
3r2 r3 or
we obtain the only root of level five: rl 2r2 r3 r 4 = el e2. rl + 13 14 is not a root. It is easy to know those vectors by adding r l , r 4 to the root of level five is not a root, because
+ +
+
+ +
+
Therefore, the D4 Lie algebra contains four roots of level one, three roots of level two, three roots of level three, one root of level four, and one root of level five. The rank of D4 is 4, and its order is 28. This result can be generalized to the Dc Lie algebra. Dc contains 2 l ( l  1) roots: f ( e j e k ) and (ej  e k ) . The rank of D is l , and its order is l ( 2 l  1). c
+
7.2 Irreducible Representations and the Chevalley Bases
Discuss an irreducible representation of a simple Lie algebra where the representation matrices of Hj are diagonal:
*
where Im) is the basis state in the representation space, and the vector m in an ldimensional space is called the weight of Im). The &dimensional space is called the weight space. A weight m is said to be multiple if the number of the basis states Im) satisfying Eq. (7.12) in the representation space is larger than one. Otherwise, it is a single weight. From Eq. (7.6) we have
280
Problems and Solutions in Group Theory
Therefore, E, is called the raising operator and E, the lowering operator, if a' is a positive root. In a given irreducible representation, one can construct a weight chain from a given weight m and a root d:
*
+ nd,  q 5 n 5 p , are weights, p 2 0, q 2 0, m + ( p + 1)G and m  ( q + 1)G are not the weights.
m
(7.13)
It can be proved that
I? (m/Z) = q  p = integer.
(7.14)
The weight m' = m  GI' (m/d) is related to the weight m by a reflection with respect to the plane through the origin and orthogonal to the root G. m and m' have the same multiplicity and are called the equivalent weights. This reflection is called the Weyl reflection. All the mutually equivalent weights constitute the Weyl orbit, and their number is called the Weyl orbital size. The Chevalley bases H p , E p and F p , 1 5 p 5 l , are another set of bases for the generators in a simple Lie algebra, which can be expressed as the linear combinations of the CartanWeyl bases Hj and E,:
*
They satisfy the following commutation relations: (7.16)
and the Serre relations:
(7.17)
The generator E, corresponding to a nonsimple root G can be calculated by the commutator (7.6). The main merit of the Chevalley bases is that three Chevalley bases H p , E, and Fp with the same subscript p span a subalgebra, denoted by d p ,which is isomorphic onto the Lie algebra A1
Lie Groups and Lie Algebras
281
(the SU(2) group). In fact, they satisfy the same commutation relations as those satisfied by the generators 22'3, T+ and 2' in A1:
In the weight space, define a set of new bases w,, called the fundamental dominant weight, satisfying
*
r (w,/r,)
= d;'
(wp  r,) = (2w,  r,)
/ (r,  r,)
= b,,.
(7.18)
In comparison with Eq. (7.9), we can relate the fundamental dominant weight wp with the simple root r, by the Cartan matrix
L
p=l
e
u=l
In the bases w,, all components of the roots and the weights are integers:
(7.20)
The weight m' equivalent to m with respect to a simple root r, can be expressed as m' = m  r,r (m/rp) = m  mpr,. (7.21)
Rewrite the weight chain in Eq. (7.13) with respect to a simple root rp where m + p r p is redefined as m:
m, m  r p ,
*
. . . ,m  qr,,
q = r(m/rp) = m,.
(7.22)
The states corresponding to the weight chain constitute a (q+l)multiplet of the subalgebra d . The basis states in the irreducible representation space , constitute some multiplets of the different subalgebras d p . If all weights in the weight chain (7.22) are single, the representation matrix entry of Fp for the multiplet is
The representation matrix of E, is the transpose of that of F,. If the , multiple weight exists, the state in the multiplet of A is a suitable com
282
Problems and Solutions in Group Theory
bination of the basis states, calculated by the commutators (7.16) and the orthonor ma1 condition. For an irreducible representation with a finite dimension, there is a state JM)with the highest weight M:
*
The highest weight M in an irreducible representation must be a single weight. Usually, an irreducible representation can be denoted by its highest weight M. A representation is called fundamental if its highest weight is the fundamental dominant weight. Any basis state Im) with the weight m in the irreducible representation space can be obtained from the highest weight state by the actions of the lowering operators. Thus,
Cm is called the height of the basis state Im). The height of the highest weight state is zero.
Two irreducible representations are conjugate with each other if the lowest weight of one representation multiplied by 1 is equal to the highest weight of another representation. The irreducible representations is selfconjugate if its lowest weight multiplied by 1 is equal to its highest weight.
*
A weight is called a dominant weight if its components are nonnegative integers. Due to Eq. (7.14), the highest weight must be a dominant weight, and any dominant weight can be the highest weight in one irreducible representation. However, for a given irreducible representation, there may exist a few dominant weights with multiplicities, among them only one dominant weight is the highest weight of this representation. The dimension of an irreducible representation is equal to the sum of the multiplicities of the dominant weights multiplied by their Weyl orbital sizes. Therefore, the dominant weights play a very important role in the calculation for the basis states and the representation matrices of the generators.
The method of the block weight diagram is effective for calculating the representation matrices of the generators in an irreducible representation. In a block weight diagram for an irreducible representation, each basis state corresponds to a block filled with its weight. For convenience, the negative component of the weight is denoted by a digit with a bar over it: Fi = m. The blocks corresponding to the multiple weight are enumerated. The block
*
*
Lie Groups and Lie Algebras
283
for the highest weight is located on the first row. The remaining blocks are arranged downwards as the height increases. The blocks for the weights with the same height are put in the same row. Beginning with the highest weight, from each positive component my > 0, we draw m, blocks downwards. They constitute a weight chain of length (m, 1) with respect to the simple root r,. The states corresponding to the weight chain constitute a multiplet of d . Two neighbored blocks , are related by a link denoting the application of the lowering operator F,. Usually, in the block weight diagram we connect two blocks by different links if their basis states are related by the different lowering operators. It is better to indicate the corresponding representation matrix entry of the lowering operator behind the link. Usually, the matrix entry is neglected when it is equal t o one. If all weights in a weight chain are single, the representation matrix entries of the lowering operator F, between two basis states corresponding to the neighbored weights can be calculated by Eq. (7.23). If a multiple weight appears in the weight chain, the state in the multiplet may be a combination of the basis states. The combination as well as the representation matrix entries of F can be determined by Eq. (7.16) and the orthonormal ’ condition. When a dominant weight appears, we determine its multiplicity which is equal to the number of paths along which the dominant weight is obtained from the highest weight. If the representation matrix entry in a path is calculated to be zero, the multiplicity decreases. All weights equivalent to the dominant weight has the same multiplicity as the dominant weight has. Continue this method until all components of the weight are not positive. In the completed block weight diagram, the number of blocks with the same height first increases and then decreases as the height increases, symmetric up and down like a spindle.
+
For a tworank Lie algebra the weights can be drawn in a planar diagram, called the planar weight diagram. The planar weight diagram of an irreducible representation is the inversion of that of its conjugate representation with respect to the origin. The planar weight diagram of a selfconjugate irreducible representations is symmetric with respect t o the inversion. Note that the simple roots as well as the fundamental dominant weights are not required to be along the coordinate axes. The representation in whose space the basis is the generator of the Lie algebra is the adjoint representation. Therefore, the root coincides with
*
*
284
Problems and Solutions in Group Theory
the weight in the adjoint representation, and the root space coincides with the weight space. To draw the block weight diagram or the planar weight diagram for the adjoint representation is another method for calculating all roots in a simple Lie algebra.
9. Draw the block weight diagrams and the planar weight diagrams of the representations (1,0), (0, l ) ,(1, l ) ,(3,0), and (0,3) of the A2 Lie algebra (the SU(3) group).
Solution. The Cartan matrix A and its inverse A' of A2 are
2 1 A  1 = 1( 2 1 3 1 2
)
*
The relations between the simple roots and the fundamental dominant weights are rl = 2wl  w2,
w 1
r2 = w1+ 2w2,
= (1/3) (2rl
+ ra) ,
w2
= (1/3) (rl
+h).
The Weyl orbits of the dominant weights ( l , O ) , ( O , l ) , ( l , l ) ,(3,O) and (0,3) respectively are: (1,o) r , ( i , i ) l
r , (o,i), Z
(0,1) r2, ( 1 , i ) %
(i,o),
(3,O) r , (3,3) % (0,3), l
(0,3) 3 ( 3 , s ) %
(5,O).
The block weight diagrams of the representations ( l , O ) , (0, I), (1, l),(3,0), and (0,3) are given in Fig. 7.2. In the diagrams, the single link means the action of Fl and the double link means that of F2. First, we discuss the fundamental representation (1,O). From the highest weight (1,O) we obtain a doublet of A1 with ( i , l ) . Then, from the weight (i, we obtain a doublet of .Ap with ( 0 , i ) . The weight (0,T) con1) tains no positive component. It is the lowest weight in the representation (1,O). The dimension of the representation (1,O) is three, where three weights are equivalent to each other. The representation matrix entries of
Lie Groups and Lie Algebras
285
the lowering operators are
Second, we discuss the fundamental representation (0,l). From the highest weight ( 0 , l ) we obtain a doublet of d 2 with (1,i). Then, from the weight (1,i) we obtain a doublet of ,A1 with (T,O). The weight (T,O) contains no positive component. It is the lowest weight in the representation ( 0 , l ) . The dimension of the representation (0,l) is three, where three weights are equivalent to each other. The representation matrix entries of the lowering operators are
It is worthy to notice that the block weight diagram of ( 0 , l ) can be
obtained from that of (1,O) by turning upside down and changing the signs of all weights. In fact, the lowest weight (0,i) in the representation ( l , O ) , times by 1, is the highest weight in the representation (0,l). Two representations are conjugate to each other.
0, i
Fig. 7.2 The block weight diagrams of some representations of A2
286
Problems and Solutions in Group Theory
Third, we discuss the adjoint representation (1,l) of A2. From the 1 and a doublet of highest weight (1,l)we obtain a doublet of A with (i,2) A with ( 2 , i ) . Then, from the weight ( 2 , i ) we obtain a triplet of A with 2 1 (0,O) and (2, l), and from the weight (i, we obtain a triplet of A with 2 ) 2 (0,O) and (1,2). There are two paths to the dominant weight (0, 0), which may be a double weight. We define two orthonormal basis states (0,O)l and (0, O, where (0,O)lbelongs to the triplet of A with (2, i)and (2,l), ), 1 and (0, O, is a singlet of 41. The state (0,O) in the triplet of A with (i, ) 2 2) and (1,2) to be a combination of (0, O), and (0,0)2: has
Applying El F = F2E1 I(1,1, (T,2)), we have 2 to )
Thus, a = Choosing the phase of 1 1 1 , 0 0 2such that b is a (,)(,)) positive real number, we have b = = Let
m.
d
m
m.
Applying
EF = FE 22 22
+ H 2 to l(1, l), (0,0)1), have we
(, Choosing the phase of l1 l), (1,g)) such that c is a positive real number, we have c = @, and then, d = @.From the weight (1,T) obtain we a doublet of A with (i,i), from the weight (2,l) we obtain a doublet 1 and of A with (1, Since the weight (i,i) equivalent to (1,I), it is a 2 i). is single weight and is the lowest weight in the representation (1,l). The representation (1,l) is selfconjugate. The dimension of the representation (1,l)is eight containing six equivalent single weights and one double weight (0,O). The representation matrix entries of the lowering operators are
Lie Groups and Lie Algebras
287
We see from the block weight diagram of the adjoint representation (1,l) that the A2 Lie algebra contains three positive roots:
2 w l  w2 = r l ,
 w1+2w2 = 12,
w1+
w2 = r1 +r2.
Fourth, we discuss the representation (3,O). From the highest weight
(3,O) we obtain a quartet of A with (1,l), ( i , 2 ) and (3,3), where the 1
representation matrix entries of F are 1 2, and respectively. The weight ( 1 , l ) is a dominant weight, which is a single weight because there is only one path to ( 1 , l ) lowered from the highest weight. All weights equivalent to (3,O) or ( 1 , l ) are single, too. Then, from the weight (1,l) we obtain a doublet of A2 with (2, i). From the weight (T, 2) we obtain a triplet of d 2 with (0,O) and ( l , z ) , and from the weight ( 2 , i ) we obtain a triplet of d with (0,O) and (2,l). There are two paths to the dominant 1 weight ( O , O ) , which may be a double weight. We define two orthonormal basis states (0, O), and (0,0)2, where (0,O)l belongs to the triplet of d with 1 ( 2 , i ) and (2, l),and (0, O), is a singlet of d1. The state (0,O) belonging the triplet of A2 with ( i , 2 ) and (1,2) has to be a combination of (0, O), and (0,0)2:
a, a,
F21(3,0),(T,2))=a1(3,0),(0,0)1) +b1(3,0),(0,0)2),
Applying E1F2 = F2E1to 1(3,0),(i, we have 2)),
a2 + b 2 = 2 .
~i1(3,0), (T,2)) = d N 3 , o), ( 2 , i ) )
= F2Ei1(3,0), ( i , 2 ) ) = 2F21(3,0), (1,1)) = 21(3,0),(2,T))Thus, a = fi and b = 0. It means that the weight (0,O) is single. The basis state ((3,0),(0,0)2) does not exist, and ((3,0),(0, O),) can be denoted by i(3, o), (0,o)). From the weight (1,2) we obtain a doublet of A1 with (i,i). From the weight (3,3) we obtain a quartet of d with (2, 1), (7,T), and (0, Z), where 2 the representation matrix entries of F are 2 2, and respectively. The weights ( 2 , l ) and (T, 1)are single because they are equivalent to ( 1 , l ) . Thus, the state ( 2 , l ) coincides with the state in the triplet of A1 with (2,T) and (0, 0), and the state (T, T) coincides with the state in the doublet
a, a,
288
Problems and Solutions in Group Theory
of d with (1,2). The block weight diagram of the representation (3,O) 1 is complete. The dimension of the representation (3,O) is ten, containing three single dominant weights (3,0), ( 1 , l ) and (0, 0), with the Weyl orbital sizes 3, 6 and 0, respectively. The representation matrix entries of the lowering operators are
The lowest weight (0,z) in the representation (3,0), multiplied by 1, is the highest weight of the representation (0,3). Therefore, the representation (0,3) is conjugate with the representation (3,0), and its block weight diagram is upside down of that of (3,0), where all weights change their signs.
t"'
(i,i)
Fig. 7.3 The planar weight diagrams of some representations of
A2.
Lie Groups and Lie Algebras
289
In the rectangular coordinate frame, the simple roots and the fundamental dominant weights of A2 are
Physically, people prefer to put el along the ordinate axis, and e2 along the abscissa axis for A2. The planar weight diagrams for the representations ( l , O ) , (0,1), (1,1),(3,0), and (0,3) are given in Fig. 7.3. Generally, the planar weight diagram for the representation ( 0 , M ) of A2 is a straight regular triangle where all weights are single, and that of ( M ,0) is a regular triangle upside down. The planar weight diagram of ( M , M ) of A2 is a regular hexagon, where the weights on the boundary are single, and the multiplicity increases as the position of the weight goes to the origin. 10. Draw the block weight diagrams and the planar weight diagrams of two fundamental representations and the adjoint representation (2,O) of the C2 Lie algebra.
Solution. The Cartan matrix A and its inverse A' in
2 2 A  l = 1 2 2 ( 2 1 2
C2
is
)

