Representations and Characters of Groups Now in its second edition, this text provides a modern introduction to the representation theory of ®nite groups. The authors have revised the popular ®rst edition and added a considerable amount of new material. The theory is developed in terms of modules, since this is appropriate for more advanced work, but considerable emphasis is placed upon constructing characters. The character tables of many groups are given, including all groups of order less than 32, and all simple groups of order less than 1000. Among the applications covered are Burnside's pa q b theorem, the use of character theory in studying subgroup structure and permutation groups, and a description of how to use representation theory to investigate molecular vibration. Each chapter is accompanied by a variety of exercises, and full solutions to all the exercises are provided at the end of the book. This will be ideal as a text for a course in representation theory, and in view of the applications of the subject, will be of interest to mathematicians, chemists and physicists alike.

R E P R E S E N TAT ION S AN D C H AR ACT E R S OF G ROU P S
G O R D O N JA M E S a n d M A RT I N L I E B E C K
Department of Mathematics, Imperial College, London

Second Edition

PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING) FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE The Pitt Building, Trumpington Street, Cambridge CB2 IRP 40 West 20th Street, New York, NY 10011-4211, USA 477 Williamstown Road, Port Melbourne, VIC 3207, Australia http://www.cambridge.org © Cambridge University Press 1993, 2001 This edition © Cambridge University Press (Virtual Publishing) 2003 First published in printed format 1993 Second edition 2001 A catalogue record for the original printed book is available from the British Library and from the Library of Congress Original ISBN 0 521 81205 4 hardback Original ISBN 0 521 00392 X paperback

ISBN 0 511 01700 6 virtual (netLibrary Edition)

Contents

Preface 1 Groups and homomorphisms 2 Vector spaces and linear transformations 3 Group representations 4 FG-modules 5 FG-submodules and reducibility 6 Group algebras 7 FG-homomorphisms 8 Maschke's Theorem 9 Schur's Lemma 10 Irreducible modules and the group algebra 11 More on the group algebra 12 Conjugacy classes 13 Characters 14 Inner products of characters 15 The number of irreducible characters 16 Character tables and orthogonality relations 17 Normal subgroups and lifted characters 18 Some elementary character tables 19 Tensor products 20 Restriction to a subgroup 21 Induced modules and characters 22 Algebraic integers 23 Real representations 24 Summary of properties of character tables 25 Characters of groups of order pq 26 Characters of some p-groups 27 Character table of the simple group of order 168 v

page vii 1 14 30 38 49 53 61 70 78 89 95 104 117 133 152 159 168 179 188 210 224 244 263 283 288 298 311

vi

Representations and characters of groups

28 Character table of GL(2, q) 322 29 Permutations and characters 337 30 Applications to group theory 348 31 Burnside's Theorem 361 32 An application of representation theory to molecular vibration 367 Solutions to exercises 397 Bibliography 454 Index 455

Preface

We have attempted in this book to provide a leisurely introduction to the representation theory of groups. But why should this subject interest you? Representation theory is concerned with the ways of writing a group as a group of matrices. Not only is the theory beautiful in its own right, but it also provides one of the keys to a proper understanding of ®nite groups. For example, it is often vital to have a concrete description of a particular group; this is achieved by ®nding a representation of the group as a group of matrices. Moreover, by studying the different representations of the group, it is possible to prove results which lie outside the framework of representation theory. One simple example: all groups of order p2 (where p is a prime number) are abelian; this can be shown quickly using only group theory, but it is also a consequence of basic results about representations. More generally, all groups of order pa q b ( p and q primes) are soluble; this again is a statement purely about groups, but the best proof, due to Burnside, is an outstanding example of the use of representation theory. In fact, the range of applications of the theory extends far beyond the boundaries of pure mathematics, and includes theoretical physics and chemistry ± we describe one such application in the last chapter. The book is suitable for students who have taken ®rst undergraduate courses involving group theory and linear algebra. We have included two preliminary chapters which cover the necessary background material. The basic theory of representations is developed in Chapters 3±23, and our methods concentrate upon the use of modules; although this accords with the more modern style of algebra, in several instances our proofs differ from those found in other textbooks. The main results are elegant and surprising, but at ®rst sight they sometimes have an air of mystery vii

viii

Representations and characters of groups

about them; we have chosen the approach which we believe to be the most transparent. We also emphasize the practical aspects of the subject, and the text is illustrated with a wealth of examples. A feature of the book is the wide variety of groups which we investigate in detail. By the end of Chapter 28, we have presented the character tables of all groups of order less than 32, of all p-groups of order at most p4 , and of all the simple groups of order less than 1000. Every chapter is accompanied by a set of Exercises, and the solutions to all of these are provided at the end of the book. We would like to thank Dr Hans Liebeck for his careful reading of our manuscript and the many helpful suggestions which he made.

In this second edition, we have included two new chapters; one (Chapter 28) deals with the character tables of an in®nite series of groups, and the other (Chapter 29) covers aspects of the representation theory of permutation groups. We have also added a considerable amount of new material to Chapters 20, 23 and 30, and made minor amendments elsewhere.

Preface to Second Edition

together with a rule for combining any two elements g. Groups A group consists of a set G. h. (3) for all g in G. which we shall use extensively to illustrate the later theory. this rule must satisfy the following axioms: (1) for all g. ( gh)k ˆ g(hk).1 Groups and homomorphisms This book is devoted to the study of an aspect of group theory. so we begin with a resume of facts about groups. eg ˆ ge ˆ g. written gh. h of G to form another element of G. we introduce several examples. One or two results which we shall use only infrequently are demoted to the exercises at the end of the chapter ± you can refer to the solutions if necessary. and any book on basic group theory will supply you with further details. such as dihedral groups and symmetric groups. In addition. most of which you should   know already. there exists an element gÀ1 in G such that gg À1 ˆ g À1 g ˆ eX We refer to the rule for combining elements of G as the product operation on G. An elementary course on abstract algebra would normally cover all the material in the chapter. 1 . (2) there exists an element e in G such that for all g in G. k in G.

then we call G a ®nite group. and an ˆ 1. The set of nth roots of unity in C. Let A be a corner of the polygon. under addition. for two symmetries f and g. The product of an element g with itself. a. g0 ˆ 1. then Cn ˆ f1. (2) The set Z of all integers. These 2n rotations and re¯ections form a group under the product operation of composition (that is. similarly 3 g ˆ g 2 g. r1 . If a ˆ e2ðia n . . . and consider the rotation and re¯ection symmetries of a regular n-sided polygon. . is a group. Usually we write 1. and gÀ1 is an inverse of g in axiom (3).1 Examples (1) Let n be a positive integer.2 Representations and characters of groups Axiom (1) states that the product operation is associative. then do g'). is a group of order n. the product fg means `®rst do f. a nÀ1 g. the number of elements in G is called the order of G. . and is written D2 n. Also. There are also n re¯ection symmetries: these are re¯ections in the n lines passing through O and a corner or the mid-point of a side of the polygon. This group is called the dihedral group of order 2n. r nÀ1 where r k is the (clockwise) rotation about the centre O through an angle 2ðkan. with the usual multiplication of complex numbers. There are n rotation symmetries: these are r0 . a2 . and denote by C the set of all complex numbers. 1. and is written |G|. and so on. (3) Let n be an integer with n > 3. gÀ2 ˆ ( gÀ1 )2 . for the identity element of G. the element e in axiom (2) is an identity element of G. is written g 2 . Write b for the re¯ection in the . gg. rather than e. It is written as Cn and is called the cyclic group of order n. and that every g in G has just one inverse. If the number of elements in G is ®nite. X X X . It is elementary to see that G has just one identity element.

This group is called the general linear group of degree n over F. b: an ˆ 1. The order of Sn is n!. F) is of course the identity matrix. we have ba j ˆ aÀ j b (using the relation ba ˆ aÀ1 b). and write a for the rotation r1 . forms a group. bÀ1 ab ˆ aÀ1 iX (4) For n a positive integer. It is an in®nite group. . ab. and is written Sn . A group G is said to be abelian if gh ˆ hg for all g and h in G. F). The set of all invertible n 3 n matrices with entries in F. Check that an ˆ 1. . and is denoted by GL(n. a2 b. . While Cn and Z are abelian. We use the notation H < G to indicate that H is a subgroup of G. Subgroups Let G be a group. a2 . a nÀ1 bX Thus all elements of D2 n are products of powers of a and b ± that is.Groups and homomorphisms 3 line through O and A. b2 ˆ 1 and bÀ1 ab ˆ aÀ1 X These relations determine the product of any two elements of the group. D2 n is generated by a and b. For example. . X X X . The identity of GL(n. X X X . b2 ˆ 1. is a group. which leaves the polygon ®xed). a nÀ1 (where 1 denotes the identity. Then the n rotations are 1. which we denote by In or just I. (5) Let F be either R (the set of real numbers) or C (the set of complex numbers). and hence (a i b)(a j b) ˆ a i ba j b ˆ a i aÀ j bb ˆ a iÀ j X We summarize all this in the presentation D2 n ˆ ha. It is called the symmetric group of degree n. under matrix multiplication. and the n re¯ections are b. under the product operation of composition. a. n}. the set of all permutations of {1. 2. most of the other examples given above are non-abelian groups. A subset H of G is said to be a subgroup if H is itself a group under the product operation inherited from G. .

Bˆ X 0 Ài À1 0 . b P G. De®ne H to be the subset of G consisting of all elements which are products of powers of a and b ± that is. The subset h gi ˆ f g n : n P Zg is a subgroup of G. such as general linear or symmetric groups. k P H then hkÀ1 P H. We illustrate the construction in the next example. we call H the subgroup generated by a and b. and (2) if h. (4) Let G ˆ GL(2. let r be the least positive integer such that g r ˆ 1. The groups Cn and Z in Examples 1. where ik .5 below. C). then r is equal to the number of elements in k gl ± indeed. the group of invertible 2 3 2 matrices with entries in C. and write H ˆ ha. 1. g 2 . In this case. If gn ˆ 1 for some n > 1. g. we can similarly de®ne hSi. (2) Let G be a group and g P G. both {1} and G are subgroups of G. g rÀ1 gX We call r the order of the element g.2 Examples (1) For every group G. and again in Example 1. X X X . biX Given any ®nite set S of elements of G. called the cyclic subgroup generated by g. the subgroup of G generated by S. (3) Let G be a group and let a. Then H is a subgroup of G. h gi ˆ f1. then k gl is ®nite. If G ˆ k gl for some g P G then we call G a cyclic group. and let     i 0 0 1 Aˆ . This construction gives a powerful method of ®nding new groups as subgroups of given groups.1 are cyclic.4 Representations and characters of groups It is easy to see that a subset H of a group G is a subgroup if and only if the following two conditions hold: (1) 1 P H. jk P Z for 1 < k < n. all elements of the form a i1 b j1 a i2 b j2 X X Xa i n b j n for some n.

h9) ˆ ( gg9. With this product operation. It can be shown that either all such expressions for g have an even number of transpositions. h)( g9. .Groups and homomorphisms Put H ˆ kA. B: A4 ˆ I. G 3 H is a group. Direct products We describe a construction which produces a new group from given ones. we see that every element of H has the form A i B j for some integers i. the subgroup of G generated by A and B. The group H is called the quaternion group of order 8. . Let G and H be groups. h9 P H. Bl. Check that A4 ˆ I. . BÀ1 AB ˆ AÀ1 X 5 Using the third relation. . h): g P G and h P HgX De®ne a product operation on G 3 H by ( g. j. or they all have an odd number of transpositions. accordingly. so we have the presentation Q8 ˆ hA. A2 ˆ B2 . Every permutation g in Sn can be expressed as a product of transpositions. in fact j Hj ˆ 8. 2. . and is written Q8. The subset An ˆ f g P Sn : g is an even permutationg is a subgroup of Sn . The above three relations determine the product of any two elements of Q8 . n and ®xes the other n À 2 numbers. Hence H has at most eight elements. hh9) for all g. Since the matrices Ai B j (0 < i < 3. called the direct product of G and H. we can take 0 < i < 3 and 0 < j < 1. A2 ˆ B2 . we call g an even or an odd permutation. and using the ®rst two relations. 0 < j < 1) are all distinct. g9 P G and all h. BÀ1 AB ˆ AÀ1 iX (5) A transposition in the symmetric group Sn is a permutation which interchanges two of the numbers 1. and consider G 3 H ˆ f( g. called the alternating group of degree n.