The relations between the simple roots and the fundamental dominant weights are
rl = 2 w l w2,
r2
=  2 w l + 2w2,
The Weyl orbits of the dominant weights (1, 0)) (0, 1)) and (2,O) respectively are:
(1,o) r ,(i, l 1)
3 (1,i) % (i,o), 3 2 3 Il (2,O). ( ',
(0, 1) r ,( 2 , i ) r ,(2,i) 3(o,i), 2 l
(2,O) r ,(2,2) l
Their Weyl orbital sizes are all 4. The block weight diagrams of the representations (1,0), (0, l),and (2,O) are given in Fig. 7.4. In the diagrams, the single link means the action of F and the double link means that of 1
F2.
290
Problems and Solutions in Group Theory
Fig. 7.4
The block weight diagrams of some representations of C2.
First, we discuss the fundamental representation (1,O). From the highest weight (1,O) we obtain a doublet of d with (i,1). Then, from the 1 weight (7,l) we obtain a doublet of A with (1,i). From the weight (1,i) 2 we obtain a doublet of d with ( i , O ) , which contains no positive component. l The dimension of the representation ( 1 , O ) is four, where four weights are equivalent to each other. The representation matrix entries of the lowering operators are
Second, we discuss the fundamental representation (0,l). From the highest weight ( 0 , l ) we obtain a doublet of d with ( 2 , i ) . Then, from a the weight (2,T) we obtain a triplet of d with (0,O) and (2,l). The 1 representation matrix entries of F1 in the triplet are both The weight (0,O) is a dominant weight. There is only one path to (0,O) lowered from the highest weight, so (0,O) is single. From the weight (2,l) we obtain a doublet of A with ( O , i ) , which contains no positive component. The 2 fivedimensional representation ( 0 , l ) contains two single dominant weights ( 0 , l ) and (0,O) with the Weyl orbital sizes four and one, respectively. The representation matrix entries of the lowering operators are
a.
Lie Groups and Lie Algebras
291
At last, we discuss the adjoint representation (2,O). From the highest weight (2,O) we obtain a triplet of A1 with (0,1> and (2,2), where the The dominant weight representation matrix entries of F are both 1 ( 0 , l ) is single because there is only one path to (0,l) lowered from the highest weight. From the weight ( 0 , l ) we obtain a doublet of A2 with (2,i). From the weight (2,2) we obtain a triplet of A2 with (0,O) and (2,2), and from the weight (2, i we obtain a triplet of A1 with (0,O) and ) (2,l). There are two paths to the dominant weight (O,O), which may be a double weight. We define two orthonormal basis states (0, O), and (0, O),, where (0,O)l belongs to the triplet of A with (2,i) and (2, l),and (0,0)2 1 is a singlet of 41. The state (0,O) in the triplet of A2 with (2,2) and (2,T) has to be a combination of (0,O)l and (0,0)2:
a.
Thus, a = 1. Choosing the phase of ((2,0),(0,0)2) such that b is a positive real number, we have b = \/2i" = 1. Applying E2F2 = F2E2 H to 2 P,01, (O,O)l), we have
+
Choosing the phase of 1(2,0), (2,2)) such that c is a positive real number, we have c = 1, and then, d = 1. All weights in the representation (2,O) are single except for (0,O). From the weight (2,2) we obtain a triplet of d 1 with (0,T) and (z,O), where the representation matrix entries of F are both 1 From the weight (2,l) we obtain a doublet of A2 with (0,T). Since the weight (0,T) is single, it also has to belong to the triplet of 41 with (2,2) and (3,O). Thus, the block weight diagram of the representation (2,O) is complete. The representation (2,O) contains two single dominant weights (2,O) and (0,1), and one double dominant weight (0,O). The dimension of the representation (2,O) is 1 x 4 1 x 4 2 x 1 = 10. The representation matrix entries of the lowering operators are
a.
+
+
292
Problems and Solutions in Group Theory
We see from the block weight diagram of the adjoint representation (2,O) that the C2 Lie algebra contains four positive roots: 2 w l  w2 = r1, w2 = rl  2 w l + 2w2 = 1'2, 2wl = 2rl
+r2.
+ 12,
In three representations (1,0), (0, 1), and (2,O) of C2, the lowest weights are equal to the highest weight times by 1, respectively. It means that three representations are all selfconjugate. In the rectangular coordinate frame, the simple roots (see Problem 6) and the fundamental dominant weights of C2 are
Thus, we obtain the planar weight diagrams for the representations ( l , O ) , ( 0 , l ) and (2,O) as follows.
11. Draw the block weight diagrams and the planar weight diagrams of three representations ( O , l ) , ( l , O ) , and (0,2) of the exceptional Lie algebra G2.
Lie Groups and Lie Algebras
293
Solution. The Cartan matrix A and its inverse Al in G2 are
A = ( 3 2
'),
1'2
i).
= w1+ 2w2,
The relations between the simple roots and the fundamental dominant weights are
rl = 2wl  3w2,
The Weyl orbits of the dominant weights (0, l ) , (1,0), and (0,2) respectively are: ( o , i ) 3 ( i , i ) %( i , 2 ) r ,( i , 2 ) 2
r ,( i , i ) 3 (o,i), l
(1,o)
3i , 3 ) % (2,s) 3 r2, (1,s) l (i,o), ( (2,3) r,
(0,2) rz, ( 2 3 1p1, (2,4) r ,(2,;I) r ,(2,2) 3 (0)Z). 2 l Their Weyl orbital sizes are all 6. First, we discuss the fundamental representation ( 0 , l ) . From the highest weight ( 0 , l ) we obtain a doublet of d 2 with ( 1 , i ) . Then, from the weight (1,i ) we obtain a doublet of d1with ( i , 2 ) . From the weight (i, 2) we obtain a triplet of d 2 with (0,O) and (1)z).The representation matrix entries of F 2 in the triplet are both The weight (0,O) is a dominant weight. There is only one path to (0,O) lowered from the highest weight, so (0,O) is single. From the weight (1,Z) we obtain a doublet of A1 with (1,l). From the weight (1,l) obtain a doublet of d 2 with (O,i), which contains we no positive component. The dimension of the representation ( 0 , l ) is seven, where six weights are equivalent to each other and (0,O) is equivalent only to itself. The representation matrix entries of the lowering operators are
a.
F21(0,
(071)) = KO, 11,( l , i > > ,FlI(0, 11,( l , i ) >= KO, 11,
m),
F21(0,1>,(i, = JZI(0, I), ( O , O ) ) , FZl(0,1), (0,O)) = JZl(O,l>, (1,2)>, 2)) F11(0,1>,(1,W = KO, 11,( i , 1 ) > ,
F2I(O,
11, (i, = I(O,l),(O,i)). 1))
Second, we discuss the fundamental representation (1,O). From the highest weight (1,O) we obtain a doublet of d with ( i , 3 ) . Then, from 1 the weight ( i , 3 ) we obtain a quartet of d 2 with (0, l), ( 1 , i ) and (2,s). The representation matrix entries of F in the quartet are 1 2, and respectively. The weight ( 0 , l ) is a dominant weight. There is only one path to ( 0 , l ) lowered from the highest weight, so ( 0 , l ) is single. All weights
a, a
294
Problems and Solutions in Group Theory
equivalent to (1,O) or (0,l) single, too. From the weight (1,T) we obtain are a doublet of A with (i, From the weight (2,z) we obtain a triplet of A 1 2). 1 with (0,O) and (2,3), and from the weight (i, we obtain a triplet of A 2) 2 with (0,O) and (1,2). There are two paths to the dominant weight (O,O), which may be a double weight. We define two orthonormal basis states (0,O)l and (0,0)2, where ( , ) belongs to the triplet of A with (2,s) 0Ol 1 and (2,3), and ( , ) is a singlet of d . The state (0,O) in the triplet of A 002 1 2 with (i,2)and (1,g)has to be a combination of ( , ) and ( , ) : 0Ol 002
Thus, a = Choosing the phase of I(l,O), ( , ) )such that b is a 002 positive real number, we have b = = Applying E 2 F 2 = FE 2 2 H 2 to 1 1 0 , 0 0 1 we have (,)(,)) ,
m.
+
d
m
m.
Choosing the phase of 1(1,0),(1,2)) such that c is a positive real number, we have c = and then, d = The remaining part of the block weight diagram is easy to calculate. We only list the results in the block diagram and the representation matrix entries of the lowering operators in the following. The representation (1,O) contains two single dominant weights (1,O) and (0,l) with the Weyl orbital size 6 and one double dominant weight (0,O). The dimension of (1,O) is 1 x 6 + 1 x 6 + 2 x 1 = 1 . 4
m,
m.
Lie Groups and Lie Algebras
295
The representation ( 1 , O ) is the adjoint representation of G2. F'rom the block weight diagram we obtain all positive roots of G2 as follows:
At last, we discuss the representation (0,2). From the highest weight (0,2) we obtain a triplet of d 2 with (1,O) and (2,2), where the representation matrix entries of F 2 are both The weight ( 1 , O ) is a dominant weight. There is only one path to ( 1 , O ) lowered from the highest weight, so ( 1 , O ) is single. From the weight ( 1 , O ) we obtain a doublet of d with 1 (i, F'rom the weight (2,2) we obtain a triplet of A1 with ( 0 , l ) and (2,4), 3). and from the weight (i,3) we obtain a quartet of d 2 with (0, l),(1,i ,and ) (2,s). There are two paths to the dominant weight (0,1), which may be a double weight. We define two orthonormal basis states (0,1)1 and (0,1)2, where ( 0 , l ) l belongs to the triplet of d with (2,2) and (2,4), and (0,1)2 1 is a singlet of d1. The state ( 0 , l ) in the quartet of d2, composed of (T, 3), ( O , l ) , (l,i'), (2,3), has to be a combination of (0,O)l and (0,0)2: and
a.
Applying
EF 1 2
= F2E1 to I(0,2), (T,3)), we have
Thus, a1 = 1. Choosing the phase of [(O, 2)) (0,1)2) such that a2 is a positive real number, we have a2 = d m= Since the weight (1,i) is equivalent to the dominant weight (0,1), it is a double weight. Define two orthonormal basis states I(0,2), (1,i)l)and I(0,2), ( l , i ) 2 ) , where I(0,2), (1,i)l)is proportional to F21(0,2), (0,1)1) with a positive real coefficient bl:
a.
Applying
EF 2 2
= F2E2
+ H2 to [(O, 2), (0,1)1) and I(0,2), (0,1)2), we have
296
Problems and Solutions in Group Theory
F21(0,2), ( 1 ) i ) i ) = b41(0,2), (2,3)), F21(0,2), ( l , i ) 2 ) = b51(0,2), (2,Z)). Applying
E F 2 2
= F2E2
+ H2 to 1(0,2),( 1 , i j l ) , we have
hb51(0,2), (1,i)z)
~%F21(0,2), 1 , i ) i ) = bz1(0,2), ( 1 , i ) i ) ( = (F2E2 + H2) I(0,2), ( 1 , i ) l )
= (2 + 1  l > l ( O , 2), (1,i)l) JZI(O,2), (hQ2).
+
Choosing the phase of I(0,2), (2,s)) such that b4 is a positive real number, we have b4 = fi and b5 = 1. F'rom the weight (2,4)we obtain a quintet of d 2 with ( i , 2 ) , (O,O), (l,T), and (2,4), where the representation matrix entries of F 2 are 2, &, fi,and 2, respectively. Since the weights (T, 2) and (1,T) are equivalent to the dominant weight (0, l ) , they are the double weights. We define the 2)2, 2)l orthonormal basis states (T, 2)1, (i, (1,2)1, and (1,2)2, where (i, and (1,2)1 belong to the quintet of A2, and ( i , 2)2 and (1,2), belong to a triplet of A2 with (0,O). Let FlI(0, 21, ( 1 J ) l )= c11(0,2), ( i , 2)l) + c21(0,2), (i, 2)2), F11(o,2)7(17i)2) C3I(o72),(i,2)i) = +c41(0,2),(i,2)2), Applying
E F 2 1
c;
+ c;
= 1, = 1
Ci + C i
= F1E2 to /(0,2),( 1 , i ) l ) and 1(0,2), (1,i)2), we have
E 2 J N m (1,i)l)= 2c11(0,2),(2,4))
Thus, c1 = 1 and c2 = c3 = 0. Choosing the phase of I(0,2), (i, such 2)2) that c4 is a positive real number, we have c4 = 1.
Lie Groups and Lie Algebras
297
In addition to a quintet and a triplet of d 2 , which both contain the weight (O,O), there is another triplet of d containing (0,O). The triplet of 1 A1 consists of (2,3),(O,O), (2,3). There are three paths to the dominant weight (0,O) lowered from the highest weight, so (0,O) may be a triple weight. We define three orthonormal basis states (0, O),, (0, O),, and (0, O),, where (0,O)l belongs to the quintet of A2 composed of (2,4), (T,2)1, (0,0)1, (1,2)1 and (2,3), (0,0)2 belongs to the triplet of A2 composed of ( i , 2 ) 2 , (0,0)2 and and ( O , O ) 3 is a singlet of A2. Thus, we have
(l,a),,
Let
F l
F l
F l
Applying
E F 2 1
= FIE2 to 1(0,2),(2,3)), have we
and d2 = Choosing the phase of I(0,2), (0,0)3) Thus, dl = such that d3 is a positive real number, we have d3 = Applying EIFI = FlEl HI to I(0,2), (0, 0)3), we have
m.
+
m.
Choosing the phase of I(0,2), (2,3)) such that d6 is a positive real number, and then d4 = and d5 = It is easy to we have d6 = calculate the remaining part of the block weight diagram.
m,
m.
298
Problems and Solutions an Group Theory
q
0 ,i
Fig. 7.5.
0 0
Ji
Ji
lT,oI
The block weight diagrams of some representations of G2
.
The results for the representation (0,2) as well as (1,O) and ( 0 , l ) are given in the block weight diagrams (Fig. 7.5). In the diagrams, the single link means the action of Fl and the double link means that of F2. The planar weight diagrams are also given, where the simple roots and the fundamental dominant weights of GZ in the rectangular coordinate frame
Lie Groups and Lie Algebras
299
are
7 3 Reduction of the Direct Product of Representations .
The ClebschGordan (CG) series for the reduction of the direct product M1 x M2 of two irreducible representations of a simple Lie algebra can be calculated by the dominant weight diagrams. The main point is to list all dominant weights N contained in the product space, to count their multiplicities n(M1 x M2,N) in the space, to calculate their Weyl orbital sizes OS(N), and to calculate the multiplicities n(M,N) of those dominant weights N in the representations M contained in the CG series, respectively. First, we list OS(N) and n(M1 x M2,N) of all dominant weights N contained in the product space at the first column of the diagram. Let M1 M2  N = C j cjrj, Cjc j is called the height of the dominant weight N. Arrange the dominant weights N downwards as the height increases. The height of M1 +M2 is zero and is listed in the first row. The representation MI+ M2 is contained in the CG series with multiplicity one. Then, the second column of the diagram lists the dominant weights N in the representation M1 M2 and their multiplicities n(M1 M2, N) (as the superscript). The difference n(M1 x M2,M’)  n(M1 M2,M‘) for the dominant weight M’ with the next smallest height is equal to the multiplicity of the representation M‘ contained in the CG series. If it is not zero, the third column of the diagram lists the dominant weights N in the representation M‘ and their multiplicities n(M’,N) (as the superscript). Continue the calculation to determine the multiplicities of the representations N in the CG series by comparing n(M1 Ma, N) with the sum of the multiplicities of N in the representations, which we have known to be contained in the CG series, until all dominant weights N are calculated. At last, calculate the dimensions d(M‘) of the representations M’ contained in the CG series by
*
+
+
+ +
+
d(M’) =
N
OS(N)n(M’,N),
(7.26)
to see whether their sum is equal t o the dimension of the direct product
300
Problems and Solutions an Group Theory
space:
d(M1 x M2) =
C OS(N)n(M1 x M2,N) = C d(M‘).
N
M‘
(7.27)
Denote by JM1, ml)JM2, ma) the basis state before reduction, and by llM,m) the basis state after reduction. Usually, the basis state IM1, ml)IM2,ma) is briefly written as Iml)lm2) for simplicity.
*
m , M2Mare l where C Mm , , m the ClebschGordan (CG) coefficients. Applying the lowering operator Fp to the basis state, we have:
The main steps of the calculation are as follows. First, the highest weight state in the representation M1 + M2 is
In terms of Eq. (7.29) we are able to calculate the expansions of all basis states in the representation MI+ M2. Then, by the orthonormal condition or by Eq. (7.24) we calculate the highest weight state in the representation M’ contained in the CG series. At last, we calculate the expansions of all basis states in the representation M’ by Eq. (7.29). When M1 = M2, the expansion of each basis state can be symmetric or antisymmetric with respect to the permutation between two basis states Iml)lm2) and Im2)lml). For simplicity, we denote by (sym. terms) or (antisym. terms) the remaining terms obtained from the preceding terms by the permutation, multiplied by 1 or 1, respectively.
12. Calculate the ClebschGordan series and the ClebschGordan coefficients for the direct product representation ( 1 , O ) x ( 1 , O ) of the C2 Lie algebra. Solution. In Problem 10 we have calculated the block weight diagrams of the representations ( l , O ) , ( 0 , l ) and (2,O) of C2. The representation (1,O) contains only one single dominant weight (1,O). The representation ( 0 , l )
Lie Groups and Lie Algebras
30 1
contains two single dominant weights (0,l) and (0,O). The representation ( 2 , O ) contains two single dominant weights (2,O) and (0, 1),and one double dominant weight (0,O). The Weyl orbital sizes of the weights (2,0), (0, l), and (0,O) are 4,4, and 1, respectively. The linearly independent basis states for the dominant weights in the product space are Dominant Weight Basis State No. of States
Thus, we draw the diagram of the dominant weights as follows.
0s x n
4x2 1x4
16
=
1
0
+
5
+
1
Now, we calculate the expansions of the basis states. First, for the representation (2,0), from Eq. (7.30) we have
From the block weight diagram of (2,O) of C2 (see Fig. 7.4) and Eq. (7.29),
we have
According to the orthonormal condition, we obtain the highest weight state in the representation (0,l)
302
Problems and Solutions in Group Theory
Applying the lowering operators to the basis states continuously, we obtain
For the basis states in the representation ( 0 , l ) we have
The expansion of the basis state in the representation (0,O) can be calculated by the orthonormal condition. But now, we prefer to calculate it by the condition (7.24) for the highest weight state. Let
Lie Groups and Lie Algebras
303
Applying El and E2 to it, we have
Thus, a = b = c = d. After normalization we obtain
It can be checked that four basis states with the weight (0,O) are orthogonal to each other. In the permutation between l l and Im2), the representam) tion (2,O) is symmetric, but ( 0 , l ) and (0,O) are antisymmetric.
13. Calculate the ClebschGordan series and the ClebschGordan coefficients for the direct product representation ( 0 , l ) x ( 0 , l ) of the C2 Lie algebra. Solution. One may solve this problem following the way used in the preceding problem. But now, we prefer to calculate the block weight diagram and the CG coefficients with another method, where the CG coefficients for (0,l) x (0,l) and the block weight diagram for (1,l) are calculated at the same time. From the block weight diagram of ( 0 , l ) given in Fig 7.4, we know that the linearly independent basis states for the dominant weights in the product space are as follows, where we neglect the basis state which can be obtained by the permutation of two states I l and I 2 . m) m)
We have calculated the block weight diagrams of the representations ( 0 , l ) and (2,O). However, we did not calculate the block weight diagram of the representation (0,2). The representation ( 0 , l ) contains two single dominant weights ( 0 , l ) and (0,O). The representation (2,O) contains two single dominant weights (2,O) and (0, l ) , and one double dominant weight (0,O). The Weyl orbital sizes of the weights (2,0), (0, l),and (0,O) are 4,
304
Problems and Solutions an Group Theory
4,and 1, respectively. The Weyl orbital of the weights (0,2) contains four weights: (0,2), (4,?), (3,2),and (O,?). F'rom Eq. (7.30) we have the expansion of the highest weight state in
the representation (0,2)
From the highest weight (0,2) we have a triplet of d2:
The dominant weight (2,O) is single in the representation (0,2) because there is only one path to it lowered from the highest weight. However, the multiplicity of the weight (2,O) in the product space is 2. It means that one representation (2,O) is contained in the CG series for ( 0 , l ) x (0,l). From the weight (2,O) we have a triplet of A1,
The dominant weight ( 0 , l ) is single in the representation (0,2) because there is only one path to it lowered from the highest weight. The representation (2,O) contains a single dominant weight (0, l), and the product space contains two basis states with (0,l). It means that the representation ( 0 , l ) is not contained in the CG series for ( 0 , l ) x ( 0 , l ) . F'rom the weight (4,2) we have a quintet of d with ( 2 , i ) , (O,O), (2, l), 1 and (4,2), and from the weight (?,2) we have a triplet of d 2 with (0,O) and (2,?). There are two paths to the dominant weight (0,O) lowered from the highest weight, so (0,O) may be a double dominant weight in the representation (0,2). We define two orthonormal basis states for the weight (0,O). "(0,2), (0,O)l) belongs to the quintet of d1,and 11(0,2),(0,0)2) has to belong to a singlet of St1 because the weight (2,T) is a single weight.
Lie Groups and Lie Algebras
305
We obtain the inner product between this basis state and Il(0,2), (0,O)l) to be Thus,
m.
After normalization we have a = In fact, a can also be calculated by the property of a triplet: u2 + 1/3 = 2. Hence,
m.
It is easy to calculate the remaining expansions of the basis states in the representation (0,2). We list the results as follows.
306
Problems and Solutions an Group Theory
In addition, we have
From the above calculation, we are able to draw the dominant weight diagram for ( 0 , l ) x ( 0 , l ) of Cz and the block weight diagram and the planar weight diagram of the representation (0,2) as follows.
0s x
n
4 x 1 4x2 4x2 1x5 25 1 4 + 1 0 + 1
=
Now, we calculate the expansions of the basis states in the representation
(2,O). Since 11(2,0), (2,O)) is orthogonal to Il(0,2), (2,0)), we have
The remaining basis states are
Lie Groups and Lie Algebras
307
We may calculate the basis state Il(O,O), (0,O)) by the orthonormal condition to the preceding four basis states with the weight (0,O). Here we calculate it by Eq. (7.24). Let
308
Problems and Solutions an Group Theory
From Eq. (7.29) we have
Thus, a = b = c = d = e. After normalization, we have
It can be checked that five basis states with the weight (0,O) are orthogonal to each other. In the reduction of the direct product representation ( 0 , l ) x ( 0 , l ) of C2, the representations (0,2) and (0,O) are symmetric with respect to the permutation between Iml) and Imz), and the representation (2,O) is antisymmetric.
14. Calculate the ClebschGordan series for the direct product representation (1,O) x ( 0 , l ) of the Ca Lie algebra and the expansion for the highest weight state of each irreducible representation in the CG series. Solution. From the block weight diagrams for the representations ( 1 , O ) and ( 0 , l ) given in Fig 7.4, we know that the linearly independent basis states for the dominant weights in the product space are as follows.
There are a few methods to determine the dominant weights and their multiplicities contained in an irreducible representation. One method is to draw the upper part of the block weight diagram until all dominant weights in the representation appear. Another method is to look over the table book, for example, the table book [Bremner et al. (1985)] contains the information for the simple Lie algebras with rank less than 13. Anyway, we know that the representation (1,l) contains one single dominant weight (1,l) and one double dominant weight (1,O). The representation (1,O) contains only one single dominant weight (1,O). The Weyl orbital size of the weight (1,O) is 4. The Weyl orbital of the weights ( 1 , l ) contains eight weights: (1,l ) , ( , ( 3 , i ) ) (3,2), (3,2), (3, l ) , (1,2) and (i,i). 12)) NOW,we
Lie Groups and Lie Algebms
309
are able to draw the diagram of the dominant weights for the direct product representation (1,O) x (0,I) of Ca.
0s x n
8 x 1 4x3 20
=
1 6 + 4
Fig. 7.6. The diagram of the dominant weights for (1,O) x ( 0 , l ) of C2.
The highest weight states of the representations (1,l) and (1,O) can be calculated by Eqs. (7.30) and (7.24). The results are
15. Calculate the CG series for the following direct product representations of the Ga Lie algebra and the expansion for the highest weight state of each irreducible representation in the CG series:
where the dimensions d(M) of some representations M, the Weyl orbital sizes OS(M) of M, and the multiplicities of the dominant weights in the representation M are listed in the following table.
6
1
1
6
6
2
3
1
2
1 1 2
3 3
12
6 6
4
5
5
4 4 3
1 2
2
2
1
1 1 1
1
1
Solution. In Problem 11 we have studied the block weight diagrams of the
310
Problems and Solutions an Group Theory
representations ( 0 , l ) and (1,O) of G2. The representation ( 0 , l ) contains seven single weights
The representation ( 1 , O ) contains 12 single weights and one double weight:
In the following, we first list the linearly independent basis states I l I 2 for each dominant weight in the product representation space, m )m ) where we neglect the basis state which can be obtained by the permutation of two states I l and Im2). Then, comparing the number of the basis m) states of the dominant weight in the product representation space with the multiplicities of the dominant weight in the relevant representations, we are able to calculate the ClebschGordan series. The expansion for the highest weight state of each irreducible representation in the CG series can be calculated by Eq. (7.24). The results are listed in the following. a) The product representation ( 0 , l ) x ( 0 , l ) . Dominant Weight Basis State
No. of States
The expansions for the highest weight states of the irreducible representations in the CG series are
In the permutation between m1) and m2), the representation (0,2)and (0,0) are symmetric, but (1,0) and (0,1 are antisymmetric.
Lie Groups and Lie Algebras
311
OSxn
49
= 2 7
+
+
14 (170)
+
+
7
(0,1)
+
+
1
(0,O)
(0,1) x (0, I)= (0,2)
Fig. 7.7. The dominant weight diagram for ( 0 , l ) x ( 0 , l ) of G2.
b) The product representation (0,l) x (1,O). Dominant Weight
Basis State
No. of States
The expansions for thje highest weight states of the irreducible representations in the CG series are
312
Problems and Solutions in Group Theory
Fig. 7.8. T h e dominant weight diagram for ( 0 , l ) x ( 1 , O ) of G2.
c) The product representation (1,O) x (1,O).
The expansions for the highest weight states of the irreducible representations in the CG series are
Lie Groups and Lie Algebras
313
In the permutation between I l and Im2), the representation (2,0), m) (0,2) and (0,O) are symmetric, but (0,3) and (1,O) are antisymmetric.
Fig. 7.9. The dominant weight diagram for ( 1 , O ) x ( 1 , O ) of G2.
16. Calculate the ClebschGordan series for the direct product representation ( O , O , 0 , l ) x (O,O, 0 , l ) of the F4 Lie algebra and the expansion for the highest weight state of each irreducible representation in the ClebschGordan series, where the dimensions d( M) of some representations M, the Weyl orbital sizes OS(M) of M, and the multiplicities of the dominant weights in the representation M are listed in the following table.
0s
1
24 24 96 24
Solution. F’rom the Cartan matrix of F4 we obtain the relations between the simple roots rj and the fundamental dominant weights w j
The representation (0, 0, 0 , l ) of F contains 24 equivalent single weights 4
314
Problems and Solutions in Group Theory
and one doublet weight (O,O, 0,O):
First, we list the linearly independent basis states I 1 l z for each domm)m) inant weight in the product representation space (0, 0, 0 , l ) x (0, 0, 0, l ) , where we neglect the basis state which can be obtained by the permutation m) of two states I l and Im2). Dominant Weight
Basis States
No. of States
Then, comparing the number of the basis states of the dominant weight in the product representation space with the multiplicities of the dominant weight in the relevant representations, we are able to calculate the ClebschGordan series, as given in the dominant weight diagram. The expansion for the highest weight state of each irreducible representation in the CG series can be calculated by Eq. (7.24). The results are listed in the following.
Lie Groups and Lie Algebras
315
In the permutation between m1) and m2), the representation (0,0,0,2), (0,0,0,1) and (0,0,0,0) are symmettric, but (0,0,1,0) and (1,0,0,0) are antisymmetric.
0s x n
24 x 1 96 x 2 24 x 6
Fig.7.10…The dominant weight diagram for (0,0,0,1)´(0,0,0,1) of F4.
This page intentionally left blank
Chapter 8
UNITARY GROUPS
8.1
The S U ( N ) Group and Its Lie Algebra
The set of all Ndimensional unimodular unitary matrices, in the multiplication rule of matrices, constitutes the SU(N) group. It is a simplyconnected compact Lie group with order ( N 2  1) and rank ( N  1). The generators in its selfrepresentation are traceless hermitian matrices:
*
6cd
[2a(a  1)]1’2
when c < a,
2
(8.1)
&d
0
[(a 1 ) / ( 2 ~ ) ] ” ~ when c = a, when c > a,
L a L N,
where the subscripts a and b are the ordinal indices of the generator, while
c and d are the row and column indices of the matrix. The matrix
Ti:)
is symmetric with respect to both ab and cd, but Tit) is antisymmetric. TA3)is diagonal. Usually, three types of the generators can be enumerated uniformly in the following order 2 1 = Ti:), T2 = Ti;), = T i 3 ) T4 = Ti:), ’ T 3 , T = Ti:), = T J i ) 27 = T23 , T = Ti3),and so on. The generators TA 5 Ts , (2) g Satisfy the normalization condition: TAT') = 6,4B/2. Any element u in SU(N) can be expressed in the exponential matrix form:
N21 N
317
318
Problems and Solutions a Group Theory n
*
Let N
E
l + 1, and define ( 1 + 1)vectors V
a
in the !dimensional space:
v1=
v = z
v 3
=
1
.a={
1
””
d2 l ( l + 1 ) ’ @(cq’
1
) O,
/=,
2a
0,
...,
‘
‘
8
)
0
1
,
....*..........
. .. . .. . .. . .. .. .
v N = {  d X
2 ( l + 1)’
0,
0,
The CartanWeyl bases for the generators in the selfrepresentation of SU(N) can be expressed as
Hj = &T,(!\+,,
Esab= Tab + iTit), (1)
when l!  j
0341
+ 2 = a,
1
[Hj,
Ezob when a
< l  j + 2 < b,
Ea,,l =
(e  j
+ 2)(! j + 1)
when l  j + 2 < a ,
1j+2 when l  j
+ 2 = b,
or l  j + 2 > b ,
where a
< b. Namely,
Unatary Groups
319
.
I
When a < b, Ezabcorresponds to the positive root Gab = &{V,  v }= b rP. Therefore, the Lie algebra of the S U ( l + 1) group is Ae. The rP = w1 we, which is the highest weight largest root in At is c3 = 1 of the adjoint representation of A t .
c/.L=l
+
1. Calculate the anticommutation relations for the generators TA in the
selfrepresentation of SU ( N ).
Solution. The anticommutator for two generators TAand TB is hermitian, but not traceless. Let
P=l
Taking the trace, we have cN = ~ T ~ ( T A T B ) Thus, = BAB.
N21
' I
2. Calculate the structure constants and d A B D in Eq. (8.7) for the
su(3)
group.
Solution. Write the commutation relations for the generators TA in the selfrepresentation of SU(N):
where CAj$'are the structure constants. Multiplying Eqs. (8.7) and (8.8) by To and taking the trace, we have
320
Problems and Solutions in Group Theory
CA,” are totally antisymmetric with respect to three indices, but ~ A B D are totally symmetric. Substituting the generators TA into Eq. (8.9), we obtain the nonvanishing structure constants CA,” and the coefficients ~ A B D of SU(3) as listed in Table 8.1. Table 8.1 The nonvanishing structure constants CAf and the coefficients ~ A B D SU(3). of
ABD
123 147 156 246 257 345 367 458 678
ABD
118
146
157
228
247
256
338
344
I A B D I 355
366
377
448
558
668
778
888
I
3. Calculate the Killing form for SU(N) group.
Solution. The generator Iidjin the adjoint representation is related to the structure constant CA,Dby
(8.10)
Thus, the Killing form g A B is
(8.11)
Being an ( N 2 1)dimensional matrix, g is commutable with any generator Iidjin the adjoint representation of SU(N):
Unitary Groups
321
Due to the Schur theorem, g A B is a constant matrix, the constant c for the SU(3) group by taking A = 3:
= C ~ A B Calculate .
Note that T = Ti:) and TG= Ti:). When generalizing the result to the 4 SU(N) group, the subscript 3 in Ti:) and Ti;) becomes a, 3 5 a 5 N , so that the number in the curve brackets becomes 1 2 ( N  2)/4. Therefore,
+
gAB
=N~AB,
for SU(N).
(8.12)
8.2
Irreducible Tensor Representations of SU(N )
The Chevalley bases for the generators in the selfrepresentation of SU(N) are
*
(EP)cd
=
(Erp)cd
= 'W6(p+l)d,
(Fp)Cd
=
(E'p)cd
= d(p+I)cdpd,
1 5p
5 e = N  1.
(8.13) Note that for a given p, three generators H,, E p and Fp constitute the subalgebra Ap,which is isomorphic onto the A 1 algebra.
*
A covariant tensor
T a l . , . a n of
rank n with respect to SU(N) transforms
as
322
Problems and Solutions an Group Theory
Since the matrix entries ?&idi in Eqs. (8.14) and (8.15) are commutable, the symmetry among the indices of a tensor keeps invariant in the SU(N) transformation. Therefore, the tensor space is reducible, and can be reduced into the direct sum of some irreducible tensor subspaces with the definite permutation symmetry among the indices. In order to describe the symmetry of the indices of a tensor, we define the permutation operators R for the tensor. R is a linear operator by which a tensor T is transformed into a new tensor T'. The product of two permutation operators should satisfy the multiplication rule of elements in the permutation group. Let us begin with a simple example. If
*
R=
(ii;) (iii)
=
=(123)=(12)(23),
then a tensor transforms as
On the other hand, a tensor basis transforms as
They transform in the different but compatible ways. In fact, expanding a tensor T in the tensor basis @&, we have
In general, if
we have
Unitary Groups
323
It is easy to show that the product order of an SU(N) transformation 0, and a permutation R among indices of a tensor are commutable
ROUT= 0,RT.
(8.17)
This property is called the Weyl reciprocity. It is the theoretical foundation for decomposing a tensor space by the Young operators.
obtained by the projection of a Young operator, = y117.Two tensor subspaces denoted by different Young patterns are inequivalent. Two tensor subspaces denoted by different standard Young tableaux of the same Young pattern are equivalent, but are linearly independent. can be An arbitrary tensor basis $?l@ul...un in a tensor subspace denoted by a tensor Young tableau with the Young pattern [A], where the box filled with j in the standard Young tableau yF1is now filled with the subscript a j . For example, for the Young operator $”’ corresponding to the tensor Young tableau of the tensor the standard Young tableau
ex]corresponding to an irreducible representation [A] of SU(N) can be
*
In general, the tensor space ‘T is reducible. The minimal tensor subspace
*
wl,
basis
$’11@ala2a3
is F
l
. For convenience we write
(8.18)
Note that the same tensor Young tableaux in different tensor subspaces denote the different tensor bases. For example, if the Young operator $”I corresponds to the standard Young tableau E, tensor Young l the
l tableau Fdenotes the different tensors in two tensor subspaces:
Due to the symmetry and the Fock condition (6.5) satisfied by the Young operator, the tensor Young tableaux in a given tensor subspace satisfy some linear relations. For example, for the Young operator y?’l1we have
324
Problems and Solutions an Group Theory
E
The relations similar to Eq. (8.19) hold for the general cases. Those relations obviously depend upon the Young pattern, but are independent of the Young operators with the same Young pattern as well as the tensor subspaces. The first equality says that the tensor Young tableau is antisymmetric with respect to any two digits filled in the same column. The tensor Young tableau is equal to zero if there are two equal digits filled in its same column. We may rearrange the digits in each column of the tensor Young tableau by the antisymmetric relation such that the digit increases downwards. The second equality is also called the Fock condition. Any tensor in a tensor subspace $ I’ is a combination of the tensor can be expressed as bases yF1Oa1...,,. yp I’[
*
Since yFIRFY the standard bases in the minimal right ideal of the are permutation group produced by the Young operator yI1,the following tensor bases
YFIR,uu@bl...b , >
bl
5 b2 5    5 bn,
(8.20)
constitute a complete set of the tensor bases in the tensor subspace $’I, where R,, is the permutation transforming the standard Young tableau y;’] to the standard Young tableau We are going to discuss the characteristic of the basis (8.20). Since
ypl.
in , , the tensor Young tableau for the tensor basis Y ~ l R p , @ ~ l . . . the tensor I’ is the same as that for the tensor basis y~’]@b,...,,, tenin the subspace $ sor subspace In fact, if R permutates ~j to j, R ~ ~ o ~ ~ . ~ . ,b , =. , ...bp,.
d’].
Therefore, the box filled with j in the standard Young tableau is filled with T j in the standard Young tableau y;’], and is filled with bj both in the tensor Young tableau y ~ l R p , @ b l . . . and in the tensor Young tableau b,
yrl
I’[ yu
tensor Young tableau is called standard if the digit in each column of the tableau increases downwards and in each row does not decrease right
*A
@b1
...b , .
Unitary Groups
325
wards. Since YLxl is a standard Young tableau and bl 5 b2 5 . . . 5 b,, the tensor Young tableau yLxl@,,,..,n well as the tensor Young tableau as yP f&v@bl...C, is standard. The standard tensor Young tableaux consti[XI tute a complete set of the tensor bases in the tensor subspace The standard tensor Young tableau is the eigenstate of HP where, due to Eq. (8.13), the eigenvalue is equal to the number of the digit p filled in the tableau, subtracting the number of the digit ( p 1). Therefore, the representation matrix of the Chevalley basis HP is diagonal in the tensor bases. The action of F, on the standard tensor Young tableau is equal to the sum of all possible tensor Young tableaux, each of which is obtained from the original one by replacing one filled digit p with the digit ( p 1). If without special indication, we use the notation for the inner product of tensors that the bases @al...a, are orthonormal. Two standard tensor Young tableaux with the different sets of the filled digits are orthogonal, but may not be normalized to the same number. Two standard tensor Young tableaux with the equal set of the filled digits but different filling order may not be orthogonal. To find a set of the orthonormal bases, one need to make suitable combination and normalization from the standard tensor Young tableaux.
+
+
Since RpvYLx1 the standard bases in the minimal left ideal of the are permutation group corresponding to the irreducible representation [A], Eq. (8.21) also shows that the same tensor Young tableaux in all tensor subspace 7;Ix1with the same Young pattern [A] constitutes a complete set of the standard bases in the irreducible representation space of the permutation group, denoted by [A]. Therefore, in the decomposition of the tensor space
*
the tensor subspace corresponds to the irreducible representation [A] of SU(N), and for a given Young pattern [A], the tensor bases with the same span the irreducible tensor Young tableau in the d[xl(S,) subspaces I$' representation space of the permutation group S, denoted by [A]. The dimension d[xl(SU(N)) of the representation [A] of SU(N) can be calculated by the hook rule. In this rule the dimension is expressed by a quotient, where the numerator is the product of ( N m i j ) , and the
*
+
326
Problems and Solutions in Group Theory
denominator is the product of hij. mij = j  i is called the content of the box located in the j t h column of the ith row of the Young pattern [A]. hij is called the hook number of that box in the Young pattern, which is equal to the number of the boxes on its right in the ith row of the Young pattern, plus the number of the boxes below it in the j t h column, and plus one.
(8.23)
where the symbol Yyl is a tableau obtained from the Young pattern [A] by filling ( N mij) into the box located in its ith row and j t h column, and the symbol YLX1 a tableau obtained from the Young pattern [A] by filling is hij into that box. The symbol Y y l means the product of the filled digits in it, so does Yh . [XI
+
The onerow Young pattern [n] denotes the totally symmetric tensor representation. The onecolumn Young pattern [In]denotes the totally antisymmetric tensor representation. The totally antisymmetric tensor space of rank N , denoted by [IN], contains only one standard tensor Young tableau, corresponds to the tensor basis
(8.24)
*
The tensor E is an invariant tensor in the SU(N) transformation, so [l”] denotes the identical representation of SU(N ) . The tensor subspace described by a Young pattern with more than N rows is a null space. The reduction of the direct product of two irreducible representations described by two Young patterns [A] and [p] can be calculated by the LittlewoodRichardson rule. However, the Young patterns with more than N rows contained in the reduction should be removed. The representation denoted by the Young pattern [A] with N rows is equivalent to the representation denoted by the Young pattern [A’] obtained from [A] by removing its first column. contravariant tensor Tal...an rank n with respect to SU(N) transof forms as
d l ...d,
*A
=
C
Tdl...dn u  ~. .) ( u  ’ ) ~~ ~~ ( . ~ ~ , ,,
u E SU(N).
(8.25)
Unitary Groups
327
d l ...d , d l ...d ,
(8.26)
The contravariant tensor space T can also be decomposed by the projection of the Young operator, T = @$T = @$"*. The tensor subspace corresponds to the representation [A]* of SU(N), which is the conjugate one of the representation [A]. The basis in the tensor subis denoted by the standard tensor Young tableau with star, for space example
$ " I '
$"*
(8.27)
*A
as
mixed tensor T$;;t; of rank (n,m) with respect to SU(N) transforms
dl
dn
u c ' ( o ~ T ) = ~ ~ * . ~ a l~l ~' * U a n d n T c l : : : c n ( U  l ) d l b l ~
Cl...Cn
' ' '
(u')dnbn
'
d l ...d n
(8.28) Obviously, the mixed tensor (D): = 6; is an invariant tensor of rank (1,l) in the SU(N) transformation. A mixed tensor is said to be made a contraction (or the trace) of its one pair of indices, one covariant and one contravariant, if to take the pair of indices to be equal and then to sum over the indices. The trace tensor of a mixed tensor of rank (n,m)is a tensor of rank ( n  1,m  1). A mixed tensor can be decomposed into the sum of a series of traceless tensors with different ranks. For example,
} ;[ c l ~ dc :
+dt
d
Ti]7
(8.29)
+D
1
T:]
7
d
where the tensor in the curve brackets is a traceless mixed tensor of rank (1,l), and the tensor in the square brackets is a scalar [the tensor of rank (0, O)]. Therefore, a mixed tensor space can be decomposed into the direct
328
Problems and Solutions in Group Theory
sum of a series of traceless tensor subspaces with different ranks, each of which is invariant in the SU(N) transformation. The traceless tensor space can be decomposed into the direct sum of some minimal tensor subspace by the projection of the Young operator YF1 on the covariant indices and the projection of the Young operator yF1 on the contravariant indices, where the Young patterns [A] and [TI contain n and m boxes, respectively. The minimal tensor subspace corresponds to an irreducible representation of SU(N), denoted by [A]\[T]*. It can be shown that the tensor subspace corresponding to [A]\[T]* is null space if the sum of the numbers of rows of [A] and [TI is larger than N , and the representation [A]\[T]* of S U ( N ) is equivalent to the representation [A']\[T']* if [TI contains t rows, [ T ' ] is obtained from [TI by removing its first column, and [A'] is obtained from [A] by attaching a new column with ( N  t ) boxes from its left. Thus, the representation [TI* of SU(N) is equivalent to the representation [A] if by rotating the Young pattern upside down and then by attaching it to the Young pattern [A] from below, it just forms a rectangle with N row.
[TI
4. Calculate the dimensions of the irreducible representations denoted by
the following Young patterns for the SU(3) group and for the SU(6) group, respectively:
Solution.
34
67
1 345
1 678
Unitary Groups
329
5 . Calculate the ClebschGordan series for the following direct product representations, and compare their dimensions by Eq. (8.23) for the
SU(3) group and for the SU(6) group, respectively:
o o l l l $ o o l l $ o0 0 1 1
Solution. a): O O 8 1 1 1 z
0
0
01
1
001 $01 , 1
SU(3) : 8 x 10 = 80 = 35
+ 27 + 10 + 8, SU(6) : 70 x 56 = 3920 = 1050 + 1134 + 840 + 896.
00011$0001$000 b): 0 0 0 ~ 1 1 1 ~ 0 0 0 1 1 1 $ 1 11 1 1 1’
S U ( 3 ) : 10 x 10 = 100 = 28 + 35
+ 27 + 10, SU(6) : 56 x 56 = 3136 = 462 + 1050 + 1134 + 490.
N
4:
00011 0001 000 000 ooolll$ooo $000 $000, l l ~ @ o o o  o o o 1 11 111 SU(3) : 10 x 10 = 100 = 64 27+ 8 1, SU(6) : 56 x 490 = 27440 = 9240 + 11340 5880
+
+
+
+ 980.
d):
0000@11
N
o o o o l l $ oooo o o l l
r)
00001 @0012
00
2
002
L
00001 00001 00001 $00 $001 $002 2 1 12
330
Problems and Solutions in Group Theory
00001 O0 @ I 2
0000 0000 @0011@001 2 1 2
0000
ool
2
0000 00 @11
2
’
SU(3) : 27 x 8 = 216 = 6 4 + 35 + 35 + 2 x 27+ 10 O + 1 0 + 8 + 0 + 0, SU(6) : 1134 x 70 = 79380 = 9240 + 11550 + 5880 2 x 11340 6000 5670 4704 5880 4536 3240.
+
+
+
+
+ +
+
+
6 . Write all tensor Young tableaux in the tensor subspace
7;(2’11 $’ll‘T = of SU(3), and prove that the standard tensor Young tableaux constitute the complete bases in this subspace.
$’11
Solution. The standard Young tableau
is F
l
. Thus,
If there is a pair of the filled digits to be equal in the tensor Young tableau, from Eq. (8.19) we have
where the left tensor Young tableau in each equality is standard. If all three filled digits in the tensor Young tableau are different, from Eq. (8.19) we
Unitary Groups
331
have
where the left tensor Young tableaux in the first two equalities are standard. The dimension of the subspace 7;[2’11 eight. There are eight standard is tensor Young tableaux. The first six standard tensor Young tableaux are normalized to 6 , but the last two to 4. In addition, the last two standard tensor Young tableaux are not orthogonal to each other.
7.Try to express each nonzero tensor Young tableau for the irreducible
representation [3,1] of SU(3) as the linear combination of the standard tensor Young tableaux.
Solution. For the SU(3) group, the digits filled in the tensor Young tableaux are 1, 2 and 3. The equal digits cannot be filled in the same column. There are three types of nonzero tensor Young tableaux. A tensor Young tableau in the first type contains three equal digits. From Eq. (8.19) we have
When a < b the left tensor Young tableaux is standard, while when a > b the right one is standard. There are 3 x 2 = 6 standard tensor Young tableaux in the first type. A tensor Young tableau in the second type contains two pairs of equal digits. From Eq. (8.19) we have
When a < b, the first tensor Young tableau is standard, while when a > b the fourth one is standard. There are 3 standard tensor Young tableaux in the second type. A tensor Young tableau in the third type contains one pair of equal digits. There are six standard tensor Young tableaux in the third
332
Problems and Solutions an Group Theory
type. The other tensor Young tableaux can be expanded in the standard tensor Young tableaux by Eq. (8.19),
U
The dimension of the irreducible representation [3,1] of SU(3) is 15. There are 15 standard tensor Young tableaux.
8. Write the explicit expansion of each standard tensor Young tableau in the tensor subspace JfP1]7, 5 is the tensor space of rank four for where ' the standard Young tableau of the Young operator
Solution. First, we write the explicit formula for the application of the