g r g9 )X 1 r 1 r If all the groups Gi are ®nite. X X X . 1. A function W from G to H is invertible if and only if it is both injective (that is. . Functions A function from one set G to another set H is a rule which assigns a unique element of H to each element of G. the image of g under a function W is written as gW. A function W: G 3 H is invertible if there is a function ö: H 3 G such that for all g P G. g r )( g9 . . . where g P G and h P H. and is written as WÀ1 . g2 P G implies that g1 ˆ g2 ) and surjective (that is. g r ): g i P Gi for 1 < i < rg. h P H. we mean that h ˆ gW. for every h P H there exists g P G such that gW ˆ h). By an expression W: g 3 h. We often indicate that W is a function from G to H by the notation W: G 3 H. . then a homomorphism from G to H is a function W: G 3 H which satis®es ( g1 g 2 )W ˆ ( g 1 W)( g 2 W) for all g 1 . . we generally apply functions on the right ± that is. g9 ) ˆ ( g 1 g9 . then G1 3 . . |Gr |. . . 3 Gr is also ®nite. . An invertible function is also called a bijection. of order |G1 | . then the direct product G1 3 X X X 3 Gr is f( g 1 . In this book. g 2 P GX . g1 W ˆ g2 W for g1 . Gr are groups. If G and H are groups. Homomorphisms Given groups G and H. not as W g. those functions from G to H which `preserve the group structure' ± the so-called homomorphisms ± are of particular importance. X X X .6 Representations and characters of groups More generally. 3 C2 (r factors) has order 2 r and all its nonidentity elements have order 2. ( gW)ö ˆ g and (hö)W ˆ hX Then ö is called the inverse of W.3 Example The group C2 3 . if G1 . with product operation de®ned by ( g1 . . X X X . X X X .

(ar bs at bu )W ˆ (a i b j )W ˆ x i y j ˆ x r y s x t y u ˆ (ar bs )W . Moreover. then G and H are said to be isomorphic. we can also deduce that xr ys xt yu ˆ x i y j X Therefore.4 Example Let G ˆ D2 n ˆ ka. Let H be any group. 1. WÀ1 is an isomorphism from H to G. Then ar bs at bu ˆ a i b j for some i. 0 < t < n À 1. (at bu )W. The following example displays a technique which can often be used to prove that certain functions are homomorphisms. 0 < j < 1) is a homomorphism. i and j are determined by repeatedly using the relations an ˆ b2 ˆ 1. and so W is a homomorphism. y ˆ (2 5)(3 4)X .4 in action. so H  G. bÀ1 ab ˆ aÀ1 l. If there is an isomorphism W from G to H. Suppose that 0 < r < n À 1. 1. j with 0 < i < n À 1. 0 < j < 1. 0 < s < 1.5 Example Let G ˆ S5 and let x. y be the following permutations in G: x ˆ (1 2 3 4 5). also. bÀ1 ab ˆ aÀ1 X Since we have x n ˆ y 2 ˆ 1. 0 < u < 1. b: an ˆ b2 ˆ 1. 0 < j < 1. and we write G  H. We now demonstrate the technique of Example 1. y À1 xy ˆ x À1 X We shall prove that the function W: G 3 H de®ned by W: a i b j 3 x i y j (0 < i < n À 1.Groups and homomorphisms 7 An invertible homomorphism is called an isomorphism. and write the 2n elements of G in the form ai b j with 0 < i < n À 1. y À1 xy ˆ x À1 . and suppose that H contains elements x and y which satisfy x n ˆ y 2 ˆ 1.

and so on. yl of G. Now recall that D10 ˆ ha. . and so j Hxi j ˆ j Hj. every element of G is in precisely one of the cosets). Suppose now that G is ®nite. the function W: D10 3 H de®ned by W: a i b j 3 x i y j (0 < i < 4. (1 2 3 4 5) denotes the permutation 1 3 2 3 3 3 4 3 5 3 1. we see that H ˆ fx i y j : 0 < i < 4. Thus. Hxr be all the distinct right cosets of H in G. b: a5 ˆ b2 ˆ 1. the subset Hx ˆ fhx: h P Hg of G is called a right coset of H in G. Since W is invertible. 0 < j < 1g. . a group of order 10. the function h 3 hxi (h P H) is a bijection from H to Hxi .8 Representations and characters of groups (Here we adopt the usual cycle notation ± thus. bÀ1 ab ˆ aÀ1 iX By Example 1. 0 < j < 1) is a homomorphism. The distinct right cosets of H in G form a partition of G (that is. For x in G. and let Hx1 . we deduce that jGj ˆ rj HjX In particular. Since G ˆ Hx1 ‘ X X X ‘ Hxr . we have . .) Check that x 5 ˆ y 2 ˆ 1. yl  D10. For all i. it is an isomorphism. and Hxi ’ Hxj is empty if i Tˆ j. H ˆ kx. Cosets Let G be a group and let H be a subgroup of G. Using the above relations.4. y À1 xy ˆ x À1 X Let H be the subgroup kx. .

9 The number r of distinct right cosets of H in G is called the index of H in G. h P G.Groups and homomorphisms 1. If n > 2 then there are just two right cosets of An in Sn . Na. Normal subgroups A subgroup N of a group G is said to be a normal subgroup of G if gÀ1 Ng ˆ N for all g P G (where gÀ1 Ng ˆ f gÀ1 ng: n P Ng). we see that GaN  C2 3 C2 . NabgX Since (Na)2 ˆ (Nb)2 ˆ (Nab)2 ˆ N. Thus jG: Hj ˆ jGjaj Hj when G is ®nite. called the factor group of G by N. we have fxy: x P Ng and y P Nhg ˆ NghX Hence we can de®ne a product operation on GaN by (Ng)(Nh) ˆ Ngh for all g. and An (1 2) ˆ f g P Sn : g oddgX Thus |Sn :An | ˆ 2. ka2 l ˆ {1. and so Sn aAn  C2 . b: a4 ˆ b2 ˆ 1. Nb. and is written as jG: Hj. we write N v G to indicate that N is a normal subgroup of G. we have An v Sn .7 Examples (1) For every group G. but the subgroup H ˆ kbl is not normal in G. Then N v G and bÀ1 ab ˆ aÀ1 l and let Nˆ GaN ˆ fN . a . (2) For n > 1.6 Lagrange's Theorem If G is a ®nite group and H is a subgroup of G. (3) Let G ˆ D8 ˆ ka. Suppose that N v G and let GaN be the set of right cosets of N in G. the sub-groups {1} and G are normal subgroups of G. namely An ˆ f g P Sn : g eveng. The subgroup kal is also normal in G. a2 }. The importance of the condition gÀ1 Ng ˆ N (for all g P G) is that it can be used to show that for all g. since b P H while aÀ1 ba ˆ a2 b P H. then j Hj divides |G|. h P GX This makes GaN into a group. 1.

Then GaKer W  Im WX An isomorphism is given by the function Kg 3 gW where K ˆ Ker W. For example.10 Representations and characters of groups Simple groups A group G is said to be simple if G Tˆ {1} and the only normal subgroups of G are {1} and G. Also. we eventually see that G is `built' out of a collection of simple groups. ( g P G) . The following result describes the way in which the kernel and image of W are related. (This is analogous to the fact that every positive integer is built out of its prime factors. with p a prime number. 1. Let G and H be groups and suppose that W: G 3 H is a homomorphism. is simple. we relate normal subgroups and factor groups to homomorphisms. We shall give examples of non-abelian simple groups in later chapters ± the smallest one is A5 . We de®ne the kernel of W by (1X8) Ker W ˆ f g P G: gW ˆ 1gX Then Ker W is a normal subgroup of G. If G is a ®nite group which is not simple. the image of W is (1X9) Im W ˆ f gW: g P Gg. the cyclic group Cp . Continuing this process with the smaller groups. then G has a normal subgroup N such that both N and G/N have smaller order than G. and Im W is a subgroup of H. Kernels and images To conclude the chapter. and in a sense. simple groups are fundamental to the study of ®nite groups.10 Theorem Suppose that G and H are groups and let W: G 3 H be a homomorphism. G is `built' out of these two smaller groups.) Thus.

Examples of groups are Cn ˆ ka: an ˆ 1l. a2 ˆ b2 . Ker W. Show that either W is an isomorphism or H ˆ {1}. G1 3 . C) ˆ the group of invertible n 3 n matrices over C. We know from Example 1. 2. . is a subgroup of H. 2. Sn ˆ the symmetric group of degree n. b: a4 ˆ 1. Q8 ˆ ka. illustrating Theorem 1. Exercises for Chapter 1 1. with multiplication (Ng)(Nh) ˆ NghX 3. and that W: G 3 H is a surjective homomorphism. D2 n ˆ ka.10. . GL(n.Groups and homomorphisms 1.11 Example The function W: Sn 3 C2 given by & 1. A normal subgroup N of G is a subgroup such that gÀ1 Ng ˆ N for all g in G. 3 Gr . Summary of Chapter 1 1. W: g 3 À1. Gr . The factor group GaKer W is isomorphic to Im W. 11 is a homomorphism. bÀ1 ab ˆ aÀ1 l. Im W. Suppose that G and H are groups. bÀ1 ab ˆ aÀ1 l. . . A homomorphism W: G 3 H is a function such that ( g1 g 2 )W ˆ ( g 1 W)( g 2 W) for all g1 . is a normal subgroup of G. if g is an odd permutation. and for n > 2. . Show that if G is an abelian group which is simple. Im W ˆ C2 . . g2 in G. if g is an even permutation. with G simple. and the image.7(2) that Sn /An  C2 . The kernel. An ˆ the alternating group of degree n. . The factor group GaN consists of the right cosets Ng ( g P G). We have Ker W ˆ An . b: an ˆ b2 ˆ 1. then G is cyclic of prime order. the direct product of the groups G1 .

yl of S4 . . 0 < s < 1). c2 ˆ d 2 . d: c4 ˆ 1. y be the permutations in S4 which are given by x ˆ (1 2). Let G ˆ D8 ˆ ha. Prove that f g P G: g n ˆ 1g is a cylic subgroup of G of order n. b: a4 ˆ b2 ˆ 1. y ˆ (3 4).12 Representations and characters of groups 3. show that x is a power of y. and H ˆ Q8 ˆ hc. and Ga (G ’ An )  C2 X 4. de®ned by ö: ar bs 3 x r y s . Y i of GL(2. Show that both the functions ö: G 3 K and ø: H 3 K. (b) Let G be a ®nite cyclic group. 6. and that G is not contained in An . de®ned by ë: ar bs 3 X r Y s . C). 5. (b) Let X. ì: cr d s 3 X r Y s (0 < r < 3. (a) Show that every subgroup of a cyclic group is cyclic.Y ˆ . and let K be the subgroup kx. bÀ1 ab ˆ aÀ1 i. ø: cr d s 3 x r y s (0 < r < 3. Find Ker ö and Ker ø. i 0 1 0 and let L be the subgroup hX . is a homomorphism. Y be the 2 3 2 matrices which are given by     0 i 0 À1 X ˆ . are homomorphisms. and let n be a positive integer which divides |G|. Suppose that G is a subgroup of Sn . y are elements of G with the same order. Show that just one of the functions ë: G 3 L and ì: H 3 L. Prove that G ’ An is a normal subgroup of G. 0 < s < 1). Prove that this homomorphism is an isomorphism. (c) If G is a ®nite cyclic group and x. d À1 cd ˆ cÀ1 iX (a) Let x. Prove that D4 m  D2 m 3 C2 if m is odd.