Unitary Groups
333
Similar to Problem 7, there are three types of standard tensor Young tableaux. In the first type there are six standard tensor Young tableaux, each of which contains three equal digits.
P= I
[3 11
 I P
= Y2 ’ Oaaba = 6@aaba  2 0 b a a a  2 0 a b a a  2 0 a a a b .
When a < b the left tensor Young tableau is standard, and when a > b the right one is standard. The module square of the standard tensor Young tableaux in the first type is 48. In the second type there are three standard tensor Young tableaux, each of which contains two pairs of equal digits:
The module square of the standard tensor Young tableau in the second type is 24. In the third type there are six standard tensor Young tableaux, each of which contains one pair of equal digits:
334
Problems and Solutions in Group Theory
where a and b are respectively taken 2 and 3. The module square of the first form, which includes two standard tensor Young tableaux, is 24, and that of the remaining standard tensor Young tableaux is 18. Two standard tensor Young tableaux with the equal set of the filled digits are not orthogonal.
9. Transform the following traceless mixed tensor representations of the SU(6) group into the covariant tensor representations, respectively, and calculate their dimensions:
Solution. (1) [3,2,I]* N [i3]\[2,I]* N [23,1]\[1]* 21 [33,2,1],
6 5 4 3 78 67 56 4
678 56
d [ 6 ,5,4,3] (su(6) =
6 5 4 3
7 6 5 4 9 7 5 3
8 9 10 11 78 9 67 5 8 6 4 2 7531 531 31 1
= 147840.
Unitary Groups
335
d[7,6,4,3,1](SU(6))
=
6 5 4 3 2
7 6 5 4
8 9 10 11 12
7 8 9 10 67 5
11
976421 6431 421 1
= 612500.
10. Prove the identity:
are where (TA),, the generators in the selfrepresentation of the SU(N) group.
Solution. Define a mixed tensor T of rank (2,2) with respect to SU(N)
It is easy to show that T is invariant in the SU(N) transformation,
Therefore, T can be expressed as the product of the invariant mixed tensors d of SU(N), :
336
Problems and Solutions in Group Theory
Since ’ is a traceless matrix, 0 = T  N p . AS
C , TE,d = ( N p + q) 6:,
we have q =
when A # N 2  1 when A = N 2  1, we have
N1 2N

Tgg = p + q =  ( N
 1)p.
The solution is p =  ( 2 N )  ’ and q = 1 / 2 . This completes the proof.
8.3
Orthonormal Bases for Irreducible Representations
Recall that the standard tensor Young tableau is the eigenstate of the Chevalley basis H p where the eigenvalue is equal to the number of the digit p filled in the tableau, subtracting the number of the digit ( p + 1 ) . The action of the lowering operator Fp on the standard tensor Young tableau is equal to the sum of all possible tensor Young tableaux, each of which is obtained from the original one by replacing one filled digit p with the digit ( p + 1 ) . Two standard tensor Young tableaux with different sets of the filled digits are orthogonal, but may not be normalized to the same number. Two standard tensor Young tableaux with the equal set of the filled digits but different filling order may not be orthogonal. The merit of the method of Young operators is that the complete tensor bases in an irreducible representation space of SU(N) have been explicitly known. The shortcoming is that the tensor bases are not orthonormal, and the obtained representation is not unitary. On the other hand, the method of the block weight diagram gives the representaion matrices of the Chevalley bases of generators in the unitary irreducible representation, but the orthonormal bases are given only symbolically. To combine two methods will offset one’s shortcoming by the other’s strong point. In the standard tensor Young tableau corresponding to the dominant weight, the number of each digit p filled in the tableau must not be less than the number of the digit (p+ 1)filled there. It is the criterion for determining which standard tensor Young tableau corresponds to the dominant weight in the representation. The states for the multiple weight correspond to the
*
*
Unitary Groups
337
standard tensor Young tableaux with the equal set of the filled digits but in different filling order. In the standard tensor Young tableau corresponding to the highest weight, each box located in the j t h row of the tableau is filled with the digit j , because each raising operator E, eliminates it. Therefore, the component MP of the highest weight M in the representation [A] is
M , = A,  &+I.
(8.30)
Gelfand presented the symbols for the orthonormal basis states of an irreducible representation [A] of SU(N), usually called the Gelfand bases, and calculated the analytic formulas for the representation matrix entries of the generators (the Chevalley bases) of SU(N) in the Gelfand bases. For a multiple weight, he chose the orthonormal basis states first according to the multiplets of the subalgebra d1,then, according to the multiplets of the subalgebra A2 if there still exist the multiple states, and so on. Gelfand described the basis state with the weight m in the representation [A] of SU(N) by N ( N 1)/2 parameters &b, 1 5 a 5 b 5 N , arranged as a regular triangle upside down:
*
+
w1N
W2N
W1(Nl)
lwab)
=
... w ( N  1 ) N ... ... ... ...
w22
(J11
W ( N  1) ( N  1)
w12
where W
~ N =
A,,
WNN
= AN = 0, and
(8.31)
i
The representation entries of the generators are
338
Problems and Solutions an Group Theory
A,,
(Wab)
=
{

P1
(&(@I)
 Wvp
t
+ v  1)
P+l
p=l
n
(Wp(p+l)
 WV,
p
+ ).
t=l
Namely, the eigenvalue of H, is equal to twice the sum of the parameters in the pth row from the bottom, minus the sum of the parameters in the ( p 1)th row and the ( p + l ) t h row from the bottom. The application of E, (F,) to a Gelfand basis leads to a linear combination of the possible basis states, each of which is obtained by increasing (decreasing) one parameter wap in the pth row from the bottom under the condition (8.31). The parameters of the highest weight state in the Gelfand bases satisfy
. (
d#v,d=l
n
,
(Wdp
 Wv,  d
+ v)
(Wd,
 Wv,  d
+ v  1)
r2
(8.32)
} I 2
11. Expand the Gelfand bases in the irreducible representation [3,0]of the SU(3) group with respect to the standard tensor Young tableaux by making use of its block weight diagram given in Fig. 7.2.
Solution. The tensor subspace described by the representation [A] = [3,0] consists of the completely symmetric tensors of rank three. There is no multiple weight in the representation. Its highest weight is M = (3,0), which corresponds to the standard tensor Young tableau and the Gelfand basis are
There are three types of the standard tensor Young tableaux which are normalized to 6,12 and 36,respectively:
where a, b and c are all different. In the application of the lowering operator F, , all possible terms, which may be nonstandard tensor Young tableaux,
Unitary Groups
339
need to be considered. For example
Now, the orthogonal bases can be calculated as follows.
2 0 0 0 ) ) 2
2 0 0 0 ) )
340
Problems and Solutions in Group Theory
The expansions of the orthonormal basis states with respect to the standard tensor Young tableaux give the wave functions of the hadrons. Substituting the expansions into the planar weight diagram of (3,O)given in Fig. 7.3, we obtained the wave functions for the baryon decuplet. In the planar weight diagram, the ordinate axis and the abscissa axis are the eigenvalues of T3 and 2'8, respectively (see the end of Problem 14).
12. Expand the Gelfand bases in the irreducible representation [3,3] of the SU(3) group with respect to the standard tensor Young tableaux by making use of its block weight diagram given in Fig. 7.2.
Solution. The representation [A] = [3,3] of SU(3) is the conjugate representation of [3,0], [3,3] N [3,0]*. This problem can be solved in the same way as that used in Problem 11. However, in the present problem we prefer to calculate the standard tensor Young tableaux of the contravariant tensor instead of those of the covariant tensor. Both results for two kinds of the standard tensor Young tableaux are given here. Note that the action of the generator on the basis state of the conjugate representation is different from that on the basis state of the original representation. Denote by D ( I A ) and D'(IA) the representation matrices of the generator I A in the representations D ( R ) and those in its conjugate one D ( R ) * ,respectively:
D(R) = 1  i
D(R)* = 1
+ix
A
C D(IA)wA,
D(IA)*wA
=1
ix
A
D'(IA)wA.
A
Thus, D ' ( I A )=  D ( I A ) * . For the Chevalley bases we have
Namely, the standard tensor Young tableau with star is the eigenstate of the Chevalley basis H p where the eigenvalue is equal to the number of the digit ( p 1) filled in the tableau, subtracting the number of the digit p. The action of the lowering operator Fp on the standard tensor Young tableau with star is equal to the sum multiplied by 1, of all possible tensor Young tableaux with star, each of which is obtained from the original one by replacing one filled digit ( p 1) with digit p. The order of the filled
+
+
Unitary Groups
341
digits in the standard tensor Young tableau changed to 3, 2, 1 instead of 1, 2) 3. The highest weight of the representation [3,3] is M = (0,3), which corresponds to the standard tensor Young tableau and the Gelfand basis are
1(0,3),(0,3))
=m]*
1 1 1 2 2 2
=wl= l3 :
3 3 O).
The normalization factor for the standard tensor Young tableau with star is similar to that for the standard tensor Young tableau. Now, the orthogonal bases can be calculated as follows.
342
Problems and Solutions in Group Theory
13. Expand the Gelfand bases in the irreducible representation [2,1] of the SU(3) group with respect to the standard tensor Young tableaux by making use of its block weight diagram given in Fig. 7.2.
Solution. The representation [A] = [2,1] is the adjoint representation of SU(3). For the tensor subspace 7;12’11 = yi2’117, standard Young the operator is $’11 = E + (1 2)  (1 3)  (2 1 3), corresponding to the Young
tableau
F l
. There are six single weight and one double weight (0,O)
in the representation [2,1]. The typical standard tensor Young tableaux and their normalization factors are
Due to the Fock condition (6.5), we have
In particle physics, a hadron is composed of three quarks (baryon) or a pair of one quark and one antiquark (meson). The expansions of the orthonormal basis states in [2,1] with respect to the standard tensor Young tableaux give the flavor wave functions for the baryon octet or for the meson octet. For the meson octet, the standard tensor Young tableau should be
Unitary Groups
343
transformed into the traceless mixed tensor by the totally antisymmetric tensor &bc:
(8.35)
The highest weight in [2,1] is M = (1,l), which corresponds to the standard tensor Young tableau and the Gelfand basis are
The remaining orthonormal bases can be calculated from the block weight diagram given in Fig. 7.2:
U
I
U
I
I
344
Problems and Solutions in Group Theory
+V F
dm
* 2
P I =I
2 0o 0 ) , l
where we use the formula calculated from Eq. (8.32):
F2I22:l0)=&I 3
1 2 l l l o )
+Ll
1
2 , l1 O ) . O
Each basis state in the representation [2,1] is normalized to 6 . Especially,
Unitary Groups
345
the basis state
14. Express each Gelfand basis in the irreducible representation [4,0] of
the SU(3) group by the standard tensor Young tableau and calculate the nonvanishing matrix entries for the lowering operators Fp. Draw the block weight diagram and the planar weight diagram for the representation [4, O] of SU(3).
Solution. The tensor subspace described by the representation [A] = [4,0] consists of the completely symmetric tensors of rank four. There are four single dominant weights in the representation:
The highest weight of the representation [4,0] is M = (4,0), which corresponds to the standard tensor Young tableau and the Gelfand basis: )(4,0),(4,0))
=
M
I
=
l4
4
0
").
Obcac
There are four types of the standard tensor Young tableaux which normalize to 48, 96, 144, and 576, respectively. The normalization factors for them are &,2, and I, respectively.
m,
346
Problems and Solutions in Group Theory
where a , b, and c all are different. Now, we calculate the remaining basis states. From the highest weight [A] = [4,0] we have a quintet of d1:
Unitary Groups
347
=
dZ[11212131=
4 3 : 0 0 ) , 1/2)F2
{ dZ pppp1)
=
2ml=/4 ,
O)
=
m= 4 l 1. o
0
O)
It is easy to check that
The block weight diagram and the planar weight diagram are now drawn as follows. From Eqs. (8.3) and (8.6), the simple roots and the fundamental dominant weights of the SU(3) group in the rectangular coordinate frame
348
Problems and Solutions in Group Theory
are
From Eq. (8.4) we know that the first and the second components of the weight m in the rectangular coordinate frame are related to the eigenvalues of T8 = Ti3) and T3 = Ti3) in SU(3), respectively. Namely, the ordinate axis and the abscissa axis are the eigenvalues of T3 and T8, respectively.
15. Express each Gelfand basis in the irreducible representation [3,1] of the SU(3) group by the standard tensor Young tableau and calculate the nonvanishing matrix entries for the lowering operators F’. Draw the block weight diagram and the planar weight diagram for the representation [3,1] of SU(3).
Solution. There are two single dominant weights ( 2 , l ) and (0,2), and one double dominant weight (1,O) in the representation [A] = [3,1] of SU(3):
Unitary Groups
349
,
The highest weight of the representation [3,1]is M = (2,1), which corresponds to the standard tensor Young tableau and the Gelfand basis:
P
= the standard Young tableau is In the tensor subspace 7;[3”1 3/1[3’117, I , and the action of the Young operator on the tensor basis
Note that
where a may be filled in the second column or the third column. There are four types of standard tensor Young tableaux which normalize to 18, 24, 24, and 48, respectively. The normalization factors for them are and 1, respectively.
m,a,a,
J“(J ( 3 G a a a d  O d a a a  O a d a a  e a a d a ) } 1/12
350
Problems and Solutions in Group Theory
From the highest weight state ( 2 , l ) we have a triplet of A1 with (0,2) and (2,3), and a doublet of A2 with ( 3 , i ) .