C) such that A has order 8. and B2 ˆ A4 and BÀ1 AB ˆ AÀ1 X Show that the group kA. Prove that H v G. Show that the set of non-zero complex numbers. Show that every group of even order contains an element of order 2.2(4). is a group. B has order 4. Prove that every ®nite subgroup of this group is cyclic. under the usual multiplication. 9. Suppose that H is a subgroup of G with |G: H| ˆ 2. . Bl has order 16.) 10. Find elements A and B of GL(2. (Hint: compare Q8 in Example 1. 8.Groups and homomorphisms 13 7.

Vector spaces Let F be either R (the set of real numbers) or C (the set of complex numbers). v in V and all ë. Most of the material will be familiar to you if you have taken a ®rst course on linear algebra. ì in F. (4) 1v ˆ v. (1) ë(u ‡ v) ˆ ëu ‡ ëv. we gather the results from linear algebra concerning vector spaces. namely group theory and linear algebra. and those of F are called scalars. together with a rule for adding any two elements u. (3) (ëì)v ˆ ë(ìv). in case you have not come across projections before. An exception occurs in the last section. so we omit the proofs. these rules must satisfy: (2. here.1) (a) V is an abelian group under addition. 14 .) Moreover. A vector space over F is a set V. linear transformations and matrices which we shall use later. where we deal with projections. v of V to form an element u ‡ v of V. (2) (ë ‡ ì)v ˆ ëv ‡ ìv. we explain in detail how the results work. For reference purposes. The elements of V are called vectors. (b) for all u.2 Vector spaces and linear transformations An attractive feature of representation theory is that it combines two strands of mainstream mathematics. and a rule for multiplying any element v of V by any element ë of F to form an element ëv of V. (The latter rule is called scalar multiplication. We write 0 for the identity element of the abelian group V under addition.

v n are linearly dependent if ë1 v1 ‡ X X X ‡ ë n v n ˆ 0 for some ë1 . . Bases of vector spaces Let v1 . . The vector space V is n-dimensional if dim V ˆ n. . . We say that v1 . xn belong to F. . . . X X X . we shall consider only vector spaces V which are ®nite-dimensional ± this means that V has a basis consisting of ®nitely many vectors. . . X X X . otherwise.2 Examples (1) Let R2 denote the set of all ordered pairs (x. . for each positive integer n. . . ë y)X Then R2 is a vector space over R. . y) where x and y are real numbers. A vector v in V is a linear combination of v1 . . as above. . . . X X X . . . . x2 . not all of which are zero. ë(x. x2 . The vectors v1 . xn ) ˆ (ëx1 . ë n in F. we consider row vectors (x1 . X X X .Vector spaces and linear transformations 15 2. . . . Throughout this book. . xn ‡ x9 ). (2) More generally. . . The number of vectors in a basis of V is called the dimension of V and is written as dim V. . . . v n form a basis of V if they span V and are linearly independent. and de®ne addition and scalar multiplication on F n by (x1 . v n are linearly independent. . . 1 n 1 n ë(x1 . It turns out that any two bases of V have the same number of vectors. . If V ˆ {0} then dim V ˆ 0. v1 . X X X . . ëxn )X Then F n is a vector space over F. . We denote the set of all such row vectors by F n. v n are said to span V if every vector in V is a linear combination of v1 . X X X . . y ‡ y9). . v n . xn ) where x1 . . v n be vectors in a vector space V over F. xn ) ‡ (x9 . . De®ne addition and scalar multiplication on R2 by (x. y) ‡ (x9. v n if v ˆ ë1 v1 ‡ X X X ‡ ë n v n for some ë1 . . x9 ) ˆ (x1 ‡ x9 . The vectors v1 . y9) ˆ (x ‡ x9. y) ˆ (ëx. . . ë n in F.

. .6 Examples (1) {0} and V are subspaces of V. that is. 0. 2. Except in the case where V ˆ {0}. 0.16 Representations and characters of groups 2. . Then (1. (2. . (1. Subspaces A subspace of a vector space V over F is a subset of V which is itself a vector space under the addition and scalar multiplication inherited from V. . . (0.4) If v1 . ur be vectors in V. . The vector v therefore determines the scalars ë1 . (2) if u. 0. . Indeed. (0. . . . 1. 1. . ur .5). For a subset U of V to be a subspace. each vector v in V can be written in a unique way as v ˆ ë1 v1 ‡ X X X ‡ ë n v n . . . v k are linearly independent vectors in V. 0). .5) (1) 0 P U. with ë1 . . ur ) to be the set of all linear combinations of u1 . . X X X . . . . 0. . X X X . . ë n in F. . X X X . then there exist v k‡1 . 1)X Given a basis v1 . ur ) ˆ fë1 u1 ‡ X X X ‡ ë r ur : ë1 . the next result says that any linearly independent vectors can be extended to a basis. v n form a basis of V. ur . X X X . . it is necessary and suf®cient that all the following conditions hold: (2. v n of a vector space V. . 0. X X X . 0). 1) is a basis of V. ë n . . . there are many bases of V. . . Another basis is (1. . . (2) Let u1 . 0. 0). . .3 Example Let V ˆ F n. X X X . X X X . . 0. . 1. so dim V ˆ n. X X X . . . v n in V such that v1 . . . . 0. and it is called the subspace spanned by u1 . v P U then u ‡ v P U. (3) if ë P F and u P U then ëu P U. . . ur ) is a subspace of V. . 1. . 0). X X X . X X X . . ë r P FgX By (2. We de®ne sp (u1 . sp (u1 . sp (u1 . (1.

. . .4)). . unless U is {0} or V. . . (2.4 if you have dif®culty with the proof. Ur are subspaces of a vector space V. (2. . . . . ws is a basis of V. v n of V (see (2. . and for 1 < i < n. You should consult the solutions to Exercises 2. . . that u1 . ‡ Ur is de®ned by U1 ‡ X X X ‡ Ur ˆ fu1 ‡ X X X ‡ ur : ui P Ui for 1 < i < rgX By (2. . . v k to a basis v1 . . . . dim U ˆ dim V if and .5). Then V ˆ U È WX From this construction it follows that there are in®nitely many subspaces W with V ˆ U È W. . only if U ˆ V.9) Suppose that V ˆ U ‡ W. . .4). then we write it as U1 È X X X È Ur X 2. .7) Suppose that U is a subspace of the vector space V . U1 ‡ .3 and 2. .Vector spaces and linear transformations Notice that the following fact is a consequence of (2. . Then V ˆ U1 È X X X È Un X (2) Let U be a subspace of V and let v1 . . ws is a basis of W. We say that the sum U1 ‡ . v n is a basis of V. Then the following three conditions are equivalent: (1) V ˆ U È W. . ‡ ur with ui P Ui for 1 < i < r. . . . ur . . The next result is frequently useful when dealing with the direct sum of two subspaces. If the sum is direct. . . w1. Then dim U < dim V Also. . (2) u1 . Direct sums of subspaces 17 If U1. let Ui be the subspace spanned by v i.8 Examples (1) Suppose that v1 . . ‡ Ur is a direct sum if every element of the sum can be written in a unique way as u1 ‡ . . . (3) U ’ W ˆ {0}. ur is a basis of U and that w1. . ‡ Ur is a subspace of V. v n ). . . . . . . v k be a basis of U. Extend v1. . . . . and let W ˆ sp (v k‡1 . then the sum U1 ‡ . . .

ur ): ui P U i for 1 < i < rgX De®ne addition and scalar multiplication on V as follows: for all ui . . A linear transformation from V to W is a function W: V 3 W which satis®es (u ‡ v)W ˆ uW ‡ vW (ëv)W ˆ ë(vW) for all u. for 1 < i < r. . and for all ë P F and v P V X . and let V ˆ f(u1 . Ur be vector spaces over F. Ur. Let U1. . X X X . If V ˆ U È W and also U ˆ U1 È X X X È Ua . .18 Representations and characters of groups Our next result. . X X X . W1. X X X . ëur )X With these de®nitions. can be deduced immediately from the de®nition of a direct sum. . X X X . X X X . ui . ur ‡ u9 ). . . X X X . and W ˆ W1 È X X X È Wb . . V is a vector space over F. . and. 1 r 1 r ë(u1 . let (u1 . X X X . u9 i in Ui (1 < i < r) and all ë in F. U1. then V ˆ U1 È X X X È Ua È W 1 È X X X È W b X We now introduce a construction for vector spaces which is analogous to the construction of direct products for groups. . Ua. then it is immediate that V ˆ U1 È X X X È U 9X 9 r We call V the external direct sum of U1. u9 ) ˆ (u1 ‡ u9 . If. W. involving the direct sum of several subspaces. we put U 9 ˆ f(0. Wb are subspaces of a vector space V. . . X X X . 0): ui P U i g i (where the ui is in the ith position).10) Suppose that U. . . ur ) ˆ (ëu1 . v P V . (2. we write V ˆ U1 È X X X È Ur X Linear transformations Let V and W be vector spaces over F. abusing notation slightly. ur ) ‡ (u9 . .

it is easy to check that Ker W is a subspace of V and Im W is a subspace of W.5). . Their dimensions are connected by the following equation. and then saying `extend the action of ö to be linear'. there is a unique linear transformation ö: V 3 W such that v i öˆ wi for all i. Kernels and images Suppose that W: V 3 W is a linear transformation. v n is a basis of V. then W is a linear transformation. The kernel of W (written Ker W) and the image of W (written Im W) are de®ned as follows: (2X11) Ker W ˆ fv P V : vW ˆ 0g. so a linear transformation preserves addition and scalar multiplication. the linear transformation ö is given by (ë1 v1 ‡ X X X ‡ ë n v n )ö ˆ ë1 w1 ‡ X X X ‡ ë n wn X We sometimes construct a linear transformation ö: V 3 W in this way. . . . . then for ë1 . Im W ˆ fvW: v P V gX Using (2. W is determined by its action on a basis. . . Furthermore. . given any basis v1 . . .Vector spaces and linear transformations 19 Just as a group homomorphism preserves the group multiplication. which is known as the Rank±Nullity Theorem: (2X12) dim V ˆ dim (Ker W) ‡ dim (Im W)X 2. . v n of V and any n vectors w1. and Ker W ˆ f0g. . Im W ˆ V X . and Ker W ˆ V .13 Examples (1) If W: V 3 W is de®ned by vW ˆ 0 for all v P V. . wn in W. . by specifying the values of ö on a basis of V. ë n in F we have (ë1 v1 ‡ X X X ‡ ë n v n )W ˆ ë1 (v1 W) ‡ X X X ‡ ë n (v n W)X Thus. Im W ˆ f0gX (2) If W: V 3 V is de®ned by vW ˆ 3v for all v P V. then W is a linear transformation. . Notice that if W: V 3 W is a linear transformation and v1. .

for all v P V. (2) Ker W ˆ {0}. By applying (2.7) into account. z P R. (3) Im W ˆ V. Invertible linear transformations Again. Endomorphisms A linear transformation from a vector space V to itself is called an endomorphism of V.12). À3. z)W ˆ (x ‡ 2 y ‡ z. A linear transformation W from V to W is injective if and only if Ker W ˆ {0}.1). By also taking (2. v(Wö) ˆ (vW)ö. . we have Ker W ˆ sp ((7. let V and W be vector spaces over F.2). We write W2 for WW. Suppose that W and ö are endomorphisms of V and ë P F. y. then W is a linear transformation. we see that isomorphic vector spaces have the same dimension. We de®ne the functions W ‡ ö. y. then V and W are said to be isomorphic vector spaces. It turns out that the inverse of an invertible linear transformation is also a linear transformation (see Exercise 2. If there exists an invertible linear transformation from V to W. and hence W is invertible precisely when W is surjective and Ker W ˆ {0}. Im W ˆ R2 . (2. Then W ‡ ö. Wö and ëW from V to V by (2X15) v(W ‡ ö) ˆ vW ‡ vö.14) Let W be a linear transformation from V to itself. Then the following three conditions are equivalent: (1) W is invertible.20 Representations and characters of groups (3) If W: R3 3 R2 is given by (x. À1)). so dim (Ker W) ˆ 1 and dim (Im W) ˆ 2. À y ‡ 3z) for all x. Wö and ëW are endomorphisms of V. we obtain the next result (see Exercise 2. v(ëW) ˆ ë(vW).

2. y)(Wö) ˆ (Àx ‡ 5 y. . y) 3 (x ‡ y. (0. y)(3W) ˆ (3x ‡ 3 y. y)ö ˆ (x À 2 y.17 De®nition The n 3 n matrix (aij ) is called the matrix of W relative to the basis B . x À 2 y).Vector spaces and linear transformations 2. If B is the basis (1. À2x ‡ 4 y)X Then W and ö are endomorphisms of V. (x. Note that Ker (W À ë1 V ) ˆ fv P V : vW ˆ ëvgX (2) Let V ˆ R2. Suppose that v1 . ö be the functions from V to V de®ned by (x. Àx ‡ 2 y). . 3W and W2 are given by (x. (x. then [W]B ˆ In for all bases B of V. . (x. and let W.16 Examples (1) The identity function 1 V de®ned by 1 V : v 3 v for all v P V 21 is an endomorphism of V. v i W ˆ a i1 v1 ‡ X X X ‡ ain v n X 2. then so is W À ë1 V . 0). where In denotes the n 3 n identity matrix. and is denoted by [W]B . 2x À 10 y). v n is a basis of V and call it B . . Wö. Àx ‡ 5 y)X Matrices Let V be a vector space over F. y)(W ‡ ö) ˆ (2x À y. y)W2 ˆ (2x À y.18 Examples (1) If W ˆ 1 V (so that vW ˆ v for all v P V). 3x À 6 y). for all ë P F. y)W ˆ (x ‡ y. 1 < j < n) such that for all i. If W is an endomorphism of V. 1) of V and B 9 is the basis . x À 2 y) of V. (2) Let V ˆ R2 and let W be the endomorphism (x. Then there are scalars aij in F (1 < i < n. and W ‡ ö. (x. and let W be an endomorphism of V.

j. As you know. their sum A ‡ B is the m 3 n matrix over F whose ij-entry is aij ‡ bij for all i. Given an m 3 n matrix A ˆ (aij ) and an n 3 p matrix B ˆ (bij ). the product of two matrices is de®ned in a less transparent way. and for ë P F. 1) of V. and that W and ö are endomorphisms of V. (1. the matrix ëA is the m 3 n matrix over F obtained from A by multiplying all the entries by ë. their product AB is the m 3 p matrix whose ij-entry is n ˆ kˆ1 aik bkj X 2. and [Wö]B ˆ [W]B [ö]B X .22 Representations and characters of groups   1 . then we describe A as a matrix over F. Then [W ‡ ö]B ˆ [W]B ‡ [ö]B . À2  [W]B 9 ˆ  1 X À1 (1. 0). AB ˆ 3 .20) Suppose that B is a basis of the vector space V.19 Example Let Aˆ Then A‡ Bˆ 2 BA ˆ  À1 3   2 0 . then [W]B ˆ 1 1 0 3 If we wish to indicate that the entries in a matrix A come from F. 2 À1 5 À2 À4 3 0 3 3 . Given two m 3 n matrices A ˆ (aij ) and B ˆ (bij ) over F. Bˆ 1 2 2 3  À4 X À1 2 . 3A ˆ X À3 9 6 3 4 2 2 À13 À12 À5 The matrix of the sum or product of two endomorphisms (relative to some basis) is related to the matrices of the individual endomorphisms in the way you would expect: (2.

20): given a basis B of V. and follows from (2. y) ˆ (x ‡ 3 y. An invertible matrix converts one basis into another. the matrix product vA also lies in V. Write det A for the determinant of A.22 Example Let Aˆ  1 3  À1 X 2 Then A gives us an endomorphism W of F 2. y)W ˆ (x.21) If A is an n 3 n matrix over F then the function . [ëW]B ˆ ë[W]B X 23 We showed you in (2. Then a necessary and suf®cient condition for A to be invertible is that det A Tˆ 0. v 3 vA (v P F n ) is an endomorphism of F n. Invertible matrices turn up when we relate two bases of a vector space. . an endomorphism W of V is invertible if and only if the matrix [W]B is invertible. Suppose that A is an n 3 n matrix over F. if it exists. xn ) with each xi in F. (2.17) how to get a matrix from an endomorphism of a vector space V. It is easy enough to reverse this process and use a matrix to de®ne an endomorphism. and this same matrix is used to describe the way in which the matrix of an . where   1 À1 (x. We concentrate on a particular way of doing this. . 2. given a basis of V. Àx ‡ 2 y)X 3 2 Invertible matrices An n 3 n matrix A is said to be invertible if there exists an n 3 n matrix B with AB ˆ BA ˆ In . it is called the inverse of A and is written as AÀ1 . for all scalars ë. The following remark is easily justi®ed. Such a matrix B. .Vector spaces and linear transformations Also. the vector space of row vectors (x1 . and let V ˆ F n. . The connection between invertible endomorphisms and invertible matrices is straightforward. Then for all v in V. is unique.

25 Example Suppose that V ˆ R2.24) below. . Such a vector v is called an eigenvector of W. (2. Let B be the basis (1. and let v9 . 2. Then     1 0 1 0 À1 Tˆ . 1) of V. The precise meaning of these remarks is revealed in the de®nition (2.23 De®nition Let v1 . 0). . . (0. . y) 3 (x ‡ y. (1. and suppose that W is an endomorphism of V. 1) and B 9 the basis (1.24) If B and B 9 are bases of V and W is an endomorphism of V. Then for 1 < i < n.23) and the result (2. The scalar ë is said to be an eigenvalue of W if vW ˆ ëv for some non-zero vector v in V. then       1 0 0 1 1 0 1 1 [W]B ˆ ˆ T À1 [W]B 9 T X ˆ À1 1 3 À1 1 1 1 À2 Eigenvalues Let V be an n-dimensional vector space over F. then [W]B ˆ T À1 [W]B 9 T . and is called the change of basis matrix from B to B 9.18(2). . v9 ˆ t i1 v1 ‡ X X X ‡ tin v n i for certain scalars tij .T ˆ X 1 1 À1 1 If W is the endomorphism W: (x. The inverse of T is the change of basis matrix from B 9 to B . The n 3 n matrix T ˆ (tij ) is invertible.24 Representations and characters of groups endomorphism depends upon the basis. as in Example 2. . v9 1 n be a basis B 9 of V. . where T is the change of basis matrix from B to B 9. . 2. 0). x À 2 y) of V. . v n be a basis B of the vector space V.

y)W ˆ (À y. Corresponding eigenvectors are (1. Then W has an eigenvalue. then the eigenvalues of W are those scalars ë in F which satisfy the equation det ([W]B À ëI n ) ˆ 0X Solving this equation involves ®nding the roots of a polynomial of degree n. (0. then   0 1 [W]B ˆ X À1 0 We have det ([W]B À ëI2 ) ˆ ë2 ‡ 1. y)W ˆ (À y. 0). i). (1. and let W be an endomorphism of V. 1) of V. Therefore. i) of V. if B is a basis of V. then   i 0 [W]B 9 ˆ X 0 Ài (2) Let V ˆ R2 and let W again be the endomorphism which is given by (x. x)X This time. Note that if B 9 is the basis (1.27 Examples (1) Let V ˆ C2 and let W be the endomorphism of V which is given by (x. x)X If B is the basis (1.26). Since every non-constant polynomial with coef®cients in C has a root in C. the element ë of F is said to be an eigenvalue of A if vA ˆ ëv for some non-zero row vector v in F n. Thus we depend upon F being C in result (2. 2. (2. which occurs if and only if W À ë1 V is not invertible. For an n 3 n matrix A over F. we deduce the following result. Ài) and (1. Ài).26) Let V be a non-zero vector space over C.Vector spaces and linear transformations 25 Now ë is an eigenvalue of W if and only if Ker (W À ë1 V ) Tˆ {0}. V is a vector space over R. The eigenvalues of A are those elements ë of F which satisfy det (A À ëI n ) ˆ 0X . so i and Ài are the eigenvalues of W. and W has no eigenvalues in R.

that aii ˆ ë i for 1 < i < n. Im ð ˆ U . and since uð ˆ u for all u in U. (ëv)ð ˆ (ëu ‡ ëw)ð ˆ ëu ˆ ë(vð)X Therefore. Projections If a vector space V is a direct sum of two subspaces U and W. ë n .29 Proposition Suppose that V ˆ U È W. the eigenvalues are ë1 . De®ne ð: V 3 V by (u ‡ w)ð ˆ u for all u P U . . in addition. . . w P W. Clearly Im ð # U. Also. Then v ˆ u ‡ w and v9 ˆ u9 ‡ w9 for some u. We often display such a matrix in the form H I ë1 0 f g FF Aˆd e F ën 0 which indicates. Ker ð ˆ W and ð2 ˆ ðX Proof Since every vector in V has a unique expression in the form u ‡ w with u P U. u9 in U and w. for ë in F. ð is an endomorphism of V.28 Example We say that an n 3 n matrix A ˆ (aij ) is diagonal if aij ˆ 0 for all i and j with i Tˆ j. w P W X Then ð is an endomorphism of V. we have Im ˆ U. . Let v and v9 belong to V.26 Representations and characters of groups 2. We have (v ‡ v9)ð ˆ (u ‡ u9 ‡ w ‡ w9)ð ˆ u ‡ u9 ˆ (u ‡ w)ð ‡ (u9 ‡ w9)ð ˆ vð ‡ v9ðX Also. Further. For this diagonal matrix A. . w9 in W. then we can construct a special endomorphism of V which depends upon the expression V ˆ U È W: 2. it follows that ð is a function on V.

29. it follows that v ˆ vð ˆ 0.30 De®nition An endomorphism ð of a vector space V which satis®es ð2 ˆ ð is called a projection of V.31 Example The endomorphism (x. (u ‡ w)ð2 ˆ uð ˆ u ˆ (u ‡ w)ð. and so Ker ð ˆ W. since (v À vð)ð ˆ vð À vð2 ˆ vð À vð ˆ 0X This establishes that V ˆ Im ð ‡ Ker ð. Then V ˆ Im ð È Ker ðX Proof If v P V then v ˆ vð ‡ (v À vð). as in Proposition 2. Clearly the ®rst term vð belongs to Im ð. we have v ˆ uð for some u P V. As v P Im ð. j . Àx À y) of R2 is a projection. Finally. 27 j 2. We now show that every projection can be constructed using a direct sum.9) now shows that V ˆ Im ð È Ker ð.32 Proposition Suppose that ð is a projection of a vector space V. Now suppose that v lies in Im ð ’ Ker ð.Vector spaces and linear transformations (u ‡ w)ð ˆ 0 D u ˆ 0 D u ‡ w P W . and the second term v À vð lies in Ker ð. and so ð2 ˆ ð. 2. Thus Im ð ’ Ker ð ˆ f0g. Therefore vð ˆ uð2 ˆ uð ˆ vX Since v P Ker ð. 2. and (2. y) 3 (2x ‡ 2 y.