2 1°), 1
I
From the weight (3,i) we have a quartet of d with ( l , O ) , (i, and 1 l), (3,2), where the dominant weight ( 1 , O ) and its equivalent weight (i, are 1) the double weights. We define one basis state for ( 1 , O ) or for ( i , l ) belongs to the quartet of d1, and the other belongs to the doublet.
Unitary Groups
351
From the weight (0,2) we have a triplet of A2 with (1,O) and (2,2), where the basis state with the weight (1,O) is a combination of 1(1,0)1) and 1(1,0)2). Letting
we have
Thus, a1 = and a2 = &!@ = H to the basis state ((2, 1), (l,O)l), we have 2
m,
m.Applying
E F 2 2
=FE 2 2
+
352
Problems and Solutions in Group Theory
f i o m the weight (2,2) we have a triplet of A with ( 0 , i ) and (2,0), 1 where the weight (0, i)is a double weight. We define one state basis I(0, i),) belongs to the triplet of A1, and the other 1 ( O 7 i ) 2 ) belongs to the singlet.
(l,s), the states with the weights (i, and ( 0 , i ) are the combinawhere 1)
tions of the. basis states. Letting
From the weight (2,3) we have a quartet of A2 with (T, 1))(0,T) and
we have
Unitary Groups
353
Thus, bl = E F =FE 2 2 2 2
m, = m,ci = m,and
b2
+ H2 to the basis state l ( 2 , 1 ) , ( i , l ) 2 ) , we have
c3
=
m. Applying
Choosing the phase of the basis state )(2,1),(0, i ) 2 ) such that c4 is real positive, we obtain c4 = fi, and then c2 = 0. Applying E2F2 = F2E2+ H 2 to the basis state )(2,1),( O , i ) 2 ) , we have
EF 2 2
P,11,( O , i ) 2 ) = dld2 1(2,1), ( 0 , i ) l ) + di; l(2,1), ( O , i ) 2 ) = (F2E2 + H2) ((2,1), ( O , i ) 2 ) = Jz I@, 11,( 0 , i ) l ) + (3  1) 1(2,1), (0,1)2).
Choosing the phase of the basis state l(2, l), such that d2 is real and then dl = 1. From the above calculation, positive, we obtain d2 = we have
a,
(1,s))
l(27 I>, O A 2 ) = r n ( F 2 1 ( 2 , 1 > , ( i , 1 ) 2 )  r n I ( 2 , 1 > ,( O J ) l ) } (
= m 2 F
{ m p !
 J . j ( 2 P l = + p = 1 = 13 1 3 3
;
P I }
0) ,
l(2, 1)) L a ) = l / W 2 1 ( 2 , 1 > , ( O , i ) 2 ) = F21(2,1), ( 0 , i ) l ) (
354
Problems and Solutions in Group Theory
Note that those representation matrix entries of F 2 related to the basis states with the multiple weights can also be calculated with the formula of Gelfand.
From Eq. (8.32) we have
Unitary Groups
355
The block weight diagram and the planar weight diagram are given in the figure. The planar weight diagram of the representation [A,, A,] of SU(3) is a hexagon with the lengths of two neighbored sides to be (A1  A,) (i.e. M I ) and A2 (i.e. M z ) . The weights on the boundary is single, and the multiplicity increases as the position of the weight in the diagram goes inside until the hexagon reduces to a triangle.
16. Calculate the ClebschGordan series for the direct product representation [2,1] x [2,1] of the SU(3) group, and expand the highest weight state of each irreducible representation in the CG series with respect to the standard tensor Young tableaux. Solution. The representation [2,13 is the adjoint representation of SU(3) and was studied in Problem 13. The highest weight of [2,1] of SU(3) is ( 1 , l ) .
356
Problems and Solutions in Group Theory
Usually, the ClebschGordan series for the direct product representation of the SU(N) group is best calculated with the LittlewoodRichardson rule, as used in Problem 5. For [2,1] x [2,1] of SU(3) we have
However, the method of the dominant weight diagram discussed in Chapter 7 is also a good method for it. In this method we need to know the dominant weights and their multiplicities in the relevant representations, which can be obtained, say, by the method of the block weight diagram, or by consulting the table book. They can also be obtained directly from the standard tensor Young tableaux. This is the aim of this problem. Dominant weight Standard tensor Young tableaux No. of states 1 2
X
2
X
6
10
Unitary Groups
357
We have known that the weight for a standard tensor Young tableau is dominant if and only if the number of each digit p filled in the tableau is not less than the number of the digit ( p 1) filled there. We can easily count by this criterion which and how many standard tensor Young tableaux correspond to the dominant weights in the direct product representation space and in the relevant representation spaces. First, we list the standard tensor Young tableaux with the dominant weights in the direct product representation space [2,1] x [2,13, where only one standard tensor Young tableau in the pair related by a permutation of two product factors is listed for simplicity. Then, we calculate the standard tensor Young tableaux with the dominant weights in the representation (2,2) which is contained once in the CG series because the weight (2,2) is the highest weight in the product space.
+
Dominant weight
Standard tensor Young tableaux 111111111
No. of states
1111
12121 By comparing respectively the multiplicities of the dominant weights (3,O) and (0,3) in the product space with those in the representation (2,2), 2 1 = 1,we conclude that there is one representation (3,O) and one representation (0,3) contained in the CG series. Third, we calculate the standard tensor Young tableaux with the dominant weights in the representations (3,O) and (0,3). Comparing the numbers of the standard tensor Young tableaux with the dominant weight (1,l) in the product space and in the three representations (2,2), (3,O) and (0,3), 6  2  1  1 = 2, we conclude that the representation (1,l)is contained in the CG series twice.
358
Problems and Solutions in Group Theory
~~
Dominant weight
Standard tensor Young tableaux in (3,O) in ?0,3)
No. of states in (3,O) in (0,3)
TTTTTq
Fourth, we calculate the tensor Young tableaux with the dominant weights in the representation (1,l). Comparing the numbers of the standard tensor Young tableaux with the dominant weight (0,O) in the product space and in the representations (2,2), (3,0), (0,3), and two ( l , l ) , 10  3  1  1  2 x 2 = 1, we conclude that the representation (0,O) is contained once in the CG series. Therefore, we complete the diagram of the dominant weights for the direct product representation [2,1] x [2, I], where the subscript S or A denotes the symmetric or antisymmetric combination in the permutation of two factors in the product, respectively.
1
Standard tensor Young tableaux
No. of states
0s x n
6x1 3x2 3x2 6x6
1 x 10
R
2
0,o
At last, we calculate the expansion for the highest weight state of each
Unitary Groups
359
representation contained in the CG series by Eq. (7.24), namely, each raising operator E, annihilates the highest weight state. The results are

F F+ l l
x
X X
U U
(antisym. terms),
terms)
U
U
In terms of the orthonormal basis states we have
360
Problems and Solutions in Group Theory
17. A neutron is composed of one u quark and two d quarks. Construct the wave function of a neutron with spin S = 1/2, satisfying the 3 correct permutation symmetry among the identical particles. Solution. In particle physics, a hadron is composed of three quarks (baryon) or a pair of one quark and one antiquark (meson). The strong interaction among quarks is the gauge interaction of the colored SU(3) group. In addition to the color, a quark brings the flavor quantum number. The u and d quarks span the selfrepresentation space of the flavor SU(2) group, the isospin symmetry group. The isospin is conserved in the strong interaction. The next quark is s quark, which is heavier than the u and d quarks. u, d and s quarks span the selfrepresentation space of the flavor SU(3) group, which describes an approximate symmetry, but played an important role in the history of particle physics. Even now, it is still useful in the rough estimation. The expansions of the orthonormal basis states with respect to the standard tensor Young tableaux give the flavor wave functions of the hadrons. The flavor SU(3) group is a compact simple Lie group with order 8 and rank 2. The two commutable generators have important physical meanings. In the selfrepresentation, T3 describes the third component of the isospin of a quark, and T8 is proportional to the supercharge Y of a quark:
The basis states in an irreducible representation space of the flavor SU(3) group can be described in the planar weight diagram with the coordinate axes T3 and Y , as shown in Fig. 7.3 and Problems 14 and 15 in this Chapter. Since the planar weight diagram is intuitive, it has widely applied in particle physics. A neutron consists of three quarks. It is a system of three identical fermions. According to the Fermi statistics, the wave function has to be totally antisymmetric in the permutation of quarks. The internal orbital angular momentum of quarks is assumed to be zero. The total wave function is composed of three parts: the color part, the flavor part, and the spinor part. Since the neutron has to be the color singlet, its wave function
Unitary Groups
361
of the color part is totally antisymmetric in the permutation. Thus, the product of the wave functions of the flavor part and the spinor part should be combined into the totally symmetric state in the permutation. The neutron belongs to a flavor octet with T = T3 = 1/2 and Y = 1, and to a spinor doublet. In this problem we discuss the state with S = 1/2. The 3 Young patterns corresponding to two parts both are [2,1]. Choosing the
standard Young tableau
F,obtain two basis states of the flavor l we
wave function by the standard tensor Young tableaux:
F= udd + dud  2ddu, l
(23)
F= udd + ddu  2dud . l
There are also two basis states of the spinor wave function:
F= (+  ) + ( + ) l
(23)
 2(  +) ,  +)
 2( f
E= (+  ) + (l
) .
In this set of the standard bases, the representation matrices of the generators of the permutation group S3 can be calculated by the tabular method (see Problem 20 in Chap. 6),
In the direct product representation,
362
Problems and Solutions in Group Theory
Their eigenvectors for the eigenvalue 1 respectively
The totally symmetric state corresponds to the common eigenvector with eigenvalue 1, which is ( 2 1 1 2 ) T . Thus, the total wave function of the flavor part and the spinor part for the neutron with S3 =  1 / 2 is
r
= {udd dud  2 d d ~ } 3 {(+  )  (  +)} {udd ddu  2 d ~ d ) 3{(+  )  ())
*
+
+
+
*
+
= 3{2~+ddd+d~d+~d~dd++2dd~+
 dud+
 ud+d
 dd+u
+2d~+d}
.
For normalization, one may replace the factor 3 with
dm.
8.4
Subduced Representations
Now, we discuss the reduction problem of the subduced representation of an irreducible representation of the SU(N) group with respect to its subgroup. There are three kinds of important subgroups for the SU(N) group. One is SU(N 1) cSU(N), the second is SU(M)xSU(N/M) cSU(N), and the third is SU(M)xSU(N  M) cSU(N). The reduction of the subduced representation from an irreducible representation [w] of SU(N) with respect to the subgroup SU(N  1 ) can be solved directly by the standard tensor Young tableau. The tensor indices taken the value N keep invariant in the subgroup SU(N  1). In a standard tensor Young tableau, the box filled with N can only be located in the lowest position of each column, where the neighbored right position has to be empty or a box filled with N . Removing the boxes filled with N in a standard tensor Young tableau of SU(N), we obtain a basis state of an
*
*
Unitary Groups
363
irreducible representation of the subgroup SU(N  1). Thus, by removing several (including zero) boxes located in the lowest positions in some columns from the Young pattern [A], we obtain the Young pattern [p],denoting an irreducible representation of SU(N  1) in the ClebschGordan series for the reduction of the subduced representation:
(8.37)
This method may be applied continuously to the reduction for the subgroups in the group chain SU(N) 3 SU(N  1) 3
. ‘ . 3 SU(3) 3 SU(2) .
Thus, the subduced representation of an irreducible representation [A] of SU(N) with respect to its subgroup SU(N’) with N’< N can be reduced. For the SU(NM) group, containing the subgroup SU(N) x SU(M), each tensor index a consists of two indices (a;),where a is the tensor index of the subgroup SU(N), and i is the tensor index of the subgroup SU(M). In the reduction of a subduced representation from an irreducible representation [w] of SU(NM) with respect to the irreducible representations [A] 8 [p] of the subgroup SU(N)xSU(M), the numbers of the boxes in three Young patterns are the same n. The three Young patterns respectively describe the permutation symmetry among n indices a , n indices a , and n indices a. When n indices i are fixed, the n indices a have the permutation symmetry denoted by [A]. When n indices a are fixed, the n indices i have the permutation symmetry denoted by [ p ] . When both sets of indices are transformed, n indices a! have the permutation symmetry denoted by [w]. Therefore, the representation [w] of the permutation group S, should be contained in the reduction of the inner product of two representations [A] and [ p ] :
*
Now, being an irreducible representation of SU(NM), the number of rows of the Young pattern [w] cannot be larger than N M . Correspondingly, the number of rows of the Young pattern [A] for SU(N) cannot be larger than N, and that of the Young pattern [p] for SU(M) cannot be larger than M .
For the SU(N M ) group, containing the subgroup SU(N)x SU(M), each tensor index a! takes two possible kinds of values, either the tensor in
*
+
364
Problems and Solutions in Group Theory
dex a of the subgroup SU(N),or the tensor index i of the subgroup SU(M). In the reduction of the subduced representation from the irreducible representation [w] of the SU(N IM) group with respect to the irreducible representation [A] 8 [p] of the subgroup SU(N)xSU(M), the number n of 1 boxes in the Young pattern [A] and the number 722 of boxes in the Young pattern [p] are not fixed, but the sum nl + n2 of their numbers of boxes is fixed and equal to the number n of boxes in the Young pattern [w]. In the permutation among the tensor indices, all indices are divided into two sets. The n1 indices a in the first set have the permutation symmetry denoted by [A], which contains n1 boxes, and the 722 indices i in the second set have the permutation symmetry denoted by [ p ] ,which contains 122 boxes. Altogether, they have the permutation symmetry denoted by [w], which contains n = n1 n boxes. Therefore, the representation [w] of the 2 permutation group S, should be contained in the reduction of the outer product of two representations [A] and [p] of the permutation group:
+
+
[A] 8 [p] = a:, [w] (3
*
 
The reduction of the subduced representation can be calculated by the LittlewoodRichardson rule.
Being an irreducible representation of SU(N M ) , the number of rows of the Young pattern [w] cannot be larger than N M . Correspondingly, the number of rows of the Young pattern [A] for SU(N) cannot be larger than N , and that of the Young pattern [p] for SU(M) cannot be larger than M .
+
+
18. Reduce the subduced representation from [3,1] of the SU(4) group with respect to the subgroup SU(3) and list the standard tensor Young tableaux as the basis states in each irreducible representation space of SU(3). Solution. The dimension of the representation [3,1] of SU(4) is
Unitary Groups
365
Its subduced representation with respect to SU(3) is reduced into the direct sum of six irreducible representations of SU(3). They and their basis states are listed as follows. 1). The representation [3,1] of SU(3) is 15dimensional with the following standard tensor Young tableaux as the basis states 111 111 1 1 2 112 113 2 ' 3 ' 2 ' 3 l 2 ' 113 122 122 123 123 3 ' 2 ' 3 l 2 l 3 ' 133 133 222 223 233 2 l 3 ' 3 j 3 ' 3
*
2). The representation [3,0] of SU(3) is 10dimensional with the following standard tensor Young tableaux as the basis states
111 1 1 2 113 122 123 ' 4 ' 4 ' 4 ' 4 ' 4 133 222 223 233 333 4 ' 4 l 4 l 4 ' 4 3). The representation [2,1] of SU(3) is 8dimensional with the following standard tensor Young tableaux as the basis states 114 2 ' 134 2 l 114 3 l 134 3 l 124 124 2 ' 3 224 234 3 l 3
*
4). The representation [2,0] of SU(3) is 6dimensional with the following standard tensor Young tableaux as the basis states 114 4 ' 124 4 ' 134 4 ' 224 4 ' 234 4 ' 334 4

5 ) . The representation [1,1]of SU(3) is 3dimensional with the following standard tensor Young tableaux as the basis states
144 2 ' 144 3 ' 244 3
*
6). The representation [l,0 of SU(3) is 3dimensional with the following 1
366
Problems and Solutions in Group Theory
standard tensor Young tableaux as the basis states
144 4 )
244 4 '
344 4

19. Reduce the subduced representations from the following irreducible representations of the SU(6) group with respect to the subgroup SU(3)@SU(2),respectively:
Solution. In particle physics, a representation of the flavor SU(3) group describes a flavor multiplet, and a representation of the spinor SU(2) group describes a spinor multiplet. All those representations are denoted by the state numbers contained in the multiplets.
6
E
(3, 2)
15 N (3*, 3) @ (6, 1)
21 21 (3*, 1) @ (6, 3)
56 N (10, 4) @ (8, 2)
Unitary Groups
367
E P
70 "
n
20
21
35 N Notice the reductions of the subduced representations from [3]and [2, 14] 2 [1]\[1]*. former gives the observed multiplets of the lowenergy baryons The in the sixties of the 20th century, namely, a decuplet with spin 3/2 and an octet with spin 1/2. Since they belong to the same representation of SU(6), their parities have to be the same (to be positive relative to the parity of a proton). The latter gives the observed multiplets of the lowenergy mesons in that years, namely, a singlet with spin 1, an octet with spin 0, and an octet with spin 1, whose parities are the same (to be negative from experiments). It is the success of the SU(6) symmetry theory for hadrons. In the SU(6) symmetry theory, the spin is described by the SU(2) group, so that this theory is only a nonrelativitic theory. This theory cannot explain many phenomena in high energy particle physics.
2 0 . Reduce the subduced representations from the following irreducible
representations of the SU(5) group with respect to the subgroup SU(3) @SU(2), respectively:
368
Problems and Solutions in Group Theory
Solution. In the grand unified model proposed in 1974, one needs to study the reduction of the subduced representations from some irreducible representations of the SU(5) group with respect to the subgroup SU(3)@SU(2). There is a U(1) subgroup in the SU(5) group with the generator Y =diag{1/3,  1/3,  1/3, 1/2, 1/2}, which is commutable with all elements in the subgroup SU(3)@SU(2). However, since there are five common elements in two subgroups in addition to the identity, the SU(5) group is not the direct product of two subgroups SU(3)@SU(2) and U(1). The color subgroup SU(3) describes the strong interaction of quarks, and the subgroup SU(2)@U( describes the weakelectric interaction. Each 1) representation in the reduction is denoted by two multiplets of SU(3) and SU(2) and by the eigenvalue of Y as the subscript. In the SU(5) grand unified model, the electric charge Q is equal to T 3 Y , where T 3 is the third generator of the subgroup SU(2):
+
Q=
7'3
+ Y = diag (0, 0, 0, 1/2,


1/2}
+ diag{1/3,
= diag{1/3,
 1/3,  1/3, 1/2, 1/2} 1/3,
 1/3,
1, 0} .
Unitary Groups
369
In the SU(5) grand unified model, the state of a particle is divided into the lefthand state and the righthand state. They are filled into the 5dimensional and 10dimensional representations, respectively. The first generation of particles are filled as follows:
where the superscript c denotes the charge conjugate state, and the subscript R and L denote the righthand state and lefthand state, respectively. The 10dimensional represent ation is the antisymmetric tensor represent ation, so only the upper half matrix is needed to be filled with the particle states. The 24dimensional representation is the adjoint representation, describing the gauge particles, which is not listed here.
8.5
Casimir Invariants of S U ( N )
The generators TA in the selfrepresentation of SU(N) satisfy the commutation relations and the anticornmutation relationship
*
370
Problems and Solutions in Group Theory
Due to the normalization condition, ” ~ ( T A T B )6,4B/2, the coefficients = c A iand d A B D can be expressed by T as given in Eq. (8.9) ’
Recall that the adjoint representation Dadj(u)of SU(N) satisfy
N21
D [ ’ ~ ( u )  ~ I ~ ~ D [ =~ ( u ) D i 2 1 ( u ) I $ 1 , u E SU(N), ’
A‘=l
C
(8.38)
where D[’] ( u )and are the representation matrices of the element u and the generator I A in the representation [A] of SU(N), respectively. When [A] is the selfrepresentation [l]of SU(N), D[’](u) = u and I?] = TA. Let CA,” and ~ A B D the totally symmetric and antisymmetric tensors of rank be three with respect to the adjoint representation. We are going to show that they are the invariant tensors. Prove it with CA; as example. CA,D
It1
+
C
A’B’D‘
Di$iD$$iD$DiCAiD,:
= 2in
= CA!.
{ U  ~ T A U U  ~ T B U U  ~ T D Uu  ~ T ~ u u  ~ T ~ u u 
(8.39) According to the LittlewoodRichardson rule, except for N = 2, the CG series for the direct product of two adjoint representations of SU(N) contains, in addition to the other representations, only one identical representation and two adjoint representations. Therefore, the identical representation only appears twice in the reduction of the direct product of three adjoint are representations of SU(N). Equation (8.39) says that CA: and ~ A B D the only invariant tensors of rank three, up to a constant factor, with respect to the adjoint representation of SU(N). For the SU(2) group, since there is only one adjoint representation contained in the CG series for the direct product of two adjoint representations, and then, there is only one identical representation contained in the CG series for the direct product of
Unitary Groups
371
three adjoint representations. On the other hand, d ~ = g for~SU(2), and 0 the invariant tensor of rank three with respect to the adjoint representation of SU(2) has to be proportional to CA,D. Now, we replace TA on the right hand side of Eq. (8.9) with They are still the symmetric or antisymmetric invariant tensors of rank three with respect to the adjoint representation. Thus, they are proportional to cA[or dAgg, respectively
It1.
From the first equality we obtain
Tz([X]) and A([X]) are respectively related to the Casimir operators of rank two and rank three of SU(N), and sometimes are simply called the Casimir invariants of rank two and rank three in the literature. Both Tz([X]) and A( [A]) are independent of the similarity transformation of the representation. It is easy to show from Eq. (8.40) that
(8.42)
In the calculation of T ! ( [ X ]with Eq. (8.41), we take A = D = 3 such ) that both generators contained in Eq. (8.41) belong to the subgroup SU(2). Denote T2([A]) in SU(2) by T2(')([X]), which is calculated to be
T 2 ( O ) ( [ X ] ) = Tr
{
=
1 12 X(X + 1)(X + 2) .
(8.43)
Calculate T2( [A]) in the subduced representation from the irreducible representation [A] of SU(N) with respect to the subgroup SU(N  2)@SU(2), which is equivalent to a direct sum of some irreducible representation [v] @ [p] of the subgroup, I (8.44)
372
Problems and Solutions in Group Theory
In the calculation of A([X]) with Eq. (8.40), we take A = B = 3 and D = 8 such that three generators are contained in Eq. (8.40) all belong to the subgroup SU(3), where d338 = Denote A([X]) in SU(3) by
m.
d o"XI), )
In the planar weight diagram, the states located in the same horizontal level have the same eigenvalues of Tix1,and the sum of (Tix1)2can be calculated by Eq. (8.43). Then, we sum up the contribution for the states For with different TiX1. a onerow Young pattern, we have
A(')([A,O]) = 2
= (1/6)
x C
mO =
C
mO =
x
(A  3m)Tz("'([A m ] )
(A  3m)(X  m)(X  m
+ 1)(X  m + 2)
(8.45)
= (1/120)X(A
+ 1)(X + 2)(X + 3)(2X + 3).

Generally,
A(O)([A + A', A']) = A(O)([A]CZI [A']*)  A(O)([A  1 CZI [A' 1
= dpll ( s W ) ) A ( O )
1
1]*)
([XI)

4 x 1(sw~))A(O)([X'I)
 ~ ~ ~ ,   ~ I ( S U ( ~ ) ) A ( '3 ( [ X ~ ~  ~ I ( S U ( ~ ) ) A ( 11) ( [ X '  1) ~ )  O)
(8.46) Calculate A( [A]) in the subduced representation from the irreducible representation [A] of SU(N) with respect to the subgroup SU(N  3)@SU(3), which is equivalent to a direct sum of some irreducible representation [v]@ [p] of the subgroup,
=  (A 120
+ 1)(A' + 1)(A

+ A')(A + A' + 2)(A + 2X' + 3)(2A + A' + 3).
21. Calculate the Casimir invariants T2([1']) and A([1']) in the representation denoted by a onecolumn Young pattern ([l'])SU(N) of the group.
Unitary Groups
373
Solution. From Eqs. (8.44) and (8.47), we obtain
1 N2 Tz([l‘]) = d[lr~l(SU(N 2)>T4O’([1])  r  1 =2
(
)’
A([lr]) = {d[,r1](SU(N 3))  d[lp21(SU(N 3))) A‘O’([1]) (N  3)! (N  3)! (r  2)!(N  r  l)! (r  l)!(N  r  2)!  (N  3)! {(N  r  1)  (T  1)) (r  l)!(N  r  l)!
22. Calculate the Casimir invariants T2([3])and A([3]) in the irreducible
representation [3] of the SU(6) group.
Solution. From Eq. (8.43) we have
T4O’([3])= 5,
Ti0’([2]) = 2,
TiO’([l]) 1/2 =
TiO’([O]) = 0.
From the reduction of the subduced representation from [3] of SU(6) with respect to the subgroup SU(4)@SU(2)
we obtain
From Eq. (8.45) we have
A(’)([3]) = 27, A(’)([2]) = 7, A(’)([l]) = 1
A(’)([O]) 0. =
From the reduction of the subduced representation from [3] of SU(6) with respect to the subgroup SU(3)@SU(3)
374
Problems and Solutions in Group Theory
we obtain
23. Calculate the Casimir invariants T2([X]) and A([X]) in the representation ([A] = [A, O,O, 01) of the SU(5) group.
Solution. The calculation method is the same as Problem 22. Here we only list the result.
+ 1)(X + 2)/12, A(O)([X]) = X(X + 1)(X + 2)(X + 3)(2X + 3)/120, d[A](SU(3)) = (A + 2)(X + 1)/2, d[A](SU(2)) = + 1.
T2(O)([X]) = X(X
Hence
Generalizing this result to the SU(N) group, we obtain
N
n=O
 X(X+l).(X+N) 2 ( N + l)!
N
)
X(X
+ 1)    (A + N ) ( U + N ) ( N + 2)!
Chapter 9
REAL ORTHOGONAL GROUPS
9.1
Tensor Representations of S O ( N )
The set of all Ndimensional unimodular real orthogonal matrices, in the multiplication rule of matrices, constitutes S O ( N ) group. In this chapter we discuss only the case of N 2 3, because S O ( 2 ) group is an Abelian group which is isomorphic onto the U(1) group. The S O ( N ) group is a doublyconnected compact Lie group with order N(N  1 ) / 2 . S O ( N ) is a subgroup of SU(N). The generators in the selfrepresentation of S O ( N ) is usually taken as the second type of generators in the selfrepresentation of SU(N) multiplied by two
*
(Tab),, = 2 (TLt))
rs
= idarSbs
 26adSbc,
+ ibasdbr,
a
< b,
(94
T ( T a b T c d ) = 2SacSbd r
where the indices a and b are the ordinal indices of the generator, while T and s are the row and column indices of the matrix. N(N  1 ) / 2 matrices Tab all are antisymmetric, and they are also antisymmetric with respect to the ordinal indices. Any matrix R in S O ( N ) can be expressed in the form of exponential matrix function: R=exp
{ iC
a<b
WabTab
1
.
(9.2)
j , The S O ( N ) group with N = 2 l or N = 2 l + 1 contains l mutually commutable generators,
375
376
Problems and Solutions in Group Theory
which span the Cartan subalgebra of the Lie algebra of S O ( N ) . Thus, the ranks of both SO(2t) and S0(2C+ 1) are C. We want to find the common eigenvectors of H j , [ H j , Ea] = ajEa. When N = 2 4 the eigenvectors are
1 E,,2 )  5 [ T ( 2 a ) ( 2 b  l ) + i T ( 2 a  1 ) ( 2 b  l ) (
(3) Eab
+ i T ( 2 a ) ( 2 b ) T ( 2 a  1 ) ( 2 b ) ]
)
=
5 [T(2a)(2b1)
 iT(2a1)(2bl)
+ i T ( 2 a ) ( 2 b )+ T ( 2 a  l ) ( 2 b ) ]
(9.4)
7
with the eigenvalues { e ,  e b } j , { e, eb}j,{ e , e b } j ,and { e,  e b } j , respectively, where a < b, and { e , } j = 6,j. When N = 2 t + 1, in addition to Eq. (9.4), there are another two types of eigenvectors:
+
+
=
8
[T(2,)(2l+l)
+i~(2al)(2C+l~l
3
(9.5)
with the eigenvalues { e , } j and  { e , } j , respectively. The simple roots for the S 0 ( 2 l + 1) group are
The remaining positive roots are
where a < b. The largest root, which is the highest weight of the adjoint representation of SO(2C + l),is
v=2
Therefore, the Lie algebra of S0(2C+ 1) is Bl,and L3 = w2. The simple roots for the SO(2t) group are
The remaining positive root are
Real Orthogonal Groups
377
where a < b. The largest root, which is the highest weight of the adjoint representation of S 0 ( 2 l ) , is
Therefore, the Lie algebra of S O ( 2 l ) is De,and c3 = w 2 . The Chevalley bases in the selfrepresentation of the S 0 ( 2 l + 1) group (the Be Lie algebra) are
*
Hp = Ep =
Fp = He = Ee =
Fe =
q2e)(2e+l)
+ iq2el)(2e+l),
(9.10) where 1 5 p < l. The Chevalley bases in the selfrepresentation of the SO(2l) group (the DCLie algebra) are the same as those of the S 0 ( 2 l + 1) group except for those when p = l , which are
He =
=
T(2[3)(2&2)
+ 7'(2ei)(2e),
7
Fe =
$ { T(212)(2el)  iT(213)(2ll) + iT(212)(21) + T(2e3)(21!)} $ {~(2e2)(2e1) + i ~ ( 2 e  3 ) p e  i )  i~(2[2)(2e) + T(ze3)pe)) 
The definition for a tensor with respect to the S O ( N ) group is the same as that to the SU(N) group, except that the transformation matrix u ESU(N) is replaced with R & O ( N ) . The Weyl reciprocity still holds for the tensors of S O ( N ) . However, this replacement makes some new characteristics for the tensors of S O ( N ) . Since the transformation matrix R E S O ( N ) is real, there is no difference between the covariant tensor and the contravariant tensor for the S O ( N ) group. As a result, first, the tensor space of S O ( N ) group can be decomposed into a direct sum of a series of the traceless tensor subspaces 7 ,which are invariant in S O ( N ) . Each 7can be projected by the Young operators into the subspace = $?IT.
*
(9.11)
378
Problems and Solutions in Group Theory
Second, since the tensor is traceless, the tensor subspace is a null space if the sum of the numbers of boxes in the first and the second columns of the Young pattern [A] is larger than N. Third, since there is no difference between the covariant tensor and the contravariant tensor, the representation described by a Young pattern [A] with the row number m larger than N / 2 is equivalent to that by [A'], where [A'] is obtained from [A] by replacing its first column with a column containing ( N  m) boxes. Fourth, when the row number of [A] is equal to N / 2 , the representation space '7$^] is reduced into two minimal invariant subspaces with the same dimension. They are the selfdual and the antiselfdual tensor subspaces, whose representations are denoted by [+A] and [A], respectively.
A tensor of rank two for S O ( N ) is easy to be decomposed into the sum of the traceless tensors:
*
where the tensor in the curve bracket is a traceless tensor of rank two, and the tensor in the square bracket is a tensor of rank zero. However, generally speaking, the decomposition of a tensor of rank n into a sum of some traceless tensors of different ranks is straightforward, but tedious. As discussed in Chap. 8, the irreducible tensor bases for SU(N) are obtained explicitly. In fact, they are the standard tensor Young tableaux. The bases can be orthonormalized by making use of the block weight diagram. However, due to the traceless condition, some standard tensor Young tableaux for S O ( N ) become linearly dependent. It becomes a new problem how to find the complete set of the independent irreducible tensor bases for S O ( N ) explicitly. Another problem is that the tensor basis Oal...,, is not the common eigenstate of the Chevalley bases H,. Since a tensor basis is a direct product of n vector bases O,, ..O,,, we only need to find the new vector bases such that H p are diagonal. Namely, we want to diagonalize H p in the selfrepresentation of S O ( N ) by a similarity transformation. Both problems are well solved for the SO(3) group. Now, we generalize the solution to the S O ( N ) group. In the selfrepresentation space of SO(2C 1) we defined a set of new orthonormal bases, called the spherical harmonic bases,
*
+
Real Orthogonal Groups
379
when c t = l + l , when ! 2 5 a 5 2 l + 1. ( @ 4 ~  2 , + 3  i@4C2a+4) (9.12) The tensor basis @ , l . . . ( y , of rank n for S 0 ( 2 l + 1) is the direct product , of n vector bases a    @ a n . In the spherical harmonic bases a the , nonvanishing matrix entries of the Chevalley bases are
@a
=
i
(l)ea+l
+
i02,)
when 1 5 a
5 l,
@2e+1
+
(9.13)
where 1 5 p 5 l  1. In the spherical harmonic bases, the standard tensor Young tableau $l+,,...(y,, in the representation [A] of S0(2!+ 1) corresponds to the highest weight if each box in its j t h row is filled with the digit j , because each raising operator E p annihilates it. Therefore, the relation between the highest weight M and the Young pattern [A] for S 0 ( 2 l + 1) is
Although the standard tensor Young tableaux is generally not traceless, the standard tensor Young tableau with the highest weight in [A] of S 0 ( 2 l + 1) must be traceless, because it only contains with a < l + 1. For example, is traceless. The remaining the tensor basis 0 1 0 1 is not traceless, but traceless tensor bases in an irreducible representation space [A] of S 0 ( 2 l + 1) can be calculated from the basis with the highest weight by the lowering operators Fp. In the calculation the block weight diagram is helpful. In the selfrepresentation space of SO(21) we defined a set of new orthonormal bases, called the spherical harmonic bases,
*
380
Problems and Solutions an Group Theory
The tensor basis @)(Y1...CYn rank n for SO(2t) is the direct product of n of vector bases aal    a . In the spherical harmonic bases a,, the nonvan, ishing matrix entries of the Chevalley bases are
(9.16)
where 1 5 p 5 C  1. In the spherical harmonic bases, the standard tensor [XI Young tableau yp @al...Qn in the representation [A] of SO(2t) corresponds to the highest weight if each box in its j t h row, j < t, is filled with the digit j and if all boxes located in its t t h row, if [A] contains k' row, are filled with the same digit l or k ' + 1, depending on whether the representation is selfdual [+A] or antiselfdual [A]. Therefore, the relation between the highest weight M and the Young pattern [A] for SO(2t) is
Since the standard tensor Young tableau with the highest weight contains with a < C, it is a traceless tensor. either @e or @!+I, in addition to The remaining traceless tensor bases in an irreducible representation space [A] of SO(2l) can be calculated from the basis with the highest weight by the lowering operators Fp. The dimension d[xl(SO(N)) the representation [A] of S O ( N ) can be of calculated by the hook rule. In this rule the dimension is expressed as a quotient, where the numerator and the denominator are denoted by the
*
Real Orthogonal Groups
381
symbols Y g l and YL”], respectively:
Let us explain the meaning of two symbols YL”]and Y P ] . The hook path (i, j) in the Young pattern [A] is defined to be a path which enters the Young pattern at the rightist of the ith row, goes leftwards in the i row, turns downward at the j column, goes downwards in the j column, and leaves from the Young pattern at the bottom of the j column. The inverse path (i, j) is the same path as the hook path (i, j) but with the opposite direction. The number of boxes contained in the path (i, j ) , as well as in its inverse, is the hook number hij. YL”’ is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the hook number hij. Define a series of the tableaux Y g l recursively by the rule given below. Y p l is a tableau of the Young pattern [A] where each box is filled with the sum of the digits which are respectively filled in the same box of each tableau Y g l in the series. The symbol Y g l means the product of the filled digits in it, so does the symbol YL”]. The tableaux Y g l are defined by the following rule: (a) Y g l is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the digit ( N j  i).
+
(b) Let [A(1)] = [A]. Beginning with [A(1)], we define recursively the Young pattern [A(”)] by removing the first row and the first column of the Young pattern until [A(”)] contains less than two columns. (c) If [A(”)] contains more than one column, define Y g l to be a tableau of the Young pattern [A] where the boxes in the first ( a  1) row and in the first ( a  1) column are filled with 0, and the remaining part of the Young pattern is nothing but [A(”)]. Let [A(“)] have r rows. Fill the first r boxes along the hook path (1, 1) of the Young pattern [A(”)], beginning with the box on the rightmost, with the
382
Problems and Solutions in Group Theory
digits (Xi"'  l ) , (A?)  l ) , ", (A?)  l),box by box, and fill the first (Xi"'  1) boxes in each inverse path (i, 1) of the Young pattern [A(")], 1 5 i 5 T , with 1. The remaining boxes are filled with 0. If a few 1 are filled in the same box, the digits are summed. The sum of all filled digits in the pattern Ygl is zero.
1. Calculate the dimensions of the irreducible representations of the SO(8) group denoted by the following Young patterns: ( a ) [4,2], (b) [3,2], (c) ~ 4 ~ 4 ( d ) [+3,2,1,11, (el [+3,3,1,11. 1,
Solution. (a).
d[4,2~(S0(8)) =
891011 78
1  1 1 3
+
2
0

781114 58 5421 2 1
5421 2 1 71012
= 4312.
8910
.
112
2 1
11 891011 78 910+21
d[4,4](S0(8))
2 1 3
1
(c).
3 0 0 0 0'01 1
0 7 8 1 3 14 2  5 6 7 12
=
= 8918.
5432 4321 0 1 2 8 9 10 00 78 6 '1 81012 78 5

5432 4321
4 1 2x 2 1
41 2x 2 1 0 2 2 0 0 0
8 9 10
789 6 5
0 0 0 0  1 1
(el.
d[+3,3,1,1](S0(6))
=
+  2 2
+ o
0
632
81112 7 7 1 0 4  3 6 3 2 = 4312. 2 x 521 2 1
521 2x 2 1
2. Calculate the ClebschGordan series for the subduced representations
of the following irreducible representations of the SU(N) group with respect to the subgroup S O ( N ) , and then check the results by their dimensions for N = 7:
Real Orthogonal Groups
383
Solution. The following results for the reductions of the subduced representations of SU(N) with respect to the subgroup S O ( N ) are independent of N , except for the Young patterns in the CG series with the row number not less than N/2, or with the digit number in the first two column larger than N.
210 = 189
+ 21.
+ 77 + 105 + 7. 1008 = 819
+P E
@
384
Problems and Solutions an Group Theory
882 = 693
+ 77 + 105 + 7.
'3
756 = 616
+ 105 + 35.
490 = 378
+ 105 + 7.
t924 = 714+ 182
224 = 189
+ 35.
+ 27+
1.
p+p@r
2646 = 1911
+ 182 + 168 + 330 + 2 x 27 + 1.
2100 = 1560
+ 189 + 330 + 21.
1176 = 825
+ 330 + 21.
Real Orthogonal Groups
385
2352 = 1617
+ 189 + 330 + 168 + 27 + 21.
V@F@P
840 = 616
+ 189 + 35.
490 = 294+ 168+ 27+ 1.
= 378
+ 189 + 21.
140 = 105 t.35.
3. Calculate the ClebschGordan series in the reductions of the following direct products of the irreducible tensor representations, and check the results by their dimensions for the SO(7) group:
(1)
Solution. (1):
PI €3 PI,
(2)
PI €3 [I,11,
(3) [31 €3 [2,11*
27 x 27 = 729 = 182
+ 330 + 168 + 27 + 21 + 1.
m
27 x 21 = 567 = 330 + 189
+ 27 + 21.
386
Problems and Solutions in Group Theory
E F
77 x 105 = 8085 = 1750 1911 1560 1617 182 330 168 330 189 27 21.
+
+
+
+ + + + + + +
4. Calculate the orthonormal bases in the irreducible representation space
[2,2] of SO(5) by the method of the block weight diagram, and then express the orthonormal bases by the standard tensor Young tableaux in the traceless tensor space of rank four for SO(5).
Solution. The Lie algebra of SO(5) is B2. The relations between the simple roots rp and the fundamental dominant weights wp are
rl = 2wl  2w2,
r2
= wl
+ 2w2,
The highest weight of the representation [2,2] of SO(5) is (0,4), and its dimension is
56
11
67
21
21
From the highest weight state (0,4) we have a quartet of A with (1,2), 2 (2,0), (3,g) and (4, where the matrix entries of F 2 are 2, &,&,and 2, respectively. The weights (1,2) and (2,O) are the single dominant weights. From (1,2) we have a doublet of A with (T, 4). From the weight (2,O) we 1 have a triplet of d with (0,2) and (z,4), and from the weight (T,4) we have 1 a quintet of A2 with (0,2), (l,O), (2,2) and (3,a). The weight (0,2) may
z),
Real Orthogonal Groups
387
be a double dominant weight. Define the first basis state I(0,2)1) belonging to the triplet of A1 and the other I(0,2)2) belonging to a singlet of CAI. Let
Applying El F 2 = F2El to the basis state
l i 4))) we have (,
Thus, a1 = &, and then, a2 = d m = 1. From the weight (3,2) we have a quartet of d with (1,0), (i, and (3,4).On the other hand, applying 1 2)) F to two states )(0,2)1) and 1(0,2)2),we obtain another two states with 2 the weight (1,O). Thus, the dominant weight (1,O) may be a triple weight. We define the first basis state (1, O), belonging to the quartet of dl and the other two belonging to two doublets of d1. The action of F 2 on 1(0,2)1) is the combination of I ( 1 , O ) l ) and I(l,0)2). The weight (2,2) is the double weight because it is equivalent to (0,2). The weight ( i , 2 ) is equivalent to (1,O). We have
Let
Applying
EF 1 2
= F2El to the basis state )(0,2)1) and 1(0,2)2), we have
388
Problems and Solutions an Group Theory
Choosing the phases of the basis states 1(1,0)2) and 1(1,O)3) such that a4 and a7 are real positive, we have a4 = 1, a6 = &, and a7 = 0. It means that the dominant weight (1,O) as well as its equivalent weight (T,2) in [2,2] are the double weights, not triple. The states 1(1,O)3) and I(T, 2)3) do not exist in [2,2]. From the weight (4, q) we have a quintet of 41 with 1 the weights (2,2)1, (0,0)1, (2,2)1 and (3,4),where the matrix entries of F are 2, &, & and 2, respectively. Another basis state 1(2,2)2) belongs to a triplet of 41 with (0, O), and (2,2)2, where both matrix entries of F are 1 Let
Jz.
Applying
EF 1 2
= F2El to the basis state I(l,O)1) and 1(1,0)2), we have
Thus, a8 = 4 and a10 = 0. Applying state I ( l , O ) 1 ) , we have
EF 2 2
=
FE 2 2
+ H2 to the basis
Choosing the phase of the basis state )(2,2)2) such that we have a9 = 1 and all = 2. Applying E2F2 = F2E2 state 1(2,2)1), we have
+ H2 to the basis
a9
is real positive,
Real Orthogonal Groups
389
Choosing the phase of the basis state 1(3,;?>) such that a12 is real positive, we have a12 = 1 and a13 = 4. Applying F2 to the states with the double weight (T,2), we have two basis states with the weight (O,O), in addition to the state bases I(0,O)l) in the quintet and I(O,O)2) in the triplet of 41. Thus, the dominant weight (0,O) may be a quadruple weight. Both I ( O , O ) 3 ) and I(0, O),) are singlet of A1. Let
2)1) 2)2) Applying ElF2 = F2E1 to the basis states 1(2, 4))) I(i, and [(i) , we have
EF 2 2
Thus, we have bl = b2 = b3 = b4 = b7 = fi and b6 = 0. Applying = F2E2 H to the state basis [(i, we have 2 2)1),
+
Thus, b5 = 0, and the basis state 1(0,0)4) does not exist. So, b g = 0. Applying E2F2 = F2E2 H to the basis state l(i, 2 ) , we have 2 2)
+
Choosing the phase of the basis state 1(0,0)3) such that b8 is real positive, we have bs = fi. The remaining basis states and the matrix entries of the lowering operators Fp can be calculated similarly. The results are given in Fig. 9.1.
390
Problems and Solutions in Group Theory
Fig. 9.1 The block weight diagram for [2,2] of SO(5).
Now, we expand the orthonormal basis states in the standard tensor Young tableaux. The Young tableau condition (8.19) we have
yi2’21 is
F l
. F’rom the Fock
The normalization factors for the typical tensor Young tableaux are
Real Orthogonal Groups
391
From Problem 2 , we know that the number of linearly independent trace is tensors in the tensor space of rank four projected by yPy2] 50  35 = 15. In fact, they belong to the representation [2,0] and [O,O] of SO(5). We can calculate the trace tensors from the highest weight condition (7.24) and in terms of the lowering operators. In the following we only list the trace tensors with the dominant weights.
392
Problems and Solutions in Group Theory
+ 2 p 2 q5 +
2
Beginning with the highest weight state, we calculate the expansions of the basis states in terms of Eq. (9.13) and the block weight diagram. Due to the WignerEckart theorem, the state basis belonging to the representation [2,2] of SO(5) is orthogonal to the basis states belonging to [2, O] and [O,O], so it is a traceless tensor. For simplicity we neglect the index (0,4) for the highest weight in the basis state I(0,4), (ml ,ma)).
Real Orthogonal Groups
393
’
394
Problems and Solutions in Group Theory
5 . Calculate the orthonormal bases in the irreducible representation space
[2,0,0,0] of SO(8) by the method of the block weight diagram, and then, express the orthonormal bases by the standard tensor Young tableaux in the traceless symmetric tensor space of rank two for SO(8).
Solution. The Lie algebra of SO(8) is D4. The relations between the simple roots rp and the fundamental dominant weights w p are
rl = 2 w l  w2,
12
= w1+ 2w2  W Q  w4,
The highest weight of the representation [2,0,0,0] of SO(8) is (2,0,0,0), and its dimension is
From the highest weight state (2,0,0,0) we have a triplet of A1 with
Real Orthogonal Groups
395
(0, l , O , 0), and (2,2,0,0), where both matrix entries of F1 are The weight (0, l , O , 0) is a single dominant weight. The weights equivalent to two dominant weights are
a.
Therefore, the representation [2,0,0,0] contains two simple dominant weights (2,0,0,0) and (0, 1,0,O) and one triple dominant weight (O,O, 0,O). The block weight diagram for the representation [2,0,0,0] is very easy except for the part related to the triple weight (O,O,O,O), which will be explained below. In Fig. 9.2 we give the block weight diagram for the representation [2,0,0,0] where only those matrix entries which are not equal to 1 are indicated. 1(0,0,0,0)1)and I(?, l , O , O ) ) constitute a triplet of A1, Let 1(2,%,0,0)), and the other two basis states I(O,O, 0,0)2)and I(O,O, O , O ) 3 ) are the singlets of C A I . We assume
where we neglect the index (2,0,0,0) for the highest weight in the basis state I(2,0,0,0), ( m l ,m2, m3, m4)) for simplicity. Applying El F = F2E1 2 to the basis state I (i, i, 2, i)), F3 = F3El to the basis state I(0, i, 0)), El 2, and ElF4 = F4E1 to the basis state I ( O , i , O , Z ) ) , we have
E1F2 ( ( i , Z , i , 1)) J Z C l ((2, i, = 0,O)) = F ~ ((i, , i ) ) = F~ ((I, E ~ ,i 2 i,i,i)) ](2,i,o,o)), =
396
Problems and Solutions a Group Theory n
Thus, c1 = d / and a1 = bl = 0. Choosing the phase of the basis state 12 I ( O , O , O , O ) 2 ) such that a2 is real positive, we have a2 = Applying E3F2 = F2E3 to the basis state I(i, 2 , i , i ) ) , and E3F4 = F4E3 to the basis state I(0, T, 0,2)), we have
a.
Thus, c2 = and b2 = 4. Then, b3 = 0. Choosing the phase of the basis state 1(0,0,0,0)3) such that c3 is real positive, we have c3 = 1. Now, we expand the orthonormal basis states in the standard tensor Young tableaux. The Young tableau yi2’o’0’01 is . The normalization factors for two typical tensor Young tableaux are
m l
There is only one trace tensor in the tensor space of rank two projected by
y/2~0~0~01 to the representation [O,O, O,O] of SO(8): . It belongs
Beginning with the highest weight state, we calculate the expansions of the basis states in terms of Eq. (9.16) and the block weight diagram. Due to the WignerEckart theorem, the basis state belonging to the representation [2,0,0,0] of SO(8) is orthogonal to the basis state belonging to [O,O, O,O], so it is a traceless tensor.
Real Orthogonal Groups
397
I2, ? (01 Iii;,i, I , o i
(?,O,O,Ol
+ 27 B = 36 + 45 C=18 + 27 + 36
A = 18
 45
Fig. 9.2 The block weight diagram for [2,0,0,0] SO(8). of
398
Problems and Solutions in Group Theory
Real Orthogonal Groups
399
6. Calculate the spherical harmonic functions space.
Ygl in
an Ndimensional
Solution. The relations between the rectangular coordinates xa and the spherical coordinates T and e b in an Ndimensional space are
(9.19)
a=l
The unit vector along x is usually denoted by j;: = x / r . The volume element of the configuration space is
N
N1
The orbital angular momentum operators Lab are the generators of the transformation operators PR for the scalar function, R ESO(N),
F'rom the second Lie theorem, Lab satisfy the same commutation relations as those of the generators Tab in the self representation of SO(N). The Chevalley bases H,(L), E,(L) and F,(L) can also be expressed by Eqs. (9.10) and (9.11), where Tab are replaced with Lab. Because
N
c=l
N
c=l
where J a b are the generators of OR. The common eigenfunctions X, of H,(L) can be obtained from @a given in Eq. (9.12) for S 0 ( 2 l + 1) and in Eq. (9.15) for SO(2l) by replacing the vector basis 0, with the rectangular coordinate z a / T , where the factor T  ~ stands for removing the dimension of length.
400
Problems and Solutions in Group Theory
The spherical harmonic function Ykl(ii) is the eigenfunction of the orbital angular momentum for a single particle. Since there is only one coordinate vector x, the representation [A] for the spherical harmonic function Ykl(P) has to be the totally symmetric representation denoted by the onerow Young pattern [A] = [A, 0 , . . . ,O]. X,, multiplied by a normalizain tion factor, is nothing but the spherical harmonic function Ykl(ii) the selfrepresentation [A] = [l]. The component mp of the weight m is the eigenvalue of H p ( L ) in the eigenfunction X,. Generally, for the highest weight state M = (A, 0,. . . , 0 ) we have
Y i l ( k ) = cjvxx: = CN,*
{ (V
(51 T d 5
+ is,)}*
7
(9.22)
where t is equal to l when N = 2 l + 1 and to ( l  1) when N = 21. The remaining spherical harmonic function Y k l ( i ) with the weight m can be calculated by the lowering operators F p ( L ) ,just like we have done in Problems 4 and 5. Here we have to calculate the normalization factor C N , and the eigenvalue of the angular momentum square L2. Both problems can be solved from the highest weight state Y&](P). Since z1 + ix2 = reiel sin 02 . . . sin ODM1and
An
in
1 I . . 5 ...( n  1) ,. 3
de (sir
('*
we have
1 . 3 . 5 . : . (4
'
when n is odd,
For SO(2l
+ 1) group, we have
 2 ' 4 4 7 1 ) A+1 x+2 2*+2WA!(A +l  l)! (2X + 2  l)! ' l
*
( ...( A + l  1 ) 2  2 *43.65 . . *(2A2 + 2+12 l 2)l ) z ) 1 ( A
*
7r
* *
(9.23)
Real Orthogonal Groups
401
For SO(2l) group, we have
 21x7r
( ) 1 ... x + l x+2
('>
(X + l  l )
7r
= 2Xl(X+&
T'X!
I)!.
(9.24)
Assuming both b and d are not equal to 1 and 2, we have
Thus, the eigenvalue of L2 in the spherical harmonic function k'(k'+ N  2).
Ykl(Ec) is
7.Calculate the complete set of the independent bases for the eigenfunctions of angular momentum in a threebody system which is spherically
symmetric in an Ndimensional space. Solution. After separating the motion of the center of mass, there are 1 two Jacobi coordinate vectors in a threebody system, denoted by R = x and R 2 = y. Thus, the eigenfunction of angular momentum in the system belongs to an irreducible representation, denoted by a Young pattern with only one or two rows, [XI, X 2 , 0 , . . . ,O] = [XI, A,]. In a threedimensional space, the eigenfunction of angular momentum is described by the representation D and the magnetic quantum number m. But, in the Ndimensional ' space, N > 3, the eigenfunction of angular momentum is described by the representation [A,, A,] and the weight m. Due to the spherical symmetry, we only need to calculate the eigenfunction of angular momentum with the highest weight, namely, m = d in the threedimensional space and m = M = (A1  X2, X 2 , 0 , . . . ,0) in the Ndimensional space. The remaining eigenfunctions can be calculated by the lowering operators Fp( L ) ,just like we have done in Problems 4 and 5. Therefore, the eigenfunction of angular momentum with the highest weight in the Ndimensional space is described by two parameters A1 and X2. Denote by Q;lX2 the independent bases of the eigenfunctions of (x, y) angular momentum with the highest weight, where q is understood as an ordinal parameter. What the complete set means is that any eigenfunction (x,y) with the angular momentum [XI, A,] and the highest weight M can be expanded with respect to Q;IX2(x, y),
$Jk
(9.25)
402
Problems and Solutions in Group Theory
where the coefficients only depend upon the internal invariants t. In the threebody system there are three internal variables: = xx, 62 = y  y and 6 3 = x  y. What the independent bases means is that any basis cannot be expanded with respect to the other bases like Eq. (9.25). Obviously, a basis is not a new independent basis if it is a product of another basis and a function of the internal variables. Consider the product of two spherical harmonic functions Ykl(ii) and (y). It belongs to the direct product [q] x [p3 of two irreducible representations, which is usually a reducible one. Its reduction can be calculated by the LittlewoodRichardson rule and the contraction of the components of x and y. In the contraction the factor of the internal variables appears. Therefore, the independent bases of angular momentum come from the calculation result by the LittlewoodRichardson rule.
+q(c)
Yk)
The representations with t > 0 correspond to the bases which are not independent. Conversely, any independent basis of angular momentum [XI, A,] can be calculated by the product (k)Y$](y) where X2 5 q 5 X1 and p = X1 X2  q. Note that the highest weight state corresponds to the standard tensor Young tableau where each box in the first row is filled with 1 and each box in the second row is filled with 2 , and the tensor indices in the same column of the tensor Young tableau are antisymmetric. Now, the digit 1 means X1 = c(z1 i 2 2 ) or Y = c(y1 + iy2) and the digit 2 means 1 X = C(Q 2 izd) or Y =  c ( y ~ iy4), where c is a constant factor. 2 Up to the normalization factor, the independent basis state Q?lX2(x, is y) expressed as a product of three parts:
+
Ykl
+
+
+
For SO(4) group the irreducible representation denoted by a tworow Young pattern can be selfdual or antiselfdual. The standard tensor Young tableau for the highest weight state of the selfdual representation is still in the form ( 9 . 2 6 ) ,but that of the antiselfdual representation is different. In fact, each box in the second row of that standard tensor Young tableau for the highest weight state of the antiselfdual representation is filled with 3,
Real Orthogonal Groups
403
not 2. In this case, the independent basis state Q;IX2 (x, is expressed as y)
For SO(3) group, due to the traceless condition, the sum of the numbers of boxes in the first two columns of the Young pattern describing the irreducible representation is not larger than 3, namely, A2 = 0 or 1. On the other hand, the representation [ A , l ] for SO(3) is equivalent to the representation [A, 01. But, the bases of angular momentum for [A, 0 and [A, 1 1 1 have different parities.
Spinor Representations of S O ( N )
9.2
*
Let N matrices yu satisfy the anticommutation relations,
The set of all their products constitutes a finite group, denoted by r N . I’N is a matrix group. Usually, the matrix “(a is defined by its representation matrix in a faithful unitary irreducible representation of r N . The matrices yu, called the irreducible yu matrices, are unitary and hermitian. When N = 2 l is even, define
yf = (i)”2y1yz.. .yp/*
(9.29)
yf is also unitary and hermitian, and yf is anticommutable with any matrix ya. Thus, the matrix yj and N matrices yu satisfy the anticommutation
relations (9.28). They are regarded as the definition of yu in that
rN+l.
Note
The lefthand side of Eq. (9.30) changes its sign if one transposes arbitrary two matrices ya and yb,or to change the sign of arbitrary one matrix yu. When N = 4rn I, the righthand side of Eq. (9.30) is not a new element, is the group r4m+1 isomorphic onto the group r4m.When N = 4m  1, the righthand side of Eq. (9.30) is a new element, the set r 4 m  2 and its
+
404
Problems and Solutions an Group Theory
product with i is isomorphic onto the group
l?4m1.
r4m+l r4m, r4m1 M = (r4m2,
ir4m2).
(9.31)
When N = 21, there are 22'+1 elements in For any element of rN, except for the constant matrix, there exists at least one element which is anticommutable with the given element. Thus, the trace of the element is zero. The dimension d2')of the irreducible ya matrices which satisfy Eq. (9.28) can be calculated to be 2' by the character formula. The matrices in r2', neglecting their signs, are linear independent, and form a complete set of bases for the 2'dimensional matrices. The equivalent theorem says that any two sets of irreducible Ta matrices which respectively satisfy Eq. (9.28) are equivalent to each other. Namely, they can be related onetoone by a similarity transformation X : = XlyaX, 1 5 a 5 21. (9.32) When N = 21 1, the dimension of matrix ^la is still 2[. The equivalent condition for two sets of the irreducible ya matrices should include, in 1 addition to Eq. (9.28), that two products y . . . y~ are equal to each other. The equivalent theorem of ya plays an important role in the spinor theory of S O ( N ) . The charge conjugate matrix C and the spacetime inversion matrix B in particle physics can be defined based on it. When N = 21, clyac =  ( y a y , detC = 1, BlyaB = (%IT detB = 1,
+
*
CtC = 1, CT = (_l)W+1)/2C,
B ~ B 1, = B T = (1)!?('1)/2B.
,
Due to the condition (9.30), only matrix C can be defined when N = 4m1, and only matrix B can be defined when N = 4m 1. The fundamental spinor representation D["]( R )of SO( N ) , which is simply called the spinor representation in literature, is also defined based on the equivalent theorem:
+
Real Orthogonal Groups
405
Its generators Sab = iYaYb/2, a < b, are called the spinor operators. The fundamental spinor representation, being a group, is homomorphic onto S O ( N ) by 2 1 correspondence. When N is odd, the fundamental spinor representation D['I(R),or denoted by [s], is irreducible. [s] is real when N = 8k f 1, but selfconjugate when N = 8k f 3. Its dimension is di,] = 2(N1)/2. The fundamental spinor representation [s] is reducible when N is even. It can be reduced into two inequivalent representations, denoted by D[*"](R) [fs],[s]N [+s] @ [s], with the same dimension or d [ k S ]= 2(N/2)1. [fs] are conjugate to each other when N = 4k + 2. [ f s ] are real when N = 8k and selfconjugate when N = 8k + 4. if it transforms by the fundamental spinor representation D[']( R ) :
* In S O ( N ) transformation, Jr is called the fundamental spinor of S O ( N )
where Q is a column matrix with d[,] components. When N is even, @ can be decomposed into two parts by the project operators Pk = (1/2)(1 fyf), Q* = P*Q.
*
The spintensor is a spinor with the tensor subscripts,
The spintensor space is reducible. In its minimal invariant subspace, the spintensor has to satisfy the usual traceless condition and the second traceless condition:
and is projected by the Young operator. The Young pattern for the Young operator has the row number not larger than N / 2 , otherwise the subspace corresponds to the null space. The representation corresponding to the minimal invariant subspace is called the irreducible spinortensor representation or simply the spinor representation, which are described by a Young diagram with a symbol s: [s,X] = [ s , X ~ , X ~ ,    for an odd N ] or [fs, = [ f s ,XI, X2, .] for an even N . The spinor representation is A] doublevalued one of SO ( N ).