Given two bases B and B 9 of V. A linear transformation W: V 3 W satis®es (u ‡ v)W ˆ uW ‡ vW and (ëv)W ˆ ë(vW) for all u. and an endomorphism W of V. Àx À y) is the projection of R2 which appears in Example 2. Show that if V and W are vector spaces and W: V 3 W is an invertible linear transformation then WÀ1 is a linear transformation. Given a basis B of the n-dimensional vector space V. and dim V ˆ dim (Ker W) ‡ dim (Im W)X 4. 5. V ˆ U È W if and only if V ˆ U ‡ W and U ’ W ˆ {0}. . F n is the set of row vectors (x1 . where F ˆ C or R. there exists an invertible matrix T such that [W]B ˆ T À1 [W]B 9 TX 6. 7. y) 3 (2x ‡ 2 y. For example. Also. x n ) with each xi in F. . Àx): x P Rg. A linear transformation W: V 3 W is invertible if and only if Ker W ˆ {0} and Im W ˆ W. Ker W is a subspace of V and Im W is a subspace of W. Àx): x P RgX Summary of Chapter 2 1. there is a correspondence between the endomorphisms W of V and the n 3 n matrices [W]B over F. Exercises for Chapter 2 1. . All our vector spaces are ®nite-dimensional over F.28 Representations and characters of groups 2. È Ur if each Ui is a subspace of V. v in V and all ë in F. 3. Ker ð ˆ f(x.33 Example If ð: (x. and dimF n ˆ n. and every element v of V has a unique expression of the form v ˆ u1 ‡ . ‡ ur (ui P Ui ). 2. . V ˆ U1 È .31. Eigenvalues ë of an endomorphism W satisfy vW ˆ ëv for some nonzero v in V. A projection is an endomorphism ð of V which satis®es ð2 ˆ ð. F F F . then Im ð ˆ f(2x. .

Show that the following are equivalent: (1) W is invertible. ws is a basis of W. and that V ˆ U1 È . . . (a) Let U1. . (2) Ker W ˆ {0}. where U ˆ fv P V : vW ˆ vg. . Let V be a vector space and let W be an endomorphism of V. Suppose that W is an endomorphism of the vector space V. 8. . 5. (3) Im W ˆ V. Let U and W be subspaces of the vector space V. w1. Show that V ˆ U È W. . Suppose that u1 . . with V ˆ U1 ‡ U2 ‡ U3. 4. È Ur. . . W ˆ fv P V : vW ˆ ÀvgX Deduce that V has a basis B such that [W]B is diagonal. with all diagonal entries equal to ‡1 or À1. but V Tˆ Im ö È Ker ö. .Vector spaces and linear transformations 29 2. . Suppose that U1. Show that V ˆ U1 È U2 È U3 D U 1 ’ (U 2 ‡ U3 ) ˆ U 2 ’ (U 1 ‡ U3 ) ˆ U 3 ’ (U 1 ‡ U2 ) ˆ f0gX (b) Give an example of a vector space V with three subspaces U1. . Ur are subspaces of the vector space V. . Prove that V ˆ U È W if and only if V ˆ U ‡ W and U ’ W = {0}. . ur . Prove that dim V ˆ dim U1 ‡ X X X ‡ dim U r X 7. . . but V Tˆ U1 È U2 È U3. . ur is a basis of U and w1. Suppose that W is an endomorphism of the vector space V and W2 ˆ 1 V . . . . . Give an example of a vector space V with endomorphisms W and ö such that V ˆ Im W È Ker W. 6. 3. ws is a basis of V. U2 and U3 such that V ˆ U1 ‡ U2 ‡ U3 and U1 ’ U 2 ˆ U1 ’ U3 ˆ U2 ’ U 3 ˆ f0g. 9. with all diagonal entries equal to 1 or 0. Show that V ˆ U È W if and only if u1 . U2 and U3 be subspaces of a vector space V. . Let U and W be subspaces of the vector space V. . Show that W is a projection if and only if there is a basis B of V such that [W]B is diagonal.

We also introduce the concept of equivalence of representations.1 De®nition A representation of G over F is a homomorphism r from G to GL (n. for some n. and consider the kernel of a representation. where In denotes the n 3 n identity matrix. To be precise. Thus if r is a function from G to GL (n. then r is a representation if and only if ( gh)r ˆ ( gr)(hr) for all g. F). The degree of r is the integer n. Recall from the ®rst chapter that GL (n. F) denotes the group of invertible n 3 n matrices with entries in F. and g À1 r ˆ ( gr)À1 for all g P G. 3. h P GX Since a representation is a homomorphism. we have 1r ˆ I n . it follows that for every representation r: G 3 GL (n. Representations Let G be a group and let F be R or C. and give some examples of representations. F). We set out this idea in more detail. F).3 Group representations A representation of a group G gives us a way of visualizing G as a group of matrices. a representation is a homomorphism from G into a group of invertible matrices. 30 .

Group representations 31 3.2 Examples (1) Let G be the dihedral group D8 ˆ ka. F) which is given by r: a i b j 3 A i B j (0 < i < 3. Equivalent representations We now look at a way of converting a given representation into another one. The degree of r is 2. BÀ1 AB ˆ AÀ1 X It follows (see Example 1. This shows that every group has representations of arbitrarily large degree. De®ne the matrices A and B by     0 1 1 0 Aˆ . . where In is the n 3 n identity matrix. F) by gr ˆ I n for all g P G. 0 < j < 1) is a representation of D8 over F.4) that the function r: G 3 GL (2. The matrices gr for g in D8 are given in the following table: g gr  1  1 0 0 1 a  0 1 À1 0 a2  À1 0 0 À1 a3  0 À1 1 0    g gr  b  1 0 0 À1  ab  0 À1 À1 0 2  a b  À1 0 0 1 3  a b  0 1 1 0 (2) Let G be any group. as usual. Then ( gh)r ˆ I n ˆ I n I n ˆ ( gr)(hr) for all g. De®ne r: G 3 GL (n. h P G. so r is a representation of G. Bˆ À1 0 0 À1 and check that A4 ˆ B2 ˆ I. bÀ1 ab ˆ aÀ1 l. b: a4 ˆ b2 ˆ 1.

3. indeed. h P G. we have (see Exercise 3. b: a4 ˆ b2 ˆ 1. Note that for all n 3 n matrices A and B. ( gh)ó ˆ T À1 (( gh)r)T ˆ T À1 (( gr)(hr))T ˆ T À1 ( gr)T . bÀ1 ab ˆ aÀ1 l.2(1). In other words. F) be representations of G over F. We say that r is equivalent to ó if n ˆ m and there exists an invertible n 3 n matrix T such that for all g P G. (2) if r is equivalent to ó then ó is equivalent to r. then r is equivalent to ô. we simply de®ne gó ˆ T À1 ( gr)T Then for all g. gó ˆ T À1 ( gr)TX Note that for all representations r.4 Examples (1) Let G ˆ D8 ˆ ka.32 Representations and characters of groups Let r: G 3 GL (n. T À1 (hr)T ˆ ( gó )(hó ).4): (1) r is equivalent to r.3 De®nition Let r: G 3 GL (m. (3) if r is equivalent to ó and ó is equivalent to ô. ó and ô of G over F. a representation. F) and ó : G 3 GL (n. and let T be an invertible n 3 n matrix over F. Thus ar ˆ A for all g P GX . equivalence of representations is an equivalence relation. we have (T À1 AT )(T À1 BT ) ˆ T À1 (AB)TX We can use this observation to produce a new representation ó from r. and so ó is. and consider the representation r of G which appears in Example 3. 3. F) be a representation.

Hence r: 1 3 I. we have     i 0 0 1 À1 À1 T AT ˆ . and de®ne 1 T ˆp 2 Then 1 i 1 T À1 ˆ p 2  In fact. Tˆ 1 À1 then T À1 AT ˆ   0 .Group representations and br ˆ B. (2) Let G ˆ C2 ˆ ka: a2 ˆ 1l and let  À5 Aˆ À2  12 X 5 Check that A2 ˆ I. where Aˆ    1 1 . bó ˆ aó ˆ 1 0 0 Ài The representations r and ó are equivalent. T has been constructed so that T À1 AT is diagonal. . T BT ˆ . 0 Ài 1 0 and so we obtain a representation ó of D8 for which     0 1 i 0 X . a 3 A is a representation of G. À1 1 0 and so we obtain a representation ó of G for which     1 0 1 0 . . aó ˆ 1ó ˆ 0 À1 0 1 and ó is equivalent to r. If   2 À3 . Bˆ 0 0   1 X Ài  1 Ài X 1 i  0 X À1 33 0 À1 Assume that F ˆ C.

and when gr ˆ In for all g in G. In agreement with De®nition 1. F). 3.8. is called the trivial representation of G.5 De®nition The representation r: G 3 GL (1. 3. these are when the degree of r is 1. that is. Kernels of representations We conclude the chapter with a discussion of the kernel of a representation r: G 3 GL (n. as is shown by the following de®nition.7 Proposition A representation r of a ®nite group G is faithful if and only if Im r is isomorphic to G. Thus Ker r ˆ f g P G: gr ˆ I n gX Note that Ker r is a normal subgroup of G. Of particular interest are those representations whose kernel is just the identity subgroup. However. 3. F) is said to be faithful if Ker r ˆ {1}. F) which is de®ned by gr ˆ (1) for all g P G. the trivial representation of G is the representation where every group element is sent to the 1 3 1 identity matrix. . To put the de®nition another way. if the identity element of G is the only element g for which gr ˆ In . this consists of the group elements g in G for which gr is the identity matrix. there are usually lots of representations which are equivalent to r.34 Representations and characters of groups There are two easily recognized situations where the only representation which is equivalent to r is r itself. It can happen that the kernel of a representation is the whole of G.6 De®nition A representation r: G 3 GL (n.

A representation is faithful if it is injective. 2. Conversely.10. gó ˆ T À1 ( gr)T X 3. A representation of a group G is a homomorphism from G into GL(n. for some n. that is. it follows that all representations which are equivalent to a faithful representation are faithful. The basic problem of representation theory is to discover and understand representations of ®nite groups. then these two groups have the same (®nite) order. Therefore. . the factor group G/ Ker r is isomorphic to Im r. Representations r and ó of G are equivalent if and only if there exists an invertible matrix T such that for all g P G. F). r is faithful.Group representations 35 Proof We know that Ker r v G and by Theorem 1. The group generated by the matrices     0 1 1 0 and À1 0 0 À1 is therefore isomorphic to D8. (3) The trivial representation of a group G if faithful if and only if G ˆ {1}. (2) Since T À1 AT ˆ In if and only if A ˆ In . if G  Im r. j 3. Summary of Chapter 3 1. since the identity is the only element g which satis®es gr ˆ I.2(1) is faithful. if Ker r ˆ {1} then G  Im r.8 Examples (1) The representation r of D8 given by j  i 1 0 0 1 i j (a b )r ˆ 0 À1 À1 0 as in Example 3. and so |Ker r| ˆ 1. In Chapter 6 we shall show that every ®nite group has a faithful representation.