The dimension of a spinor representation [s,A] of S 0 ( 2 l + 1) or [fs, A] of SO(2l) can be calculated by the hook rule. In this rule the dimension
*
406
Problems and Solutions in Group Theory
is expressed as a quotient, where the numerator and the denominator are denoted by the symbols Ygl YiA1, and respectively:
(9.35)
We still use the concept of the hook path (i, j ) in the Young pattern [A], which enters the Young pattern at the rightmost of the ith row, goes leftwards in the i row, turns downwards at the j column, goes downwards in the j column, and leaves from the Young pattern at the bottom of the j column. The inverse path (i, j ) is the same as the hook path (i, j ) except for the opposite direction. The number of boxes contained in the hook path (i, j ) is the hook number hij of the box in the j t h column of the ith row. YiA1is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the hook number hij. Define a series of the tableaux Yitl recursively by the rule given below. Ykl a tableau of the is Young pattern [A] where each box is filled with the sum of the digits which are respectively filled in the same box of each tableau Yg' in the series. The symbol Ypl means the product of the filled digits in it, so does the symbol YiA1. The tableaux YE1are defined by the following rule:
(a) Y t ' is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the digit ( N  1 + j  i).
(b) Let [A(1)] = [A]. Beginning with [A(1)], we define recursively the Young pattern [A(")] by removing the first row and the first column of the Young pattern [A("')] until [A(")] contains less than two rows. (c) If [A(")] contains more than one row, define YE1to be a tableau of the Young pattern [A] where the boxes in the first ( a  1) row and in the first ( a  1) column are filled with 0, and the remaining part of the Young pattern is nothing but [A(")]. Let [A(")] be T row. Fill the first (T  1) boxes along the hook path (1, 1) of the Young pattern [A(")], beginning with the box on the rightmost, with the digits A?',
At),
Real Orthogonal Groups
407
   A?), box by box, and fill the first A!)' boxes in each , inverse path (i, 1) of the Young pattern [A(")], 2 5 i 5 r , with 1. The remaining boxes are filled with 0. If a few 1 are filled in the same box, the digits are summed. The sum of all filled digits in the pattern Y x l is zero.
8. Calculate the dimension of the irreducible spinor representation of the SO(7) group denoted by the following Young patterns:
Solution.
5431 321 678 5 $ 4
0 1 1 0 2
521 2 1
=8x
689 5 2 521 2 1
= 1728.
678 0 2 2 0 0 0 5 6 +I 0 $0 1 45 21 01 541 32 21
6 9 10 47
= 8~
L} 541 = 3024.
32 21
9. Calculate the highest weight of the fundamental spinor representations
of the S O ( N ) Solution. Fkom the second Lie theorem, the spinor operators Sub satisfy the same commutation relations as those of the generators Tab in the self representation of S O ( N ) . The Chevalley bases H,(S), E,(S) and F,(S) in
408
n Problems and Solutions a Group Theory
the fundamental spinor representation can also be expressed by Eqs. (9.10) and (9.11)) where T a b are replaced with S a b . It is convenient to choose an explicit forms of the ya matrices for calculating the highest weight of the fundamental spinor representations, although the calculated result is independent of the chosen forms. Since the dimension of ya for the SO(2l) group is 2e, we may define the ya matrices of l?2e to be the direct product of l Pauli matrices such that Eq. (9.28) is satisfied:
In fact, they are also ya matrices of r 2 0 + 1 with 7 2 e + 1 = yf. Thus, the Chevalley bases in the fundamental spinor representation of SO(2l + 1)) whose Lie algebra is Be, are
HP= 1x . . x 1xdiag(0, 1,  1, 0) x 1 x .. x 1
)
E P = IX ~ X
P1
F P = IX ~ x { ~  x o + } x ~x xl , x
P1
ePi
He=
Ee =
  y2P1
= 1x =
... x
1x01 x
;x02
03
x
... x
03
,
y2P
t x .;
03
P1
1P
x
x
03
x
... x
03
,
(9.36)
P1
yf =
x
... x 0 3 .
e
eP
P1
ePi
, "
X ( O + X O  } X ~ X.
1
ePi
.xl
(9.37)
IX~ 1 x 0 3 ,
e
1
03
x
. * *
x
0 3 XQ+
)
Fe =
03
x *..x
0 3 XCT
)
e
1
e
1
where 1 5 p < l . The Chevalley bases in the fundamental spinor representation of the SO(2l) group, whose Lie algebra is De,are the same as those of the S 0 ( 2 l + 1) group except for those when p = ! which are ,
He = 1x
Ee=
Fe=  1~. . . x I x { a _ x a  } .
e2
e2
x l x d i a g ( 1 , 0, 0, 1)
)
 l x
... x l x { o + x a + } ,
(9.38)
e2
Real Orthogonal Groups
409
Let a and
p be two eigenfunctions of 03,
a=(;),
ff3a= a ,
respectively,
P=(;),
f f  a = p,
aP = 0.
ff+a 0, =
asp = p,
a+p = a ,
Denote by x(m) the basis states with the weight m in the fundamental spinor representation [s]of S 0 ( 2 & +1) and by x*(m) the basis states with the weight m in the fundamental spinor representation [ks] of SO(2l). From the condition that each raising operator annihilates the highest weight state, we obtain the basis state x(M) for the highest weight for [s] of S0(2k‘+ 1) and the basis state x,t(M) for the highest weight for [ks]of SO(2C) as follows:
x(M) =
CXX.XQI
1 ’
e
M = (0, 0, ..., O , l ) , x+(M) =
for [s] of SO(2&+I),
’ e M = (0, 0, ..., O , l ) ,
x(M) =
e1
C Y X .  . X ~
M = (0, 0, ..., l,O),

(9.39)
for [+s] of S0(2&),
C Y X .  . X C X X ~ ,
for [s] of SO(2l).
10. Calculate the highest weight of the spinor representation [s,A] of
S0(2&+1),and the highest weights of the spinor representations [fs, A]
of SO(2t)
Solution. The spintensor of S O ( N ) corresponds to the irreducible representation [s,A] when N = 2 & + l or to [ks, when N = 2&if the spintensor A] satisfy the usual traceless condition and the second traceless condition, given in Eq. (9.34). The generators in the irreducible spinor representation a are the total angular momentum operators Jab = L a b Sub, < b. The orbital angular momentum operators Lab and the spinor angular momentum operators S,b are respectively applied to the tensor part and the spinor part of the spintensor. The total angular momentum operators Jab satisfy the same commutation relations as those of L a b and Sab. The Chevalley bases H,(J), E,(J) and F,(J) in the irreducible representation [s,A] can also be expressed by Eqs. (9.10) and (9.11), where Tubare replaced with
+
410
Problems and Solutions in Group Theory
Jab. Therefore, the weight of a spintensor basis is the sum of the weights of the tensor part and the spinor part. Let [A] = [A,, Aa, . . . , A,]. The highest weight M of the spinor representation [s,A] of S0(2e+ 1) is
The highest weight M of the spinor representation [+s,A] of SO(2e) is
The highest weight M of the spinor representation [s,A] of SO(2e) is
11. Calculate the ClebschGordan series for the direct product of the tensor representation [A] and the fundamental spinor representation [s] of S0(2C+ 1) or [fs] of SO(24, where [A] is a single row Young pattern or a single column Young pattern.
Solution. If the tensor part of a spintensor qal...,,, belongs to the tensor representation [A] but it does not satisfy the second traceless condition, namely, its tensor part satisfies the traceless condition for each pair of the subscripts a and aj and is projected by a Young operator, then, the spini tensor space corresponds to the direct product of the tensor representation (A] and the fundamental spinor representation. The method for reducing the direct product representation is to decompose the spintensor space into a series of subspaces satisfying the second traceless condition (9.34). In fact, x b TbQal...b...a,, is a new spintensor where the tensor rank decreases by one. If [A] is a single row Young pattern, the tensor part of the spintensor is a totally symmetric traceless tensor such that twice application of the Tb~c*al...b...c...a, = 0 due to Eq. second traceless condition is trivial, (9.28). If [A] is a single column Young pattern, the tensor part of the spintensor is a totally antisymmetric tensor. Note that the antisymmetric tensor space of rank n of S O ( N ) is equivalent to that of rank N  n. Thus, the reductions of the direct product representations for SO(2.t 1) are
zbc
+
[A,O,.
..,O]
x [s] 2 [s,A,O,.
. ,O] c3 [s, (A
*
 l),O,
* * *
,O],
[PIx
[s] 21 [s, I”] CEI [s, l n   l ]
CB .. .
[s],
n 5 .t
,
(9.40)
and those for SO(2l) are
Real Orthogonal Groups
411
[A)O,.
. . ,O]
x [h] N [As,
x,o, . . . ,O] a3 [ys,(A  l ) , O , . . . ,O],
where P = (1 f y f ) / 2 . * The sign in f s changes because P k y , = Those formulas can be shown with the method of the dominant weight diagram. The reader is encouraged to check them for the groups SO(7) and SO(8). Equations (9.40) and (9.41) can be partly checked by their dimensions. From Eqs. (9.18) and (9.35), the dimensions of [A] and [s,X] of S O ( 2 t + l ) ,for example, are ( 2 t ) ( 2 t + 1).. . ( 2 t + x  2 ) ( 2 t + 2 x  1) qx,o, ...,01 = A! ( 2 t + x  2 ) ! ( 2 t+ 2 x  1) > (21  l)!A! ( 2 l f l)! ' when n 5 t , d ~ l n ~( 2 t  n = 1 ) ! n !7 (2!+ (2!)(2t 1).. . ( 2 t  1) =2 +,x,o ,...,0 = 2 1 ' A!
+
+
+x
'
x  l)!
Then, for Eq. (9.40) we have 2e (2t+
( 2 t  l)!X! ( 2 t  l)!X! ( 2 t + l)! ( 2 t + 1 ) ! ( 2 t  2a + 2 ) 2e =2 e x ( 2 t  a 2)!a! (21 1  n)!n! a=O
x  2 ) ! ( 2 t + 2 x  1) = 2 '( 2 t + A  l)!
+
2t
( 2 t A  2)! ( 2 t  1)!(A  l ) !
)
+
+
+
)
when n
5 t.
12. Calculate the basis states in the fundamental spinor representation [s]
of SO(7).
Solution. The Lie algebra of the SO(7) group is B3. The highest weight of the fundamental spinor representation [s]of SO(7) is (0, 0 , l ) . Its dimension is 8. Its block weight diagram is as follows. All nonvanishing matrix entries of the lowering operators FD are 1.
412
Problems and Solutions an Group Theory
By the notation used in Problem 9, we have
i , i ,1
I
JI i , i 1,
x ( i 7 i 1) = 7
x(i,o,i) = 
x p x a,
x
p x p,
x(o,i,i)=  p x p x a , ~ ( oo, i ) =  p x p x p.
Fig. 9.3 T h e block weight diagram of [s] of B3.
13. Calculate the basis states in the fundamental spinor representations [AS] SO(8). of
Solution. The Lie algebra of SO(8) group is Dq.The highest weights of the fundamental spinor representations [+s] and [s] of SO(8) are (O,O, 0 , l ) and (0, 0,1,0), respectively. The dimensions of both representations are 8. Their block weight diagrams are as follows. All nonvanishing matrix entries of the lowering operators Fp are 1. By the notation used in Problem 9, we have
x + ( O , O , O , 1) = a x a x a x a ,
x(O,O,l,O)
=a x
x a x p,
0,
x+(o,i,o,i) = 
p x p, x + ( i , i , i , o ) =  x p x x p, x + ( i , o , i , o ) =  p x a x x p,
x x
x(o,i,i,o) = a x a x p x
x+(i,o,i,o) =  a x p x p x a,
x(i,i,o,i) = a x p x a x a , x(i,o,o,i) = p x x x a , x(i,o,o,i) =  a x p x p x P ,
x(o,i,i,o) =  p x p x x(o,o,i,o) =  p x
x + ( i , i , i , o ) =  p ~ ~ ~ p ~ i ~i , o , i ) =  p x c ~ x p x p , x ( ,
x+(o,i,o,i)=  p x p x a x a , X+(o,o,o,i) = p x p x p x p,
x p,
p x p x a.
Real Orthogonal Groups
413
a) [+sl
Fig. 9.4
b)
[Sl
The block weight diagrams of [fs]of Dq.
14. Expand the eigenfunction of the total angular momentum with the highest weight in terms of the product of the spherical harmonic function Y:’(EE) and the spinor basis x(m). Solution. In Problem 11 the ClebschGordan series for the direct product representation [X,O, . . . ,O] x [s] is given. For the SO(2t 1) group, we express the representations in Eq. (9.40) by their highest weights,
+
(A,O..
.,O)
x ( 0 . . . ) O , 1) 21 (A,O..
. ) O ) 1)@ (A  1 , o . . . , O ) 1).
Let (J) = (A,O.,.)0,1), (L)= ( A , O , . , , O ) , @ + I ) = ( A + 1 , 0 ...,O), and (S) = ( 0 . . . , O , 1). There are two ways to obtain the total angular momentum state ( J ) : ( L ) x (S) and ( L 1) x (S). The highest weight state can be calculated by the condition that it should be annihilated by each raising operator. The highest weight state for the first case is easy to calculate:
+
where C(z~+l),x given in Eq. (9.23). The highest weight state for the is second case is
414
Problems and Solutions in Group Theory
For the SO(2C) group, we express the representations in Eq. (9.41) by their highest weights,
(A, 0 . . . , O ) x ( 0 . . . ,o, 1) 2 (A, 0 . . . 0 , l ) @ (A  1)0 . . . ,o, 1)O),
)
(A,O
* *
 ,O)
x (0.  ) O , 1 , O )
*
N
(A,O.. . , O ) 1,O) 6 (A  1 ) O . . . ,o, 1). 3
Let ( J + ) = (A$. . . , O , l),( J  ) = (A$. . . , O , 1,0), ( L ) = (A$. . . , O ) , ( L 1) = (A 1 , O . . . , O ) , (+S) = ( 0 . . . , 0 , l),and (S) = ( 0 . . . ,O,l,O). There are two ways to obtain the total angular momentum state (J*): ( L )x (fS) and ( L 1) x ( r S ) . It is easy to calculate the highest weight state for the first case:
+
+
+
where C ( 2 1 )is~ , given in Eq. (9.24). The highest weight state for the second case is
Real Orthogonal Groups
415
+ (z3+ i ~ &  [ ( i , i ,0 , . . . ,o, i)] + (21 + iz2)x[(i, 0, ,0, l)]}* * *
9.3
SO(4) Group and the Lorentz Group
The inequivalent irreducible representations of the Lorentz group can be obtained from those of SO(4) group. This is the general method for studying the inequivalent irreducible representations of a noncompact group.
*
*
Find new bases of generators of SO(4) from six generators Tab,
416
Problems and Solutions an Group Theory
The new bases are separated into two sets. Two generators in different sets are commutable. Express them in the form of the direct product of the Pauli matrices:
By the similarity transformation N , we have
N =  1(
Jz
10 0 i 0 0 0 1 1 0 ii
1 i 0 0
)

Therefore, any element R of SO(4) can be expressed in the direct product of two matrices
R = exp i
a<b=2 3
(
4
\
WubTab
)
1
= expli
I
where
3
Thus, matrix R is explicitly expressed in the product u1 x 212 of twodimensional unimodular unitary matrices. If two matrices u1 and u 2 change their signs at the same time, R keeps invariant. Therefore, Eq. (9.43) gives
Real Orthogonal Groups
417
a onetotwo correspondence between an element R in SO(4) and two elements u1 and u2 in SU(2). Furthermore, this correspondence keeps invariant in the multiplication of the group elements. Thus,
SO(4)

SU(2) x SU(2)'.
(9.44)
We restrict the varied area of the parameters of the SO(4) group such that there is a onetoone correspondence between a set of parameters and a group element, at least in the region with nonzero measure: (9.45) where O(*) and v(*)are the polar angle and the azimuthal angle of the direction n(*). Since the group space of the second SU(2)' group is reduced to that similar to the group space of the SO(3) group, namely, the two ends of a diameter in the group space correspond to the same element, the group space of SO(4) is doublyconnected. Any irreducible representation of SO(4) can be expressed as the direct product of two irreducible representations of two SU(2) groups, denoted by
Djk :
Djk
(fi(+),u(+);~(),u())
= ~ (fi(+),J+)) . i
~k
(&),J)).
(9.46) The row (column) index of Djk is denoted by two indices ( p v ) . The dimension of D j k is ( 2 j + 1)(2k 1). Its generator I:"*) is expressed by the generator I: of Su(2):
+
(9.47)
+ Ic) is an integer, D j k is a singlevalued representation of S0(4), i.e., the irreducible tensor representation. When ( j + k  1/2) is an integer, D j k is a
(j
Djk
with different superscripts jIc are inequivalent to each other. When
doublevalued representation of S0(4), i.e., the spinor representation. The correspondence between D j k and the irreducible representation denoted by
418
Problems and Solutions in Group Theory
a Young pattern [X,, A,] is as follows. When j
+ k is an integer, we have
when j = k , when j when j
Djj
[ 2 j , 01,
Djk N [ + ( j Djk 21 [  ( j
and when j
+ k ) ,( j  k ) ] + k),(k  j ) ]
> k, < k,
when j > k, when j
+ k  1/2 is an integer, we have Djk 21 [+s, ( j + k  1/2), ( j  k  1/2)], Djk 21 [s, ( j + k  1/2), (k  j  1/2)],
< k.
Note that [+X,,Xz] is the selfdual representation, and [X,,Xz] is the antiselfdual one. DOo is the identical representation, Dqo and Do* respectively are two fundamental spinor representations [+s] and [s]) and D f 3 is equivalent to the selfrepresentation.
*
The Lorentz transformation matrix A between two inertial frames is a
4 x 4 orthogonal matrix:
A*A = A A = I. ~
The matrix entries of A satisfy the following condition: Aab and A44 are real
Aa4
(9.48)
and A4a are imaginary
a and b = 1, 2, 3.
This condition keeps invariant in the product of two matrices A. The set of all such orthogonal matrices A, in the multiplication rule of matrices, constitutes the homogeneous Lorentz group, denoted by O(3,l) or Lh. The orthogonal condition (9.48) gives detA = f l , Ai4 = 1 +
3
IA,4I2 2 1.
a=l
These two discontinuous constraints separate the group space of O ( 3 , l ) into four unconnected sections. The element in the section where the identity belongs to satisfies detA = 1, A44 2 1.
(9.49)
Those elements satisfying Eq. (9.49) constitute a simple Lie subgroup, called the proper Lorentz group, denoted by L,. Since there is no upper limit of A44, the group space of L, is a n open area in the Euclidean space. Thus, L, is a noncompact Lie group. The representative elements in three
Real Orthogonal Grozlps
419
cosets of L, are usually chosen to be the space inversion sion r and the whole inversion p:
o = diag(1,

CT,
the time inver
1,  1, 1),
 1).
r = diag(1, 1, 1,  1), p = diag(1,  1,  1,
rt Let A be an infinitesemal
A = I iaX,
element of L,:
1 = ATA = 1  ia ( X
+XT),
1 = d e t A = 1 i a T r X ,
AT = 1  i a X T , XT = X, Tx = 0. r
Thus, X is a traceless antisymmetric matrix. Expand X with respect to the generators in the selfrepresentation of SO(4):
3
a<b=2
3
a=l
(9.50)
a=l
where
dab
is real,
wa4
is imaginary, and
3
(9.51)
b<c=2
Except for the imaginary parameters, the generators in the selfrepresentations of SO(4) and L, are completely the same, so are the generators in the corresponding irreducible representations of two groups. The finitedimensional inequivalent irreducible representations of L, are also denoted by Djk(L,), whose generators are given in Eq. (9.47). Since the parameters for SO(4) and L, are different, the global properties of the two groups are very different. Any finitedimensional irreducible representation D j k of L,, except for the identical representation, is not unitary. There exist infinitedimensional unitary representations for L,. The transformation generated by T a b is obviously a pure rotation, belonging to the subgroup SO(3). Let us study the transformation generated
*
420
Problems and Solutions in Group Theory
by
T4 3
with the parameter
w34
= iw:
A(G3,iw) rexp{i(iw)T34} =
("
0
0
0 0 coshw isinhw
0 0 isinhw coshw
)
(9.52)
w = ctanhw,
coshw = (1  w2/c2)1'2,
sinh w = (w/c) (1  'u2/c2) 1'2
.
It describes the Lorentz boost along the direction of x axis with the relative velocity w = c tanh w. Any Lorentz transformation A satisfying Eqs. (9.48) and (9.49) can be decomposed into the product of some rotational transformation and the Lorentz boost along the x direction. Let A 4 4 = coshw, from which 13 is determined. Extracting a factor i sinh w from Aa4, we obtain a unit vector n(8, cp) in the threedimensional space, whose rectangular coordinates are (i/ sinhw)(A14rA24,A34): = COShw, iA24/ sinh w = sin 8 sin cp,
A44
iA14/sinhw = sinOcoscp, iA34/ sinh w = cos 8.
(9.53)
Extending the rotational matrix into a 4 x 4 matrix, we have coscpcos8  sincp coscpsin8 0 sincpcoso coscp sincpsin8 0 0
R(g3 ,p)R(s2)0) =
0
0
0
1
Leftmultiplying to A the inverse matrix of R(G3, cp)R(&, 8)A(G3)i w ) ) we obtain a rotational matrix R(a,p, T),
from which the Euler angles a , P and y can be calculated. Thus,
The geometrical meaning of the decomposition is evident. Two rotations in twosides of A(&, i w ) rotate two z axes in two frames before and after the Lorentz transformation A to the direction of the relative motion, and the remaining axes to be parallel to each other, respectively. After two rotations, the Lorentz transformation A is simplified into A(&, '1'13).
Real Orthogonal Groups
42 1
Since all inequivalent irreducible representations of L, have been known, the inequivalent irreducible representations of the homogeneous Lorentz group O(3,l) can be obtained by determining the representation matrices of r and p. The representation matrix of o can be calculated by the formula o =~ p . Since p is commutable with any element in O(3,l) and the p2 is equal to the identity, the representation matrix of p in an irreducible representation of O(3,l) is a constant matrix, where the constant is f in l a singlevalued representation. The property of the representation matrix of r can be determined by the relations in the selfrepresentation
*
Let V ( x ) X = 1, 2, 3 and 4, be four inequivalent irreducible represen, tations of the fourorder inversion group V4. When j = k , The irreducible .. representation D J J of L, induces four inequivalent irreducible representation Ajjx of O(3,l):
AjjX(A)= D j j ( A ) , A E L,, (7) V(’) (T)S,,~ 1 5 X 5 4. Ajjx(p) = V(’)(p)1,
SvcLj,
(9.55)
When j # k , while j k is an integer, the representation D j k @ Dkj of L, induces two inequivalent irreducible representation Ajk* of O(3,l):
+
where a and P are used to identify two representation spaces of D j k and D k j . Since Ajk*(r) has no diagonal entries, changing its sign leads to an equivalent representation. When j k is a halfodd number, we only discuss the Dirac spinor representation D(O(3,l)) here. For definiteness, the index p runs from 1 to 4, while the index a runs from 1 to 3. Let T~ be four anticommutable matrices and C be the chargeconjugate transformation matrix:
+
cly,c = ; y,
C+C= 1,
CT = c,
d e t C = 1.
422
Problems and Solutions in Group Theory
Thus, D(A) satisfies
4
D(A)'ypD(A) =
u=l
Apuyu,
detD(A) = 1, (9.57)
ClD(A)C =
IA44 I
{D(Al)}*.
The additional factor A44/IA441 comes from the physical reason. Its generators are
The representation matrices of the representative elements in the cosets of L, are
D ( a ) = fiy4,
D ( 7 ) = fy4y5,
D ( p ) = fiy5.
(9.59)
15. Discuss the classes in the SO(4) group and calculate their characters
in the irreducible representation
Djk.
Solution. Any element R of SO(4) can be transformed into the standard form by a real orthogonal similarity transformation X:
where Eq. (9.43) is used. Therefore, the class of SO(4) is described by two parameters cpl and y2. Its character in the representation D j k ( R )is
Real Orthogonal Groups
423
16. Calculate six parameters of the following proper Lorentz transformation A, and write its representation matrix in the irreducible representation Djk ( A )of the proper Lorentz group L,:
A(cp7 w 7 "
"
/1 0 0 0 \ 0 &/2 (cosh w ) / 2 i(sinh w ) / 2 = 0  1 / 2 &(cosh w ) / 2 i&(sinh w ) / 2 \O 0 isinhw coshw 1
Solution. From the fourth column of A(cp,8, w , a ,P, the boost paramer), ter is w , and 6' = 7r/6 and cp = 7r/2 because sin 6' cos cp = 0, sin0 sin cp = 1 / 2 and cos 0 = &/2. The matrix form of the rotation R(7r/2,7r/6,0) is 0100 1 0 00 R(7r'2,n/6,0)= 0 0 l o ) ( 0 0 01
(
fi/2 0 ;/2 0
:
0
1/2 0
&2 0
;)
/
R(n/2,7r/6,0)' =
0 &/2
1/2
o\
Thus,
From this we obtain representation matrix of A in
and
Thus, we obtain the
424
Problems and Solutions in Group Theory
17. Prove that the Dirac spinor representation satisfy
Solution. The Dirac spinor representation is not unitary. The aim of this problem is to study the property of its conjugate representation. From the definition (9.57) of the Dirac spinor representation we have:
To make the formulas uniformly, we make a similarity transformation
74:
From the equivalent theorem of 7a, obtain we
Substituting Eq. (9.59) into it, we determine the constant c :
This is the reason why the matrix 7 4 should be inserted between two spinors $t and 1(, to construct an invariant quantity of the Lorentz transformation in the quantum field theory:
18. Discuss the classes of the proper Lorentz group L,.
Solution. In the selfrepresentation of L, we write the pure rotation R ( MP, 7 ) and the Lorentz boost A(&, i w ) with the relative velocity along ,
the z axis in terms of Eq. (9.43)
Real Orthogonal Groups
425
Thus, any element A(cp,8, w , a , P, r) in L, can be expressed as
where
The matrix M is a twodimensional matrix with determinant +1, which belongs to the twodimensional unimodular complex matrix group SL(2 ,C). We are going to show the homomorphism between SL(2,C) and L, with a 2:l correspondence. First, we have known from Eq. (9.60) that an arbitrary Lorentz transformation A = A(cp,8,w , a , P, 7) corresponds to a matrix M E SL(2,C). Conversely, any M given in Eq. (9.61) determines a Lorentz transformation A. We will show that any element P in SL(2,C) can be expressed in the form (9.61) later. If two matrices M and M' correspond to the same Lorentz transformation A, then
Thus, M  l M ' = c l . Since detM = 1 and detM' = 1, c2 = 1 and c = f l . There is a 2:l correspondence between fM given in Eq. (9.61) and the Lorentz transformation A. This correspondence keeps invariant in the multiplication of elements. In fact, if
we have
AA' = N { M M ' x
[ 0 2 ( M M ' ) * ~ 2N } l . ]
Now, we have proved that the twodimensional unimodular complex matrix group SL(2,C) is homomorphic onto the proper Lorentz group L, with a 2:l correspondence. Thus, we can determine the class of L, by SL(2,C). Recall that two matrices fM eSL(2,C) correspond to the same Lorentz transformation A E L . Two classes in SL(2,C) different by a sign correspond to , the same class in L,. If two eigenvalues of M eSL(2,C) are different, M can be diagonalized through a similarity transformation X eSL(2,C)
426
Problems and Solutions in Group Theory
Thus, we have
A=
(
0 coscp  s i n 9 sincp coscp 0 0 0 coshw 0 0 isinhw coshw
n
I cp I
r 1
The class is described by two parameters w and cp. If two eigenvalues of M are equal, the eigenvalues have to be fl.fl correspond to the identity element of L,, which constitutes one class in L,. If M is not equal to fl, M can be transformed into the Jordan form by a similarity transformation Y eSL(2,C) (see Problem 16 in Chap. 1):
Y%Y=f(,
Since
12
1>.
1 we obtain 0 = 22, 0 2 = 2, 0 3 = 0. In order to calculate the six parameters of A , we make a transformation
The solutions are cp = and
7r,
6 = r/4, coshu = 3, a = T , p = 37r/4, y = 0, '
0 2
 ei2T31 2Tl4
1 2 2i
0 1 2i 0  2i 3
It is easy to check that A is a proper Lorentz transformation with all four eigenvalues to be 1. A has only two linearly independent eigenvectors for the eigenvalue 1. A is transformed into the Jordan form by the similarity
Real Orthogonal Groups
427
transformation 2:
0 0 21 040 04io
1 0
0 0
1 0
O
0 0 0 i/4
l i
ZlAZ=
( )
1 0 0 0 0 1 0 0 0 1 1 0 0 0 1 1
O
*
Finally, we are going to show that, any twodimensional unimodual complex matrix P ESL(2,C) can be expressed into the form of Eq. (9.61), namely
Solving the equation, we have
where c,g = cos(O/2), = sin(O/2) and cp = cos(P/2) and so on. The direct calculation leads to
+ c$cgew+ sis$e*, IB12 = (sin8 sinp cosa) /2 + cis$ew + sic$ew, ICI2 = (sin8 s i n p cosa) /2 + s;4cgew+ czsge", 1OI2 =  (sin 8 sin p cos a ) /2 + sis$ew + cic$ew, AB = eiCP {  cos p sin 8 + sin p (cieweia + siew ia> > / 2 , AC = eiy {case s i n p + sin8 (c$eweia  s$eweia)} /2,
IAI2 =  (sin 8 sin p cos a ) /2 AD = s i s $ +tic$  secespcp (eweia  e+eia), [A12 IBI2
+
+ [C12+
= ew ew,
+
428
Problems and Solutions in Group Theory
w , 8 and
p can be respectively calculated from the last three formulas. Then, a is determined by the formula for AD. At last, y and 9 can be calculated from the formulas for AC and AB. Here, the calculated method is not unique due to d e t P = 1. This completes the proof. For example, we calculate the parameters of a matrix P :
P=( 1 4
4
[a 1 i (a l)] a 1 i 3 (a 1) + + + a+ i 3 ( A 1) d [ A +1 + i ( A I ) ] 1

coshw = 3,
sinhw =
a,
ew/2=
Jz+ 1,
1/2,
ew/2 =
Jz 1
case = 112, e = 7r/3,
Then, from A D = 3(1  i&)/S,
C O S = ~
p = 21~/3.
we obtain
+ eweiff = 6 c o s a  i 4 a s i n a =
The solution is a = 7r/2. Finally, from
AC = a [  1  2i(3  &)]  eiy
8
we have ezv = i and last, we have
eeiY
= 1, and determine cp = 7r/2 and y = 7r. At
19. Express an arbitrary element in the proper Lorentz group L, in the form of exponential matrix function.
Solution. In Problem 18, we have shown that L,

SL(2,C) with the
Real Orthogonal Groups
429
correspondence (9.60):
An arbitrary element A in L, can be written in the form of exponential matrix function if its corresponding elements f M in SL(2,C) can. Since detM = 1, its two eigenvalues have the same sign. If its eigenvalues are positive, M can be transformed into the form of exponential matrix function by a unimodular similarity transformation. If its eigenvalues are negative,  M can be transformed into the form of exponential matrix function by a unimodular similarity transformation. Ignoring the irrelevant sign, we can express an arbitrary element M in SL(2, C) as
M
= fexp ( 4 6  $12)
,
02
M*u2 = fexp (46*. $12)
. M
By Eq. (9.60), the proper Lorentz transformation A corresponding to can be expressed as
a<b
a=l
where Ti*), Tab and Ta4 are the generators in the selfrepresentation of S0(4), and the parameters Wab and wc4 are the real part and the imaginary part of O,, respectively:
1 Wab = 2 c=
3 &bc
1
(0,
+ n:),
Wc4
=  (a,  n:).
1 2i
430
Problems and Solutions in Group Theory
Thus, an arbitrary Lorentz transformation A can be expressed in the form of exponential matrix function, where the parameters u a b are real, and ma4 are imaginary. The set of parameters up,is mainly for the theoretical study, and the set of parameters (cp,8, u,a , P, 7 )is convenient for calculation. The relation between two sets of parameters are given by Eq. (9.60), namely
20. Let
x = ix41+
C
a=l
3
xaga
=
23
x1
+ ix2 23  ix4 ) ’
x1 3
 ix4
 ixa
where xa is real, 24 = ict is imaginary, and X is a hermitian matrix. Let M E S L ( ~ , C ) ,
M X M ~ X I = ix&l+ =
C
a=l
XLua.
(a) Calculate Tr X and det X . (b) Prove xh = C , Ap,x,, where A (c) Prove that L, SL(2,C). is a proper Lorentz transformation. (d) Calculate the corresponding matrix M , when A is equal to B(G3,cp), R(&,8), and A ( & ,i w ) , respectively. (e) Calculate the matrix M corresponding to an arbitrary proper Lorentz transformation 4% u,a , P, 7 ) . 6,
4
Solution. (a) TrX = i2x4,
det X = p=l
x:.
(b) Since XI = M X M t is hermitian, it can be expanded with respect to the Pauli matrices and 1, where the coefficients xb, are real, and x i is imaginary. Further, x; can be expressed as the linear combinations of x p ,
4
u=l
where A a b and det X’ = det X ,
A44
are real, and
A4a
and
Aa4
are imaginary. Due to
4
p=l
4
p=l
Real Orthogonal Groups
43 1
Therefore, A is a Lorentz transformation matrix, which is an orthogonal matrix with the matrix entries depending upon M . When M = 1, A is the identity. Since the group space of SL(2,C) is connected, the set of A can be continuously extended from the identity, namely, A is a proper Lorentz transformation matrix. (c) From a given matrix M , we can calculate X' = MXMt, and then, determine a proper Lorentz transformation matrix A uniquely. If the matrices X ' calculated from M and M' are the same, so are the matrices A , then MlM' is commutable with three Pauli matrices ua. Thus, it is a constant matrix. Due to detM=detM' = 1, M' = & M . Namely, there is a 2:l correspondence between & M and A . This mapping is obviously invariant in the multiplication of elements. Therefore, SL(2,C) is homomorphic onto L,, L, SL(2,C). (d) Ignoring the sign in front of the matrix M , we can express the matrix M and A by the same parameters. If A ESO(3) is a pure rotation, M obviously belongs to SU(2),

M(&, cp) = u(e'3, cp) = 1cos (cp/2)  i 0 3 sin (cp/2), M(e'2,O) = u(e'2, 8) = 1cos (8/2)  iaasin (e/2),
For a boost A = A(&,iw), we have M(&,iw)XM(e'3,iw)t = il(ix3 sinhw + 2 4 coshw)
+ 03(x3 coshw  ix4 sinhw) + 01x1 + 02x2.
If only one component xp is nonvanishing, we have
M(G3, iU)OlM(e'3, i w ) t = 01, M(e'3, iw)a2hf(e'3,iw)t = 0 2 , Ad(&, iw)a3M(Z3, iw)t = 1 sinhw M(G3,iw)M(G3,iw)t= 1 coshw
x1 = 1,
+ 0 3 coshw,
x2 = 1,
23
24
+ 0 3 sinhw,
= 1,
= i.
Let M(e'3,iw) = 1Mo+alM1 +a2M2+a3M3. Extracting the Pauli matrix from the lefthand side of the first two formulas, and then, moving it to
432
Problems and Solutions an Group Theory
their righthand side, we have
(1Mo + a l M 1  02M2 (1Mo  a & + a2M2
 03M3)
M t = 1,
 a3M3) M += 1.
Adding and subtracting two equations, we obtain MI= M2 = 0, IMo12 lM3I2 = 1, and MOM,* are real. Thus, the last two formulas become the same, and turn to lMOl2 lM3I2 = coshw and 2MoM,* = sinhw. Recall that det M = 1. The solution is MO = coshw/2 and M3 = sinhw/2, namely
+
M(e"3,iw)=
( eow/2)
ewoi2
= u(Z3,iu).
(e) For an arbitrary proper Lorentz transformation A(p, 8, u,a, p,?), we have
Chapter 10
THE SYMPLECTIC GROUPS
10.1 The Groups Sp(24 R ) and Usp(21)
j r Take the subscript

a of a vector in a (2l)dimensional space to be p or
a=l,i,2,2,
p, 1 5 p
5 i?, in the following order: ,e,Z.
(10.1)
Define a (2l)dimensional antisymmetric matrix J : when a = p, b = p when a = p , b = p otherwise, det J = 1 . J = l c x (ia2) = Jl = JT,
Jab
=
{
1 1 0
(10.2)
The set of all (2t) x ( 2 4 real matrices R satisfying R T J R = J,
R* = R,
(10.3)
in the multiplication rule of matrices, constitutes the real symplectic group Sp(2l, R ) . Both the inverse R' and the transpose RT of any element R of Sp(2l, R) satisfy Eq. (10.3):
R1
= JRTJ,
( ~  1 ) R~ 1 J 
= J,
J (  J R T J ) ~ J (JRTJ) J = RJRT = J.
(10.4)
Sp(2[, R) is not a compact Lie group, because the general form for a diagonal matrix & E Sp(2l, R) contains the parameters up without an upper limit:
& = diag {ewl,
ewl,
ew2, e  W 2 ,
433
. . . ,e W t , e  w t } .
(10.5)
434
Problems and Solutions in Group Theory
*
The set of all ( 2 4 x ( 2 4 unitary matrices u satisfying
uT J u = J,
ut = u ,
'
(10.6)
in the multiplication rule of matrices, constitutes the unitary symplectic group USp(2.t). Similarly, both the inverse ul and the transpose uT of any element u of USp(2l) satisfy Eq. (10.6). Since any matrix entry of a unitary matrix is finite, USp(2.t) is a compact Lie group. As an example, the general form of a diagonal matrix uo EUSp(2.t) is:
It can be shown that the determinant of R ~ S p ( 2 . t , R ) u ~ U S p ( 2 t ) or is +l. The group spaces of both Sp(2.t, R) and USp(2.t) are connected.
*
For the infinitesimal elements R ~ S p ( 2 [R ) and u E U S P ( ~ ~ ? ) , ,
R = 1iaX, u = 1  ipY,
XT =JXJ, YT = J Y J ,
X* = X, Y t = Y.
(10.8)
An imaginary matrix X of 2.t dimensions contains 4.t2 real parameters, so does a hermitian matrix Y . The component form of the condition XT = J X J is
which gives .t2 + .t(.t  1) real constraints. The first equation gives T r X = 0, which means detR = 1. The component form of Y T = J Y J is
yV P =yP"
7
yP =y" P"'
Due to Y t = Y , the first equation contains .t2 real constraints, including T r y = 0, and the second equation contains l ( 1  1) complex constraints. Therefore, there are l ( 2 t 1) real parameters both for R and for u . The order of Sp(2t, R) and the order of USp(2.t) are both t(2.t+ 1).
+
*
We choose the generators TA in the selfrepresentation of USp(2l)
where 7':;)
and T::) are the generators in the selfrepresentation of the
su(1)group, and
(Ti:))
bd
= &d&b.
The generators TA are hermitian and
The Syrnplectic Groups
435
normalized:
Thus, the structure constants of USp(2l) are totally antisymmetric with respect to all three indices. The diagonal generators span the Cartan subalgebra of USp(2E):
H p = Ti;) x
03/&,
15
/L
5 l?.
(10.11)
Their simultaneous eigenvectors in the Lie algebra, [H,, E,] = apE,, are
where a < b and the eigenvalues (roots) are and
a
< b, respectively. The simple roots are
The first ( l  1) simple roots are shorter, d, = (rp  rp) = 1/2, and the /2 last one is longer, de = (re re) /2 = 1. The remaining positive roots are
a
b1
/= A,
b 1
e
1
p=,
p=b
e
1
Thus, the Lie algebra of USp(2t) is Ce. The largest root, which is the highest weight of the adjoint representation of Ce,is
ei
W‘ = h e 1 = 2
C rp +re =
DL=1
2 ~ 1 .
(10.13)
436
Problems and Solutions a Group Theory n
* The Chevalley bases of USp(2l) are
(10.14)
The generators in the selfrepresentation of the Sp(24 R ) group are pure imaginary, X * =  X . The explicit forms of the generators can be obtained from Eq. (10.9) by replacing 0 a with T a : r1 = iu1,
72
*
=02,
73
= ia3.
(10.15)
Namely, some generators change by a factor i so that the Sp(2!, R ) group is a noncompact Lie group. USp(2l) and Sp(24 R) have the same complex Lie algebra, but different real Lie algebras. In another viewpoint, one may change the relevant parameters of Sp(2l, R ) to be imaginary such that the generators in the corresponding irreducible representations of both USp(2l) and Sp(2e,R) are the same.
1. Prove that the determinant of R in Sp(2l, R) and the determinant of u in USp(2l) are both +l.
Solution. In the following proof for the determinant of R in Sp(2t, R ) ,we did not use the conjugate operation so that the proof is also effective for u in USp(2t). Let R ~ S p ( 2 t , R be the transformation matrix for a vector x a in the ) 21dimensional real space:
Define a pseudoproduct for two real vectors x and y :
=
p=1
c
e
(X,Y,
X D P )
*
(10.17)
Obviously, the pseudoproduct is invariant in the transformation R:
= {Rx, RY1,
*
(10.18)
The Symplectic Groups
437
The selfpseudoproduct of one vector is vanishing, {x, x } = 0. ~ For the totally antisymmetric tensor of rank 2t, Eala..aze, we can prove the following identity by Eq. (10.2)
In fact, the nonvanishing terms in the sum come from the permutations among J's ( l !terms) and the transpositions between two subscripts of any J (2' terms), and all are equal to one. Moving the column index of a 2G dimensional matrix X to be a superscript, we rewrite the determinant of X as
Permutating the order of X:, and changing the summing indices to restore the order of the superscripts, we obtain
Thus, we have det X = (2'l!)
= (2el!)I
a1
...a 2 ~
Since the righthand side is invariant in the transformation we have det (RX) = det X, det R = 1 .
In the same reason, det u = 1. Both Sp(2l, R) and USp(2l) are the simplyconnected Lie groups.
2. Count the number of the independent real parameters of R in Sp(2l, R ) and u in USp(21) directly from their definitions (10.3) and (10.6).
Solution. A (2l)dimensional real matrix contains (2t)2 real parameters. Equations (10.3) and (10.17) show that the pseudoproduct of two column
438
Problems and Solutions in Group Theory
matrices of R ESp(2l,R) is equal to
Jab:
It is an identity when a = b. It gives l ( 2 l  1) independent real constraints when a # b, so that the order of Sp(2l, R) is (21)2  l ( 2 l  1) = l ( 2 l + 1). A (2l)dimensional complex matrix contains 2( 21)2 real parameters. From ul = J u T J , we have u * =  J u J ,
Each gives 212 independent real constraints. From the unitary condition Ed U;,Udb = S a b , we have the following constraints. The constraints for a = b = p are the same as those for a = b = 1. There are l real independent . constraints :
e
V=l
e
v= I
c
When p # u, the constraints for a = p and b = u are the same as those for a = V , b = j!Z as well as those for exchanging a and b. There are l ( l  1)/2 complex independent constraints:
When p # u, the constraints for a = p, b = i are the same as those for 7 a = u , b = as well as those for exchanging a and b. There are l ( l  1)/2 complex independent constraints:
The Symplectic Groups
439
One complex constraint is equivalent to two real constraints. Altogether there are ( 2 t ) 2 e+ 2t(!  1) = 6e2  i! real constraints. The number of the independent real parameters of u in USp(2l) is 812  (612  l ) = l(2C l), which is the order of the USp(C) group.
+
+
3. Express the simple roots of USp(2C) by the vectors V, given in Eq. (8.3) for the SU(e 1) group, and then, write their CartanWeyl bases of generators in the selfrepresentation of USp(2l)
+
Solution. From the CartanWeyl bases (10.11)) we have
If we choose another set of the normalized bases HL in the Cartan subalgebra:
p=
1
H: = Ti!:+, x
03
(10.20) er+1
HeT+2,
the component (rL)uof the simple root can be calculated from the formula Hh by replacing H p with (dpp  d p ( p + l l ) when p < e and &dpe when p = e.
440
Problems and Solutions in Group Theory
Namely,
(10.21) The remaining positive roots are
b 1
p=a
10.2
Irreducible Representations of Sp(2C)
Since the generators in the corresponding irreducible representations of USp(2e) and Sp(2l, R ) ,as well as the orthonormal bases in the representation space are the same, we will only discuss the irreducible representations of Sp(2l, R ) . In the following, Sp(2l) stands for both USp(2t) and Sp(24 R) for simplicity. Let R E Sp(2l,R) be the coordinate transformation in a real (2t)dimensional space: (10.22) From Eq. (10.22) we can define the tensors for Sp(2t, R ) transformations:
*
*
Similar to the tensors of SU(N) and S O ( N ) ,the Weyl reciprocity holds for the tensors of Sp(2l,R). The tensor space can be reduced by the Young operators. In addition, there are two invariant tensors for Sp(2l,R). One is the antisymmetric tensor Jab of rank two
The Symplectic Groups
441
Tabc ...
The contraction two:
ab
cab
JabTabc...
decreases the rank of the tensor
by
ab a'b'c'
...
=
C
C ... '
Rot..
.
(z
Ja'b'Ta'b'c'
...
(10.25) Sometimes, this operation is also called the trace. The trace subspace is invariant in Sp(2&,R). In order to reduce the tensor space, we have to decompose the tensor space into the direct sum of a series of traceless tensor subspaces with decreasing rank two by two. Applying the Young operator yp to a traceless tensor space 7 ,we obtain the minimal tensor subspace 1XI
)
*
yrl7, corresponding to an irreducible representation [A] of Sp(2l,
R).
The other invariant tensor of Sp(2l, R) is the totally antisymmetric tensor € al . . . a z t of rank 2&. However, this tensor violates the new traceless condition. It is irrelevant to the reduction of the tensor space for Sp(24 R). The traceless tensor subspace denoted by a Young pattern [A] with the row number larger than & is empty. The irreducible representation of the Sp(2l,R) group is described by a Young pattern with at most l rows. The standard tensor Young tableau is still the common eigenstate of the Chevalley bases H p . The problem is that it is not necessary traceless. Fortunately, the contraction occurs only when a pair subscripts p and F appears. It is convenient to rearrange the bases @ a in the vector space:
@ ,
*
=@ ,,
@e+p = @ ,
15 P
5 e.
(10.26)
From Eq. (10.14) the nonvanishing matrix entries of the Chevalley bases in the bases @ a are
(10.27)
In the bases
@a,
1
5 a 5 2&, the standard tensor Young tableau
442
Problems and Solutions in Group Theory
Y [XI @ a l . . , a n in the representation [A] of Sp(2l, R ) corresponds to the high,
est weight if each box in its j t h row, j L, is filled with the digit j. The standard tensor Young tableau with the highest weight is a traceless tensor because all subscripts take the values less than l 1, but the trace occurs only when a pair of subscripts p and 2 l  p + 1 appears. The relation between the highest weight M and the Young pattern [A] for Sp(2l,R) is
<
+
The remaining traceless tensor bases in an irreducible representation space [A] of Sp(2L,R) can be calculated from the basis with the highest weight by the lowering operators F,. The dimension of an irreducible representation [A] of Sp(2l) can be calculated by the hook rule. In this rule, the dimension is expressed as a quotient, where the numerator and the denominator are denoted by the symbols Y p l and YiA1, respectively: (10.29) We still use the concept of the hook path (i, j ) in the Young pattern [A], which enters the Young pattern at the rightmost of the ith row, goes leftwards in the i row, turns downwards a t the j column, goes downwards in the j column, and leaves from the Young pattern at the bottom of the j column. The inverse path (i, j) is the same path as the hook path (i, j) except for the opposite direction. The number of boxes contained in the hook path (i, j) is the hook number hij of the box in the j t h column of the ith row. YiA1is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the hook number hij. Define a series of the tableaux Yktl recursively by the rule given below. YF1is a tableau of the Young pattern [A] where each box is filled with the sum of the digits which are respectively filled in the same box of each tableau Y g l in the series. The symbol Ypl means the product of the filled digits in it, so does the symbol YiA1. The tableaux Y g l are defined by the following rule:
*
(a) Y g l is a tableau of the Young pattern [A] where the box in the j t h column of the ith row is filled with the digit (2& j  i).
+
The Symplectic Groups
443
(b) Let [A(1)] = [A]. Beginning with [A(1)], we define recursively the Young pattern [A(")] by removing the first row and the first column of the Young pattern [A("')] until [A(")] contains less than two rows. (c) If [A(")] contains more than one row, define YE1to be a tableau of the Young pattern [A] where the boxes in the first (a  1) rows and in the first (a  1) columns are filled with 0, and the remaining part of the Young pattern is nothing but [A(")]. Let [A(")] have r rows. Fill the first ( r  1) boxes along the hook path (1, 1) of the Young pattern [A(")], beginning with the box on the rightmost, with the digits A?),   ., A?), box by box, and fill the first A:") boxes in each inverse path (i, 1) of the Young pattern [A(")], 2 5 i 5 r , with 1. The remaining boxes are filled with 0. If a few 1 are filled in the same box, the digits are summed. The sum of all filled digits in the pattern YAtl is zero. 4. Calculate the dimensions of the irreducible representations of the Sp(6) group denoted by the following Young patterns:
At),
(1) [4,2],
(2) [ 3 J l ,
(3) [4,417
(4) [3,3,21,
( 5 ) [4,4, 31.
Solution.
6789
0
0
0 2
(1):
d[4,2](SP(6))
=
56
t 41 1  1 5 2 21
} { %}
67811
=
=924.
678
(2):
+
0 0 2 1  1
6 7 10
(3):
d[4,4](SP(6))
=
6789 0 0 0 4 5 6 7 8 +  14 3 2 1  1 } = { 5 14321 678 567 45
67813 4 5 6 2 ) = 1274. 543 7 4321
+
(4):
d[3,3,2] (SP(6))
=
0 2 3 11 0 2  1 542 431 21
+
0 0 0 0 0 1 01
6 9 11 45 8 542 431 21
444
Problems and Solutions in Group Theory
6789 0 0 3 4 0 0 + 0 5678+111 456 211 6542 5431 321
0 0 0 0 0 2
(5):
d[4,4,3](SP(6)) =
6 7 11 13 4 5 6 10 23 4 6542 5431 321
= 2002.
5 . Calculate the orthonormal bases in the irreducible representation [1,1,0]
of the Sp(6) group by the method of the block weight diagram, and then, express the orthonormal bases by the standard tensor Young tableaux in the traceless tensor space of rank two for Sp(6).
Solution. The Lie algebra of Sp(6) is C3. The relations between the simple roots rp and the fundamental dominant weights wp are
rl = 2wl  w2,
r2
= w1+ 2w2  w3,
r3
= 2w2
+ 2w3,
w1 = rl
1 + + r3, 2
12
w2 = rl
+ 2r2 + r3,
w3 = rl
3 + 2r2 + r3. 2
The highest weight of the representation [l,1,0] of Sp(6) is M = (0, 1 , O ) . The dimension of [l, 1,0] is d[l,l,o~ (7  4)/(2  1) = 14. = The weights equivalent to the highest weight are
1 Therefore, the representation [1,1,0 contains one simple dominant weight (0, 1 , O ) and one double dominant weight (O,O,O). From the highest weight (0,1,0) we have a doublet of d 2 with (1, T, 1). From (1,i,1) we have a doublet of A with ( i , O , 1) and a doublet of d 3 1 with ( 1 , l ) i ) . From ( l , l , i )we have a doublet of .A1 with ( i , 2 , i ) and a doublet of d2with ( 2 , i , 0). From @,O, 1) we have a doublet of ds with (i, i). 2, Although (i, i)comes from two paths, it is still a single weight 2, because it is equivalent to (0, 1 , O ) . Now, we meet the double weight (O,O, 0) by applying Fl to 1(2,i,O)) and applying F 2 to l ( i , 2 , i ) ) . Letting 1(2,i,O)), I(O,O, O),) and 1(2,1,0))constitute a triplet of d1,and the other basis state I(O,O, 0)2) is a singlet of d1,we have
The Symplectic Groups
445
where we neglect the first index (0, 1,0), which denotes the representation, in the basis state I ( O , l , 0), ( m l ,m2, m3)) for simplicity. Applying ElF2 = F2El to the basis state I ( % , 2, i)),we have
Choosing the phase of the state basis (0,0,0)2)such Thus, u = that b is real positive, we have b = = Applying E2F2 = F2E2 H2 to 1(0,0,0)1)we have
0.
+
d
m
m.
such that c is real positive, Choosing the phase of the basis state 1(1,2,1)) we have c = and d = The remaining basis states and the matrix entries of the lowering operators F, can be calculated similarly. The results are given in Fig. 10.1, where only those matrix entries of F, which are not equal to 1 are indicated.
m.
II
Fig. 10.1
The block weight diagram and the state bases for [ l ,1,0] of Sp(6).
446
Problems and Solutions in Group Theory
Now, we expand the orthonormal basis states in the standard tensor Young tableaux. The Young tableau Young tableau is expanded as follows:
y[ljljO] is
. A standard tensor
There is only one trace tensor in the tensor space of rank two projected by y[ljl*O]. It belongs to the representation [O,O,O] of Sp(6):
Beginning with the highest weight state, we calculate the expansions of the basis states in terms of Eq. (10.27) and the block weight diagram. Due to the WignerEckart theorem, the basis state belonging to the representation [1,1,0] of Sp(6) is orthogonal to the basis state belonging to [O,O,O], so it is a traceless tensor.
m{B 1