Prove: (1) r is equivalent to r. B. bÀ1 ab ˆ aÀ1 l. D over C by     0 1 eiða3 0 Aˆ . C) (k ˆ 1. C). b: a6 ˆ b2 ˆ 1. and de®ne r: G 3 GL (n.36 Representations and characters of groups Exercises for Chapter 3 1. 2. Which of these representations are faithful? 3. then ó is equivalent to r. de®ned by r1 : ar 3 Ar . De®ne the matrices A. (3) if r is equivalent to ó. F) such that ar ˆ (1) and br ˆ (À1).  1 0   1 0 . say G ˆ ka: am ˆ 1l. Let Aˆ (0 < r < m À 1)X Show that r is a representation of G over C if and only if Am ˆ I. Suppose that G ˆ D2 n ˆ ka. 4. 5. Suppose that r. 3. (2) if r is equivalent to ó. Let G be the cyclic group of order m. r2 : ar 3 Br . ó and ô are representations of G over F.Cˆ  0 À1 1 À1  e2ðia3 and let G ˆ ka: a3 ˆ 1l  C3 . Suppose that A P GL (n. Bˆ 0 1 0  . C) (1 < j < 3). Dˆ X À 3a2 1a2 0 À1 Prove that each of the functions r k : G 3 GL (2. bÀ1 ab ˆ aÀ1 l. then r is equivalent to ô. is a representation of G over C. Show that each of the functions r j : G 3 GL (2. b: an ˆ b2 ˆ 1. Show that there is a representation r: G 3 GL (1. 4). Bˆ . C) by r: ar 3 Ar 2. Let G ˆ D12 ˆ ka. and ó is equivalent to ô. C. r3 : a r 3 C r (0 < r < 2). given by . and F ˆ R or C. 1 0 0 eÀiða3     p 1a2 3a2 1 0 p Cˆ .

37 is a representation of G. Which of these representations are faithful? Which are equivalent? 6. r3 : ar bs 3 (ÀA) r Bs . Give an example of a faithful representation of D8 of degree 3. r 4 : ar bs 3 C r D s (0 < r < 5. r2 : ar bs 3 A3 r (ÀB) s . Suppose that g and h are elements of G such that ( gr)(hr) ˆ (hr)( gr). Let r be a representation of the group G. Suppose that r is a representation of G of degree 1. 7. Prove that Ga Ker r is abelian. Does it follow that gh ˆ hg? .Group representations r1 : ar bs 3 Ar Bs . 0 < s < 1). 8.

. since 1r is the identity matrix. 38 . FG-modules Let G be a group and let F be R or C. the vector space of all row vectors (ë1 . Suppose that r: G 3 GL (n. F) is a representation of G. For all v P V and g P G. and show that there is a close connection between FG-modules and representations of G over F. the fact that r is a homomorphism shows that v(( gh)r) ˆ v( gr)(hr) for all v P V and all g. h P G. Much of the material in the remainder of the book will be presented in terms of FG-modules. . the matrix product v( gr). is a row vector in V (since the product of a 1 3 n matrix with an n 3 n matrix is again a 1 3 n matrix). Write V ˆ F n . we have v(1r) ˆ v for all v P V. Next. the properties of matrix multiplication give (ëv)( gr) ˆ ë(v( gr)).4 FG-modules We now introduce the concept of an FG-module. as there are several advantages to this approach to representation theory. We now list some basic properties of the multiplication v( gr). . . ë n ) with ë i P F. Finally. of the row vector v with the n 3 n matrix gr. First.

(v P V ) . 39 4. v P V. ë P F and g P G.2 De®nition Let V be a vector space over F and let G be a group. g P G) is de®ned. v( gh) ˆ (v g)h. Àë1 )X Motivated by the above observations on the product v( gr). we now de®ne an FG-module. b: a4 ˆ b2 ˆ 1. (ëv) g ˆ ë(v g). bÀ1 ab ˆ aÀ1 l. Note that conditions (1). br ˆ ar ˆ 0 À1 À1 0 If v ˆ (ë1 . ë2 ) P F 2 then.FG-modules (u ‡ v)( gr) ˆ u( gr) ‡ v( gr) for all u. 4.2(1).1 Example Let G ˆ D8 ˆ ka. v1 ˆ v. for example. Àë2 ). v P V. (u ‡ v) g ˆ ug ‡ v g. F) be the representation of G over F given in Example 3. ë1 ). Thus     1 0 0 1 X . v(a3 r) ˆ (ë2 . h P G: (1) (2) (3) (4) (5) v g P V. and let r: G 3 GL (2. (4) and (5) in the de®nition ensure that for all g P G. Then V is an FG-module if a multiplication v g (v P V. v(ar) ˆ (Àë2 . satisfying the following conditions for all u. We use the letters F and G in the name `FG-module' to indicate that V is a vector space over F and that G is the group from which we are taking the elements g to form the products v g (v P V). ë P F and g. v(br) ˆ (ë1 . the function v 3 vg is an endomorphism of V.

X X X . For each g P G.40 Representations and characters of groups 4. Then the function g 3 [ g]B is a representation of G over F. F) is a representation of G over F. then gr ˆ [ g]B for all g P G. relative to the basis B . X X X . The connection between FG-modules and representations of G over F is revealed in the following basic result. (0. (ëv)( gr) ˆ ë(v( gr)). let [ g]B denote the matrix of the endomorphism v 3 v g of V. 0. g P GX Moreover. ë P F and g. and V ˆ F n. 0. 4. (0.4 Theorem (1) If r: G 3 GL(n. 0. v(( gh)r) ˆ (v( gr))(hr). 0. v P F n. F n becomes an FG-module if we de®ne v g ˆ v( gr) for all v P F n . there is a basis B of V such that (2) Assume that V is an FG-module and let B be a basis of V. v(1r) ˆ v. if we let B be the basis (1. X X X . g P G)X for all g P GX Moreover. 1) of F n. 0. (u ‡ v)( gr) ˆ u( gr) ‡ v( gr)X Therefore. 0). 1. then V becomes an FG-module if we de®ne the multiplication v g by v g ˆ v( gr) gr ˆ [ g]B (v P V . h P G. and let B be a basis of V. X X X .3 De®nition Let V be an FG-module. we have v( gr) P F n . ( g P G) . 0). Proof (1) We have already observed that for all u.

  ˆ (0. j Our next example illustrates part (1) of Theorem 4. then we have v1 a ˆ v2 . 0)a ˆ (1. F) (where n ˆ dim V ). [1]B ˆ [ g]B [ g À1 ]B for all g P G. Therefore each matrix [ g]B is invertible (with inverse [ gÀ1 ]B ).2(1). We have proved that the function g 3 [ g]B is a homomorphism from G to GL (n. a2 ˆ b2 . Since v( gh) ˆ (v g)h for all g. Now v1 ˆ v for all v P V. v2 b ˆ Àv2 X ( g P G) If B denotes the basis v1 . g P G)X For instance. br ˆ X À1 0 0 À1 Write V ˆ F 2.4(1). By Theorem 4. 0). it follows that [ gh]B ˆ [ g]B [h]B X In particular. then the representation is just the representation r (see Theorem 4.5 Examples (1) Let G ˆ D8 ˆ ka. (2) Let G ˆ Q8 ˆ ka. h P G and all v in the basis B of V. bÀ1 ab ˆ aÀ1 l. b: a4 ˆ b2 ˆ 1. so [1]B is the identity matrix. v2 is the basis (1. bÀ1 ab ˆ aÀ1 l and let r be the representation of G over F given in Example 3. 1)X 0 (1. and hence is a representation of G over F. so     0 1 1 0 ar ˆ .4. (0. v2 .FG-modules 41 (2) Let V be an FG-module with basis B .4(1) again). v2 a ˆ Àv1 . V becomes an FG-module if we de®ne v g ˆ v( gr) (v P V . 1) of V. b: a4 ˆ 1. In Example . 0) À1 1 0 If v1 . 4. g 3 [ g]B v1 b ˆ v1 .

. . . . that is. in Example 4.2(4) we de®ned Q8 to be the subgroup of GL (2. v n is a basis of V and g1 .4(1) we must this time take F ˆ C. v n of V and then extending the action to be linear on the whole of V. v2 such that v1 a ˆ iv1 . we shall show you various ways of constructing FG-modules directly.42 Representations and characters of groups 1. v1 b ˆ v2 . . . . . v1 b and v2 b determine v g for all v P V and g P G. (v1 ‡ 2v2 )ab ˆ v1 ab ‡ 2v2 ab ˆ v2 b À 2v1 b ˆ Àv2 À 2v1 X A similar remark holds for all FG-modules V: if v1 . there are restrictions on how we may de®ne the vectors v i g. . To illustrate Theorem 4. without using a representation. 1 < j < r) determine v g for all v P V and g P G. 4. and B ˆ Aˆ À1 0 0 Ài so we already have a representation of G over C. C) generated by     0 1 i 0 . Shortly. the vectors v1 a. we ®rst de®ne v i g for each i and each g in G. then the vectors v i g j (1 < i < n. v2 a ˆ Àiv2 . v2 b ˆ Àv1 X Notice that in the above examples. v2 a.5(1). . To do this. We then obtain a CG-module with basis v1 . . we turn a vector space V over F into an FG-module by specifying the action of group elements on a basis v1 . . The next result will often be used to show that our chosen multiplication turns V into an FG-module. that we have a multiplication v g for all v in V and g in G which (ë i P F) . and then de®ne (ë1 v1 ‡ X X X ‡ ë n v n ) g to be ë1 (v1 g) ‡ X X X ‡ ë n (v n g)X As you might expect. For instance. v n is a basis of a vector space V over F Suppose . gr generate G.6 Proposition Assume that v1 . F F F .

FG-modules

43

satis®es the following conditions for all i with 1 < i < n, for all g, h P G, and for all ë1 , F F F , ë n P F: (1) (2) (3) (4) v i g P V; v i ( gh) ˆ (v i g)h; vi 1 ˆ vi ; (ë1 v1 ‡ . . . ‡ ë n v n ) g ˆ ë1 (v1 g) ‡ . . . ‡ ë n (v n g).