+2
The Symplectic Groups
44 7
U
l(o,i,o)) = p21(i,, i ) )= i
6. Calculate the orthonormal bases in the irreducible representation [I,1,1]
of the Sp(6) group by the method of the block weight diagram, and then, express the orthonormal bases by the standard tensor Young tableaux in the traceless tensor space of rank two for Sp(6). Solution. The Lie algebra of Sp(6) is C3. The relations between the simple roots rp and the fundamental dominant weights wp was given in Problem 5. The dimension of [l,1,1]is
131
= 14.
The highest weight of the representation [l,1,1]of Sp(6) is M = (O,O, 1). The weights equivalent to the dominant weight (O,O, 1) are
From the highest weight state (0, 0 , l ) we have a doublet of d 3 with (0,2, i), From ( 0 , 2 , i ) we have a triplet of d 2 with (1,0,0) and (2,2,1). Both the matrix entries of F2are A. The weight (1,0,0) is a single dominant weight. The weights equivalent to the dominant weight (1,0,0) are
448
Problems and Solutions in Group Theory
Therefore, all weights in the representation [I,1,1]are single. It is easy to draw the block weight diagram for the representation [l,1,1]. In Fig. 10.2 we give the block weight diagram for the representation [l,1,1] where only those matrix entries of Fp which are not equal to 1 are indicated.
11,090
I
I
IIJZ
Fig. 10.2 The block weight diagram and the state bases for [l,1,1] of Sp(6).
Now, we expand the orthonormal basis states in the standard tensor . Young tableaux. The Young tableau y[lilsllis Young tableau is
. The standard tensor
The trace tensors in the tensor space of rank three projected by Y['v1i1] belong to the representation [l, 0 of Sp(6): 0, 1
The Symplectic Groups
449
Beginning with the highest weight state, we calculate the expansions of the basis states in terms of Eq. (10.27) and the block weight diagram. Due to the WignerEckart theorem, the basis state belonging to the representation [l,1,1]of Sp(6) is orthogonal to the basis state belonging to [l,O,O], so it is a traceless tensor.
450
Problems and Solutions in Group Theory
m{i

m { I [}
+i}
The Symplectic Groups
45 1
7. Calculate the ClebschGordan series for the reduction of the direct product representation [l,1,0] x [l,l , O ] of the Sp(6) group and the highest weight states for the representations contained in the series by the method of the standard tensor Young tableau.
Solution. There are a few methods to calculate the ClebschGordan series of a reducible representation. The method of the standard tensor Young tableau is one of them. The main point in the calculation is to count the multiplicities of the dominant weights in the relevant representations. The difficulty in Sp(2l) group, as well as S O ( N ) group, is how to remove the trace tensor subspace. In fact, the dimension of the tensor space projected by a Young operator [A] is nothing but the dimension d[A](SU(2l)) of SU(2l) group. Subtracting the dimension 4x1 (Sp(2t)) of Sp(2l) group, we obtain the dimension of the trace tensor subspace. Conversely, we can calculate the multiplicities of the dominant weights in the representation [A] of Sp(2l) by counting the multiplicities both in the tensor space projected by the Young operator Y[’] and in the trace tensor subspace. As far as the highest weight state is concerned, the condition that each raising operator E p annihilates the highest weight state is useful to calculate the basis state. First, we count the multiplicities of the dominant weights in the space of the direct product representation [l,l , O ] x [I,1,0] of the Sp(6) group. Since the block weight diagram for the representation [I, l , O ] of Sp(6) was given in Fig. 10.1 of Problem 5, it is easy to count the multiplicities of the dominant weights. In the direct product space, there are one basis state with the dominant weight (0,2,0), two basis states with the dominant weight ( l , O , l), four basis states with the dominant weight (2,0,0), eight basis states with the dominant weight (0, 1,0), and 16 basis states with the dominant weight (0, 0,O). Second, since there is one basis state with the dominant weight (0,2,0) in the direct product space, one representation [2,2,0] is contained in the ClebschGordan series. Its highest weight state is
The decomposition of the tensor subspace projected by [2,2,0] +
y[2)2)o] is
[a, 2,0] CB [I,1 , O ] CB [O,O, 01,
105 = 90
+ 14 + 1.
The tensor subspace contains the basis states with the dominant weights
452
Problems and Solutions in Group Theory
.as follows Weight
No. State bases
We have to subtract the basis states in the representations [I, l , 0 and 1 [O,O, 01: one state with (0, 1,O) and three states with (O,O, 0). Third, since the difference between the multiplicities of the dominant weight (1,0,1) in the direct product space and in the representation space [2,2,0] is 2  1 = 1, one representation [2,1,1] is contained in the ClebschGordan series. Its highest weight state is
The coefficients are calculated by the condition that each raising operator E, annihilates the highest weight state and the normalization condition. The decomposition of the tensor subspace projected by Y[’i1r1] is
[Z, 1,1]
+
[2,1,1]@ [2,0,0]@ [l,1,0],
105 = 70
+ 21 + 14.
The tensor subspace contains the basis states with the dominant weights
The Symplectic Groups
453
as follows.
No. State bases
Weight
1
2
5
9
We have to subtract the basis states in the representations [2,0,0] and [I,1,0]: one state with (2,0,0), two states with (0,1,0) and five states with (0, 0,O). Note that the tensor space for [2,0,0] is traceless and contains the basis states with the dominant weights as follows.
Weight
(2)O)O) :
No. State bases
1
1
(0,1,0):
I1I 1I) I 1 1 2 1,
Fourth, since the difference between the multiplicities of the dominant weight (2,0,0) in the direct product space and in the representation spaces [2,2,0] and [2,1,1) is 4  2  (2  1) = 1, one representation [2,0,0] is
454
Problems and Solutions in Group Theory
contained in the ClebschGordan series. Its highest weight state is
The coefficients are calculated by the condition that each raising operator E, annihilates the highest weight state and the normalization condition. The basis states with the dominant weights contained in the traceless tensor space of [2,0,O] was given. Fifth, since the difference between the multiplicities of the dominant weight (0,1,0) in the direct product space and in the representation spaces [2,2,0], [2,1,1], and [2,0,0] is 8(41)(52)1 = 1, onerepresentation [I,1,0] is contained in the ClebschGordan series. Its highest weight state is
The coefficients are calculated by the condition that each raising operator E, annihilates the highest weight state and the normalization condition. We have known from Fig. 10.1 of Problem 5 that the traceless tensor space for [l,1,0] contains one basis state with the dominant weights ( O , l , 0) and two basis states with the dominant weight (O,O,O). Finally, counting the multiplicities of the dominant weight (0, 0,O) in the direct product space and in the representation spaces [2,2,0], [2,1,1], [2,0,0], and [I,1,0], we have 16  (9  3)  (9  5)  3  2 = 1. One representation [0, 0, 0 is contained in the ClebschGordan series. The 1 expression for its highest weight state is quite long. We write it in the form of the basis states instead of the standard tensor Young tableaux. The
The Symplectic Groups
455
latter form can be calculated from Fig. 10.1.
In summary, the dominant weight diagram for the direct product representation [1,1,0] ´ [1,1,0] is given in Fig. 10.3.
1 4 1~=196= 4
90
+
70
+
21
+
14
+
1
Fig. 10.3 The dominant weight diagram for [l,1,0] x [l,1,0] of Sp(6).
In the permutation of two factors in the direct product representation space [l, , O ] x [l, , O ] , the basis states in the representations [2,2, 01, [l,1,0] l l and [O,O, 0 are symmetric, and the representations [2,1,1] and [2,0,0] are 1 antisymmetric.
This page intentionally left blank
Bibliography
Adams, B. G., Cizek, J. and Paldus, J. (1987). Lie Algebraic Methods and Their Applications to Simple Quantum Systems, Advances in Quantum Chemistry, 19 Academic Press, New York. Berenson, R. and Birman, J. L.( 1975). ClebschGordan coefficients for crystal space group, J. Math. Phys 16,p. 227. Biedenharn, L. C. and Louck, J. D. (1981). Angular Momentum in Quantum Physics, Theory and Application, Encyclopedia of Mathematics and its Application, 8, Ed. G. C. Rota, AddisonWesley, Massachusetts. Boerner, H. (1963). Representations of Groups, NorthHolland, Amsterdam. Bourbaki, N. (1989). Elements of Mathematics, Lie Groups and Lie Algebras, SpringerVerlag, New York. Bradley, C. J. and Cracknell, A. P. (1972). The Mathematical Theory of Symmetry in Solids, Oxford: Clarendon Press. Bremner, M. R., Moody, R. V. and Patera, J. (1985). Tables of Dominant Weight Multiplicities for Representations of Simple Lie Algebras, Pure and Applied Mathematics, A Series of Monographs and Textbooks 90, Marcel Dekker, New York. Burns, G. and Glazer, A. M. (1978). Space Groups for Solid State Scientists, Academic Press, New York. Chen, J. Q. (1989). Group Representation Theory for Physicists, World Scientific, Singapore. The second version: Chen, J.Q., Ping, J. L. and Wang, F. (2002). Group Representation Theory for Physicists, 2nd edition, World Scientific, Singapore. Chen, J. Q. and Ping, J. L. (1997). Algebraic expressions for irreducible bases of icosahedral group, J . Math. Phys 38, 387. de Swart, J. J. (1963). The octet model and its ClebschGordan coefficients, Rev. Mod. Phys. 35, 916. Deng Y. F. and Yang, C. N. (1992). Eigenvalues and eigenfunctions of the Hiickel Hamiltonian for carbon60, Phys. Lett. A 170,116. Dirac, P. A. M. (1958). The Principle of Quantum Mechanics, Clarendon Press, Oxford. Dong, S. H., Hou, X. W. and Ma, Z. Q. (1998). Irreducible bases and correlations
457
458
Problems and Solutions in Group Theory
of spin states for double point groups, Inter. J. Theor. Phys. 37, 841. Dong, S.H., Xie, M. and Ma, Z. Q. (1998). Irreducible bases in icosahedral group space, Inter. J. Theor. Phys 37, 2135. Dong, S. H., Hou, X. W. and Ma, Z. Q. (2001). Correlations of spin states for icosahedral double group, Inter. J . Theor. Phys 40,569. Duan, B., Gu, X. Y. and Ma, Z. Q. (2001). Precise calculation for energy levels of a helium atom in P states, Phys. Lett. A 283, 229. Duan, B., Gu, X. Y., and Ma, Z. Q. (2002). Numerical calculation of energies of some excited states in a helium atom, Eur. Phys. J . D 19,9. Eckart, C. (1934). The kinetic energy of polyatomic molecules, Phys. Rev. 46, 383. Edmonds, A. R. (1957). Augular Momentum in Quantum Mechanics, Princeton University Press, Princeton. F’ronsdal, C. (1963). Group theory and applications to particle physics, 1962, Brandies Lectures, 1,427. Ed. K.W. Ford, Benjamin, New York. Gel’fand, I. M., Minlos, R. A. and Shapiro, Z. Ya. (1963). Representations of the Rotation and Lorentz Groups and Their Applications, translated from Russian by G. Cummins and T. Boddington, Pergamon Press, New York. Gel’fand, I. M. and Zetlin, M. L. (1950). Matrix elements for the unitary groups, Dokl. Akad. Nauk 71,825. GellMann, M. and Ne’eman, Y. (1964). The Eightfold Way, Benjamin, New York. Georgi, H. (1964). Lie Algebras in Particle Physics, Benjamin, New York. Gilmore, R. (1974). Lie Groups, Lie Algebras and Some of Their Applications, Wiley, New York. Gu, X. Y., Duan, B. and Ma, Z. Q. (2001). Conservation of angular momentum and separation of global rotation in a quantum Nbody system, Phys. Lett. A 281,168. Gu, X. Y., Duan, B. and Ma, Z. Q. (2001). Independent eigenstates of angular momentum in a quantum Nbody system, Phys. Rev. A 64,042108(114). Gu, X. Y., Duan, B. and Ma, Z. Q. (2001). Quantum threebody system in D dimensions, J . Math. Phys. 43,2895. Gu, X. Y., Ma, Z. Q. and Duan, B. (2003). Interdimensional degeneracies for a quantum threebody system in D dimensions, Phys. Lett. A 307,55. Gu, X.Y., Ma, Z. Q. and Sun, J. Q. (2003). Quantum fourbody system in D dimensions, J . Math. Phys. 44, 3763. Gu, X. Y., Ma, Z. Q. and Sun, J. Q. (2003). Interdimensional degeneracies in a quantum isolated fourbody system, Phys. Lett. A 314,156. Gu, X. Y., Ma, Z. Q. and Sun, J. Q. (2003). Interdimensional degeneracies in a quantum Nbody system, Europhys. Lett. 64,586. Gu, X. Y., Ma, Z. Q. and Dong, S. H. (2002). Exact solutions to the Dirac equation for a Coulomb potential in D 1 dimensions, Inter. J. Mod. Phys. E 11, 335. Gu, X. Y., Ma, Z. Q. and Dong, S. H. (2003). The Levinson theorem for the Dirac equation in D 1 dimensions, Phys. Rev. A 67,062715(112). Hamermesh, M. (1962). Group Theory and its Application to Physical Problems, AddisonWesley, Massachusetts.
+
+
Bibliography
459
Heine, V. (1960). Group Theory in Quantum Mechanics, Pergamon Press, London. Hirschfelder, J. 0. and Wigner, E. P. (1935). Separation of rotational coordinates from the Schrodinger equation for N particles, Proc. Nutl. Acud. Sci. U,S.A. 21,113. Hou, X. W., Xie, M., Dong, S. H. and Ma, Z. Q. (1998). Overtone spectra and intensities of tetrahedral molecules in bosonrealization models, Ann. Phys. (N. Y.) 263,340. Itzykson, C. and Nauenberg, M. (1966). Unitary groups: Representations and decompositions, Rev. Mod. Phys. 38, 95. Joshi, A. W. (1977). Elements of Group Theory for Physicists, Wiley. Koster, G. F. (1957). Space Groups and Their Representations in Solid State Physics, Eds. F. Seitz and D. Turnbull, Academic Press, New York, 5 , 174. Kovalev, 0. V. (1961). Irreducible Representations of Space Groups, translated from Russian by A. M. Gross, Gordon & Breach. Lipkin, H. J. (1965). Lie Groups for Pedestrians, NorthHolland, Amsterdam. Littlewood, D. E. (1958). The Theory of Group Characters, Oxford University Press, Oxford. Liu, F., Ping, J. L. and Chen, J. Q. (1990). Application of the eigenfunction method to the icosahedral group, J. Math. Phys. 31, 1065. Ma, 2. Q., Group Theory for Physicists, World Scientific, to be published. Ma, 2. Q. (1993). YangBaxter Equation and Quantum Enveloping Algebras, World Scientific, Singapore. Ma, Z. Q., Hou, X. W. and Xie, M. (1996). Bosonrealization model for the vibrational spectra of tetrahedral molecules, Phys. Rev. A 53, 2173. Miller, W., Jr. (1972). Symmetry Groups and Their Applications, Academic Press, New York. Racah, G. (1951). Group Theory and Spectroscopy, Lecture Notes in Princeton. Roman, P. (1964). Theory of Elementary Particles, NorthHolland, Amsterdam. Rose, M. E. (1957). Elementary Theory of Angular Momentum, Wiley, New York. Salam, A. (1963). The Formalism of Lie Groups, in Theoretical Physics, Director: A. Salam, International Atomic Energy Agency, Vienna, 173. Schiff, L. I. (1968). Quantum Mechanics, Third Edition, McGrawHill, New York. Serre, J . P. (1965). Lie Algebras and Lie Groups, Benjamin, New York. Tinkham, M. (1964). Group Theory and Quantum Mechanics, McGrawHill, New York. Tung, W. K. (1985). Group Theory in Physics, World Scientific, Singapore. Weyl, H. (1931). The Theory of Groups and Quantum Mechanics, translated from German by H. P. Robertson, Dover Publications. Weyl, H. (1946). The Classical Groups, Princeton University Press, Princeton. Wigner, E. P. (1959). Group Theory and its Applications to the Quantum Mechanics of Atomic Spectra, Academic Press, New York. Wybourne, B. G. (1974). Classical Groups for Physicists, Wiley, New York. Yamanouchi, T. (1937). On the construction of unitary irreducible representation of the symmetric group, Proc. Phys. Math. SOC.J p n 19, 436.
This page intentionally left blank
Index
l?N matrix group, 403 ya matrix, 403
anticommutator, 319 associative law, 27 basis, 7 orthogonal, 212 standard, 211 block weight diagram, 282, 285, 289, 292, 298, 303, 309, 336, 345, 348, 355, 378, 386, 394, 412, 445, 448 Bravais lattices, 174, 180 Cartan criteria, 270 Cartan matrix, 272 Cartan subalgebra, 271, 376, 435 CartanWeyl bases, 271, 318, 439 character, 43 graphic method, 213 character table, 48 Chevalley bases, 280, 321, 378, 409, 436, 441 class, 30, 48, 54, 139, 422, 424 selfreciprocal, 54 mutually reciprocal, 30 selfreciprocal, 30 ClebschGordan coefficients, 80, 101, 106, 149, 154, 215, 243 ClebschGordan series, 80, 240, 299 composition functions, 140 coset, 29
461
left, 29 right, 29 crystal, 173 crystal lattice, 173 basis vector of the reciprocal, 175 besis vector, 173 vector, 173 crystal system, 178 crystallographic point group, 174 cycle, 193 cycle structure, 194 determinant, 1 dominant weight diagram, 299, 301, 356 doublevector form, 175 Dynkin diagram, 272 eigenequation, 1 eigenvalue, 1 eigenvector, 1 element, 27 conjugate, 30 equivalent theorem, 404 Euler angles, 123, 125, 127 filling parity, 214 Fock condition, 199, 323 F’robenius theorem, 67, 239 Gelfand bases, 337 generator, 27, 269
462
Problems and Solutions in Group Theory
gliding plane, 178 gliding vector, 177, 178 group, 27 C4, 32 c8, 35 D2n+1, 68 D2n, 68 D4, 33 D N , 32 v4, 32 direct product, 55 homomorphism, 33 I, 73, 75, 106, 121, 134 0, 59, 70, 85, 101 T, 81 Abelian, 27 c 0 37 1 , C2n, 38 C4h, 35 c6, 32 C N , 65, 68, 69 Dn, 38 S O ( N ) ,375 S0(3), 115 SU(N), 317 SU(2), 117 Sp(2C, R), 433 USp(2C), 434 v2, 34 covering, 117, 140 direct product, 51 improper point, 51 proper point, 51 quaternion, 35 quotient, 31 T, 40 group space, 52, 140, 269 height, 282 homomorphism, 121 kernel of, 33 hook number, 198 hook rule, 198, 325, 380, 405, 442 horizontal operator, 199 ideal, 205
idempotent, 205 equivalent, 205 orthogonal, 205 primitive, 205 identity, 27 infinitesimal element, 141 international notation, 178 inverse, 27 irreducible tensor operator, 146 isomorphic, 27 Jordan forms, 18 Killing form, 270, 320 Lie algebra, 270 compact, 270 rank, 271 simisimple, 270 classical, 272 exceptional, 272 Lie group, 115, 140, 269 compact, 140, 269 order, 269 semisimple, 269 simple, 269 Lie product, 270 Lie Theorem, 141, 142 LittlewoodRichardson rule, 239, 326, 356, 364, 370, 402 Lorentz group, 415 homogeneous, 418 proper, 418 lowering operator, 280 matrix, 2 hermitian, 2, 5 negative definite, 2, 270 positive definite, 2 positive semidefinite 2 real orthogonal, 2, 5 real symmetric, 2, 6 unitary, 2, 5 multiplication rule, 27 multiplication table, 27
Index
463
multiplicity, 49, 53, 67, 77, 80, 97, 280, 282, 289, 299, 304, 308, 313, 355, 356, 451, 454 order, 140 of the element, 29 of the group, 27 Pauli matrices, 3, 33 period, 29 permutation, 193 even, 194 horizontal, 199 odd, 194 vertical, 199 permutation parity, 194 planar weight diagram, 283, 288, 289, 292, 298, 306, 340, 345, 348, 355, 360, 372 projection operator, 81 raising operator, 280 rank, 2, 142, 146, 317, 321, 326, 327, 376, 379, 380, 440, 441 rearrangement theorem, 27 reduced matrix entry, 147, 149 regular application, 214 representation, 43 regular, 52, 81, 85 subduced, 130 adjoint, 142, 283 conjugate, 43 equivalent, 47 faithful, 43 identical, 43 induced, 66, 237 inner product, 237 irreducible, 48, 123, 211, 279, 440 outer product, 238 real orthogonal, 43 reduced, 47 reducible, 47 subduced, 66, 237, 362 unitary, 43 unitary of infinite dimension, 166 root, 271
positive, 271 simple, 271, 319, 435 root chain, 271 rotation, 45, 115 improper, 51 proper, 51 Schur theorem, 47, 143, 245, 321 screw axis, 177 secular equation, 1 Serre relations, 280 similarity transformation, 8 space group, 174, 186 symmorphic, 174 spherical harmonic function, 399 spin tensor, 405 spinor, 405 spinor operators, 407 spinor representation, 404 structure constants, 141, 269, 319 subgroup, 29 cyclic, 29 index, 30 invariant, 30 normal, 30 symmetric center, 177 symmetric operation, 173 symmetric plane, 177 symmetric straight line, 177 symmetry transformation, 44 symplectic group, 433 real, 433 unitary, 434 tabular method, 212 trace, 1 Transformation Operators for a Scalar Function, 43 translation group, 173 transposition, 194 vertical operator, 199 weight, 279 dominant, 282, 336 equivalent, 280, 281
464
Problems and Solutions an Group Theory
fundamental dominant, 281 highest, 282, 337, 379, 442 multiple, 279 single, 279 Weyl orbit, 280 Weyl orbital size, 280 Weyl reciprocity, 323, 377, 440 Weyl reflection, 280 WignerEckart theorem, 51, 147 Young operator, 323, 377, 440 Young pattern, 198, 211, 323, 379, 441, 442 associated, 214 Young tableau, 198 standard, 198 standard tensor, 324, 337, 362, 441 tensor, 323
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.