Then V is an FG-module. Proof It is clear from (3) and (4) that v1 ˆ v for all v P V. Conditions (1) and (4) ensure that for all g in G, the function v 3 v g (v P V) is an endomorphism of V. That is, vg P V, (ëv) g ˆ ë(v g), (u ‡ v) g ˆ ug ‡ v g, for all u, v P V, ë P F and g P G. Hence (4X7) (ë1 u1 ‡ X X X ‡ ë n u n )h ˆ ë1 (u1 h) ‡ X X X ‡ ë n (un h) for all ë1 , . . . , ë n P F, all u1 , . . . , un P V and all h P G. Now let v P V and g, h P G. Then v ˆ ë1 v1 ‡ . . . ‡ ë n v n for some ë1 , . . . , ë n P F, and v( gh) ˆ ë1 (v1 ( gh)) ‡ X X X ‡ ë n (v n ( gh)) ˆ (ë1 (v1 g) ‡ X X X ‡ ë n (v n g))h ˆ (v g)h by condition (4)X by condition (4) ˆ ë1 ((v1 g)h) ‡ X X X ‡ ë n ((v n g)h) by condition (2) by (4X7)

We have now checked all the axioms which are required for V to be an FG-module. j Our next de®nitions translate the concepts of the trivial representation and a faithful representation into module terms. 4.8 De®nitions (1) The trivial FG-module is the 1-dimensional vector space V over F with vg ˆ v for all v P V , g P GX

44

Representations and characters of groups

(2) An FG-module V is faithful if the identity element of G is the only element g for which v g ˆ v for all v P V X For instance, the FD8 -module which appears in Example 4.5(1) is faithful. Our next aim is to use Proposition 4.6 to construct faithful FGmodules for all subgroups of symmetric groups. Permutation modules Let G be a subgroup of Sn , so that G is a group of permutations of {1, . . . , n}. Let V be an n-dimensional vector space over F, with basis v1 , . . . , v n . For each i with 1 < i < n and each permutation g in G, de®ne v i g ˆ v ig X Then v i g P V and v i 1 ˆ v i . Also, for g, h in G, v i ( gh) ˆ v i( gh) ˆ v(ig) h ˆ (v i g)hX We now extend the action of each g linearly to the whole of V; that is, for all ë1 , . . . , ë n in F and g in G, we de®ne (ë1 v1 ‡ X X X ‡ ë n v n ) g ˆ ë1 (v1 g) ‡ X X X ‡ ë n (v n g)X Then V is an FG-module, by Proposition 4.6. 4.9 Example Let G ˆ S4 and let B denote the basis v1 , v2 , v3 , v4 of V. If g ˆ (1 2), then v1 g ˆ v2 , v2 g ˆ v1 , v3 g ˆ v3 , v4 g ˆ v4 X And if h ˆ (1 3 4), then v1 h ˆ v3 , v2 h ˆ v2 , v3 h ˆ v4 , v4 h ˆ v1 X We have [ g]B 0 f1 ˆf d0 0 H 1 0 0 0 0 0 1 0 H I 0 0 f0 0g g, [h]B ˆ f d0 0e 1 1 0 1 0 0 1 0 0 0 I 0 0g gX 1e 0

FG-modules

45

4.10 De®nition Let G be a subgroup of Sn . The FG-module V with basis v1 , . . . , v n such that v i g ˆ v ig for all i, and all g P G, is called the permutation module for G over F. We call v1 , . . . , v n the natural basis of V. Note that if we write B for the basis v1 , . . . , v n of the permutation module, then for all g in G, the matrix [ g]B has precisely one nonzero entry in each row and column, and this entry is 1. Such a matrix is called a permutation matrix. Since the only element of G which ®xes every v i is the identity, we see that the permutation module is a faithful FG-module. If you are aware of the fact that every group G of order n is isomorphic to a subgroup of Sn , then you should be able to see that G has a faithful FG-module of dimension n. We shall go into this in more detail in Chapter 6. 4.11 Example Let G ˆ C3 ˆ ka: a3 ˆ 1l. Then G is isomorphic to the cyclic subgroup of S3 which is generated by the permutation (1 2 3). This alerts us to the fact that if V is a 3-dimensional vector space over F, with basis v1 , v2 , v3 , then we may make V into an FG-module in which v1 1 ˆ v1 , v2 1 ˆ v2 , v3 1 ˆ v3 , v1 a ˆ v2 , v2 a ˆ v3 , v3 a ˆ v1 , v1 a2 ˆ v3 , v2 a2 ˆ v1 , v3 a2 ˆ v2 X Of course, we de®ne v g, for v an arbitrary vector in V and g ˆ 1, a or a2 , by (ë1 v1 ‡ ë2 v2 ‡ ë3 v3 ) g ˆ ë1 (v1 g) ‡ ë2 (v2 g) ‡ ë3 (v3 g) for all ë1 , ë2 , ë3 P F. Proposition 4.6 can be used to verify that V is an FG-module, but we have been motivated by the de®nition of permutation modules in our construction. FG-modules and equivalent representations We conclude the chapter with a discussion of the relationship between FG-modules and equivalent representations of G over F. An FG-

46

Representations and characters of groups

module gives us many representations, all of the form g 3 [ g]B ( g P G) for some basis B of V. The next result shows that all these representations are equivalent to each other (see De®nition 3.3); and moreover, any two equivalent representations of G arise from some FG-module in this way. 4.12 Theorem Suppose that V is an FG-module with basis B , and let r be the representation of G over F de®ned by r: g 3 [ g]B ö: g 3 [ g]B 9 ( g P G)X ( g P G) (1) If B 9 is a basis of V, then the representation of G is equivalent to r. (2) If ó is a representation of G which is equivalent to r, then there is a basis B 0 of V such that ó : g 3 [ g]B 0 ( g P G)X

Proof (1) Let T be the change of basis matrix from B to B 9 (see De®nition 2.23). Then by (2.24), for all g P G, we have [ g]B ˆ T À1 [ g]B 9 TX Therefore ö is equivalent to r. (2) Suppose that r and ó are equivalent representations of G. Then for some invertible matrix T, we have gr ˆ T À1 ( gó )T for all g P GX Let B 0 be the basis of V such that the change of basis matrix from B to B 0 is T. Then for all g P G, [ g]B ˆ T À1 [ g]B 0 T , and so gó ˆ [ g]B 0 .
j

4.13 Example Again let G ˆ C3 ˆ ka: a3 ˆ 1l. There is a representation r of G which is given by

 1r ˆ

1 0

FG-modules     0 1 0 À1 À1 2 ,a rˆ , ar ˆ X À1 À1 1 1 0 

47

(To see this, simply note that (ar)2 ˆ a2 r and (ar)3 ˆ I; see Exercise 3.2.) If V is a 2-dimensional vector space over C, with basis v1 , v2 (which we call B ), then we can turn V into a CG-module as in Theorem 4.4(1) by de®ning v1 1 ˆ v1 , v2 1 ˆ v2 , v1 a ˆ v2 , v1 a2 ˆ Àv1 À v2 , v2 a2 ˆ v1 X  À1 X 0

v2 a ˆ Àv1 À v2 ,

We then have    1 0 0 [1]B ˆ , [a]B ˆ 0 1 À1 

 1 À1 2 , [a ]B ˆ À1 1

Now let u1 ˆ v1 and u2 ˆ v1 ‡ v2 . Then u1 , u2 is another basis of V, which we call B 9. Since u1 1 ˆ u 1 , u2 1 ˆ u 2 , u1 a ˆ Àu1 ‡ u2 , u2 a ˆ Àu1 , u1 a2 ˆ Àu2 , u2 a2 ˆ u1 À u2 ,

we obtain the representation ö: g 3 [ g]B 9 where       1 0 À1 1 0 À1 2 [1]B 9 ˆ , [a]B 9 ˆ , [a ]B 9 ˆ X 0 1 À1 0 1 À1 Note that if Tˆ then for all g in G, we have [ g]B ˆ T À1 [ g]B 9 T , and so r and ö are equivalent, in agreement with Theorem 4.12(1).  

1 0 1 1

Summary of Chapter 4 1. An FG-module is a vector space over F, together with a multiplication by elements of G on the right. The multiplication satis®es properties (1)±(5) of De®nition 4.2.

48

Representations and characters of groups

2. There is a correspondence between representations of G over F and FG-modules, as follows. (a) Suppose that r: G 3 GL (n, F) is a representation of G. Then F n is an FG-module, if we de®ne v g ˆ v( gr) (v P F n , g P G)X (b) If V is an FG-module, with basis B , then r: g 3 [ g]B is a representation of G over F.

3. If G is a subgroup of Sn , then the permutation FG-module has basis v1 , F F F , v n , and v i g ˆ v ig for all i with 1 < i < n, and all g in G. Exercises for Chapter 4 1. Suppose that G ˆ S3 , and that V ˆ sp (v1 , v2 , v3 ) is the permutation module for G over C, as in De®nition 4.10. Let B 1 be the basis v1 , v2 , v3 of V and let B 2 be the basis v1 ‡ v2 ‡ v3 , v1 À v2 , v1 À v3 . Calculate the 3 3 3 matrices [ g]B 1 and [ g]B 2 for all g in S3 . What do you notice about the matices [ g]B 2 ? 2. Let G ˆ Sn and let V becomes an FG-module & v, vg ˆ Àv, be a vector space over F. Show that V if we de®ne, for all v in V, if g is an even permutation, if g is an odd permutation.

3. Let Q8 ˆ ka, b: a4 ˆ 1, b2 ˆ a2 , bÀ1 ab ˆ aÀ1 l, the quaternion group of order 8. Show that there is an RQ8 -module V of dimension 4 with basis v1 , v2 , v3 , v4 such that v1 a ˆ v 2 , v1 b ˆ v 3 , v2 a ˆ Àv1 , v2 b ˆ v4 , v3 a ˆ Àv4 , v4 a ˆ v3 , and v3 b ˆ Àv1 , v4 b ˆ Àv2 X

4. Let A be an n 3 n matrix and let B be a matrix obtained from A by permuting the rows. Show that there is an n 3 n permutation matrix P such that B ˆ PA. Find a similar result for a matrix obtained from A by permuting the columns.

5 FG-submodules and reducibility

We begin the study of FG-modules by introducing the basic building blocks of the theory ± the irreducible FG-modules. First we require the notion of an FG-submodule of an FG-module. Throughout, G is a group and F is R or C. FG-submodules 5.1 De®nition Let V be an FG-module. A subset W of V is said to be an FGsubmodule of V if W is a subspace and wg P W for all w P W and all g P G. Thus an FG-submodule of V is a subspace which is also an FGmodule. 5.2 Examples (1) For every FG-module V, the zero subspace {0}, and V itself, are FG-submodules of V. (2) Let G ˆ C3 ˆ ka: a3 ˆ 1l, and let V be the 3-dimensional FGmodule de®ned in Example 4.11. Thus, V has basis v1 , v2 , v3 , and v1 1 ˆ v1 , v2 1 ˆ v2 , v3 1 ˆ v3 , v1 a ˆ v2 , v2 a ˆ v3 , v3 a ˆ v1 , v1 a2 ˆ v3 , v2 a2 ˆ v1 , v3 a2 ˆ v2 X Put w ˆ v1 ‡ v2 ‡ v3, and let W ˆ sp (w), the 1-dimensional subspace spanned by w. Since 49

50

Representations and characters of groups w1 ˆ wa ˆ wa2 ˆ w,

W is an FG-submodule of V. However, sp (v1 ‡ v2 ) is not an FGsubmodule, since (v1 ‡ v2 )a ˆ v2 ‡ v3 P sp (v1 ‡ v2 )X a Irreducible FG-modules 5.3 De®nition An FG-module V is said to be irreducible if it is non-zero and it has no FG-submodules apart from {0} and V. If V has an FG-submodule W with W not equal to {0} or V, then V is reducible. Similarly, a representation r: G 3 GL (n, F) is irreducible if the corresponding FG-module F n given by v g ˆ v( gr) (v P F n , g P G) (see Theorem 4.4(1)) is irreducible; and r is reducible if F n is reducible. Suppose that V is a reducible FG-module, so that there is an FGsubmodule W with 0 , dim W , dim V. Take a basis B 1 of W and extend it to a basis B of V. Then for all g in G, the matrix [ g]B has the form H I Xg 0 d e (5X4) Yg Z g for some matrices Xg , Yg and Zg , where Xg is k 3 k (k ˆ dim W). A representation of degree n is reducible if and only if it is equivalent to a representation of the form (5.4), where X g is k 3 k and 0 , k , n. Notice that in (5.4), the functions g 3 Xg and g 3 Z g are representations of G: to see this, let g, h P G and multiply the matrices [ g]B and [h]B given by (5.4). Notice also that if V is reducible then dim V > 2. 5.5 Examples (1) Let G ˆ C3 ˆ ka: a3 ˆ 1l and let V be the 3-dimensional FGmodule with basis v1 , v2 , v3 such that

FG-submodules and reducibility v1 a ˆ v2 , v2 a ˆ v3 , v3 a ˆ v1 ,

51

as in Example 4.11. We saw in Example 5.2(2) that V is a reducible FG-module, and has an FG-submodule W ˆ sp (v1 ‡ v2 ‡ v3 ). Let B be the basis v1 ‡ v2 ‡ v3 , v1 , v2 of V. Then H H I I 1 0 0 1 0 0 f f g g [1]B ˆ f 0 1 0 g, [a]B ˆ f 0 0 1 g, d d e e 0 0 1 1 À1 À1 H I 1 0 0 f g [a2 ]B ˆ f 1 À1 À1 gX d e 0 1 0 This reducible representation gives us two other representations: at the `top left' we have the trivial representation and at the `bottom right' we have the representation which is given by       0 1 1 0 À1 À1 2 ,a 3 ,a3 X 13 À1 À1 0 1 1 0 (2) Let G ˆ D8 and let V ˆ F 2 be the 2-dimensional FG-module described in Example 4.5(1). Thus G ˆ ka, bl, and for all (ë, ì) P V we have (ë, ì)a ˆ (Àì, ë), (ë, ì)b ˆ (ë, Àì)X

We claim that V is an irreducible FG-module. To see this, suppose that there is an FG-submodule U which is not equal to V. Then dim U < 1, so U ˆ sp ((á, â)) for some á, â P F. As U is an FGmodule, (á, â)b is a scalar multiple of (á, â), and hence either á ˆ 0 or ⠈ 0. Since (á, â)a is also a scalar multiple of (á, â), this forces á ˆ ⠈ 0, so U ˆ {0}. Consequently V is irreducible, as claimed.

Summary of Chapter 5 1. If V is an FG-module, and W is a subspace of V which is itself an FG-module, then W is an FG-submodule of V. 2. The FG-module V is irreducible if it is non-zero and the only FGsubmodules are {0} and V.

Let G ˆ C2 ˆ ka: a ˆ 1l. 2. and let V ˆ F 2.5 are irreducible? 4. (b) Suppose that å and ç are complex cube roots of unity. á). Which of the four representations of D12 de®ned in Exercise 3. Let r and ó be equivalent representations of the group G over F. Find a CG-module which is neither reducible nor irreducible. 2 . br ˆ . b. Prove that if r is reducible then ó is reducible. ab ˆ ba. and let G ˆ ka. For (á. â) and (á. De®ne the permutations a. b ˆ (4 5 6). Prove that there is a representation r of G over C such that       å 0 ç 0 0 1 ar ˆ X . cl. cr ˆ 0 å À1 0 çÀ1 1 0 (c) For which values of å. â) P V. 3. c ˆ (2 3)(4 5). ç is r irreducible? 5. ç is r faithful? (d) For which values of å. â)a ˆ (â. c P S6 by a ˆ (1 2 3). cÀ1 ac ˆ aÀ1 and cÀ1 bc ˆ bÀ1 X Deduce that G has order 18.52 Representations and characters of groups Exercises for Chapter 5 1. Let G ˆ C13 . b. Verify that V is an FG-module and ®nd all the FG-submodules of V. â)1 ˆ (á. (a) Check that a3 ˆ b3 ˆ c2 ˆ 1. de®ne (á.

In particular. and let F be R or C. g n . Take as the elements of FG all expressions of the form ë1 g 1 ‡ X X X ‡ ë n g n (all ë i P F)X The rules for addition and scalar multiplication in FG are the natural ones: namely. . After de®ning the group algebra of G. We de®ne a vector space over F with g1 .6 Group algebras The group algebra of a ®nite group G is a vector space of dimension |G| which also carries extra structure involving the product operation on G. . . which will be explored in greater detail later on. . Group algebras are therefore of great interest. . we shall use it to construct an important faithful representation. and we call this vector space FG. the ultimate goal of representation theory ± that of understanding all the representations of ®nite groups ± would be achieved if group algebras could be fully analysed. . . g n as a basis. and ë P F. . then 53 . if uˆ n ˆ iˆ1 ë i g i and v ˆ n ˆ iˆ1 ìi g i are elements of FG. group algebras are the source of all you need to know about representation theory. In a sense. known as the regular representation of G. The group algebra of G Let G be a ®nite group whose elements are g1 .

. then uv ˆ (e À a ‡ 2a2 )(1 e ‡ 5a) 2 ˆ ˆ 1 2e ‡ 5a À 1 a À 5a2 ‡ a2 ‡ 10a3 2 ‡ 9 a À 4a2 X 2 21 2e . The basis g1 . ˆˆ (ë h ì hÀ1 g ) g gPG hPG 6. .1. . . we write e for the identity element of G. g n .hPG ˆ where all ë g .54 Representations and characters of groups u‡vˆ n n ˆ ˆ (ë i ‡ ì i ) g i and ëu ˆ (ëë i ) g i X iˆ1 iˆ1 With these rules. ì h P F. 1 u ˆ 1 e À 1 a ‡ 2 a2 X 2 3 3 3 3 Sometimes we write elements of FG in the form ˆ ë g g (ë g P F)X gPG Now. (To avoid confusion with the element 1 of F. with basis g1 . 6. FG carries more structure than that of a vector space ± we can use the product operation on G to de®ne multiplication in FG as follows: 3 2ˆ 32ˆ ˆ ëg g ìh h ˆ ë g ì h ( gh) gPG hPG g. FG is a vector space over F of dimension n.1 Example Let G ˆ C3 ˆ ka: a3 ˆ el. .) The vector space CG contains u ˆ e À a ‡ 2a2 and v ˆ 1 e ‡ 5aX 2 We have u ‡ v ˆ 3 e ‡ 4a ‡ 2a2 . . v are the elements of CG which appear in Example 6. in this example. g n is called the natural basis of FG.2 Example If G ˆ C3 and u. . .

(2) Let ˆ ˆ ˆ rˆ ë g g.hPG 55 (ë g .Group algebras 6.h. for all r.h. t P FG and ë P F: (1) (2) (3) (4) (5) (6) (7) rs P FG. s. ì g . r(st) ˆ (rs)t. 6. ì h P F). r1 ˆ 1r ˆ r.kPG ë g ì h í k ( gh)k ë g ì h í k g(hk) ˆ g. Then (rs)t ˆ ˆ ˆ g. í g P F). s ˆ ì g g. (ër)s ˆ ë(rs) ˆ r(ës). (r ‡ s)t ˆ rt ‡ st. The group algebra FG contains an identity for multiplication.3 De®nition The vector space FG.kPG ˆ r(st)X We leave the proofs of the other equations as easy exercises. Proof (1) It follows immediately from the de®nition of rs that rs P FG. r(s ‡ t) ˆ rs ‡ rt. is called the group algebra of G over F. j .4 Proposition Multiplication in FG satis®es the following properties. namely the element 1e (where 1 is the identity of F and e is the identity of G). We write this element simply as 1. gPG gPG gPG (ë g . t ˆ í g g. r0 ˆ 0r ˆ 0. with multiplication de®ned by 3 2ˆ 32ˆ ˆ ëg g ìh h ˆ ë g ì h ( gh) gPG hPG g.

j 6. For all u.4.5 De®nition Let G be a ®nite group and F be R or C. with the natural multiplication v g (v P FG. 6. so g ˆ 1. 6. Then 1 g ˆ 1. (3).7 Example Let G ˆ C3 ˆ ka: a3 ˆ el. (ëv) g ˆ ë(v g). any vector space equipped with a multiplication satisfying properties (1)±(7) of Proposition 6. v1 ˆ v. (4) and (5) of Proposition 6.6 Proposition The regular FG-module is faithful. g P G). Let V ˆ FG. ë P F and g.56 Representations and characters of groups In fact. but it is worth pointing out that the axioms for an algebra mean that it is both a vector space and a ring. v( gh) ˆ (v g)h. The representation g 3 [ g]B obtained by taking B to be the natural basis of FG is called the regular representation of G over F. The regular FG-module We now use the group algebra to de®ne an important FG-module. is called the regular FG-module. by parts (1). We shall be concerned only with group algebras. so that V is a vector space of dimension n over F. (u ‡ v) g ˆ ug ‡ v g. respectively. The vector space FG. we have vg P V. Therefore V is an FG-module. (2). where n ˆ |G|. h P G. Proof Suppose that g P G and v g ˆ v for all v P FG. Note that the regular FG-module has dimension equal to |G|. The elements of FG have the form . v P V.4 is called an algebra. and the result follows.

Group algebras ë1 e ‡ ë2 a ‡ ë3 a2 We have (ë1 e ‡ ë2 a ‡ ë3 a2 )e ˆ ë1 e ‡ ë2 a ‡ ë3 a2 . Now. it is sometimes helpful to extend the de®nition of the multiplication so that we have an element vr of V for all elements r in the group algebra FG. and that v P V and r P FG. v2 r ˆ ëv1 ‡ ìv2 . say € r ˆ gPG ì g g (ì g P F). (ë1 e ‡ ë2 a ‡ ë3 a2 )a ˆ ë3 e ‡ ë1 a ‡ ë2 a2 . a. as described in Example 4. a2 of FG. If r ˆ ë(1 2) ‡ ì(1 3 4) then v1 r ˆ ëv1 (1 2) ‡ ìv1 (1 3 4) ˆ ëv2 ‡ ìv3 . 6. a2 3 d 1 0 0 eX 0 0 1 1 0 0 0 1 0 FG acts on an FG-module You will remember that an FG-module is a vector space over F. ì P F) . This is done in the following natural way.9 Examples (1) Let V be the permutation module for S4 .8 De®nition Suppose that V is an FG-module.9. a 3 d 0 0 1 e. we obtain the regular representation of G: H I H I H I 1 0 0 0 1 0 0 0 1 e 3 d 0 1 0 e. (ë1 e ‡ ë2 a ‡ ë3 a2 )a2 ˆ ë2 e ‡ ë3 a ‡ ë1 a2 X (ë i P F)X 57 By taking matrices relative to the basis e. (2v1 ‡ v2 )r ˆ ëv1 ‡ (2ë ‡ ì)v2 ‡ 2ìv3 X (ë. De®ne vr by ˆ vr ˆ ì g (v g)X gPG 6. together with a multiplication v g for v P V and g P G (and the multiplication satis®es various axioms).

(5).3. assuming the other parts. v P V. Compare the next result with Proposition 6. the element vr is simply the product of v and r as elements of the group algebra.58 Representations and characters of groups (2) If V is the regular FG-module. then for all v P V and r P FG. s ˆ ì h hX gPG hPG Then 2ˆ 3 v(rs) ˆ v ë g ì h ( gh) g. v(r ‡ s) ˆ vr ‡ vs. v(rs) ˆ (vr)s. and let r.4. all ë P F and all r.h ë g ì h (v( gh)) ë g ì h ((v g)h) by (4) and (6) ˆ g. s P FG: (1) (2) (3) (4) (5) (6) (7) vr P V. 6.h ˆ ˆ ˆ ˆ g.h 2ˆ g 3 32ˆ ë g (v g) ìh h h by (4). We shall give a proof of part (2). (6) j ˆ (vr)sX . given by De®nition 6. Then the following properties hold for all u. Let v P V.10 Proposition Suppose that V is an FG-module. s P FG with ˆ ˆ rˆ ë g g. v1 ˆ v. (u ‡ v)r ˆ ur ‡ vr. v0 ˆ 0r ˆ 0. (ëv)r ˆ ë(vr) ˆ v(ër). and we leave them to you. Proof All parts except (2) are straightforward.

What is the matrix [W]B . where the symbol 0 is used for the zero elements of V and FG. 3. Show that zg ˆ gz for all g in G. . Assume that G is a ®nite group. b: a4 ˆ b2 ˆ 1. r P FG such that vr ˆ 0. The vector space FG. . (a) Let x and y be the following elements of CG: x ˆ a ‡ 2a2 . (b) Deduce that c2 ˆ |G|c. Suppose that G ˆ D8 ˆ ka. gn of CG? 5. For r and s in CG. The regular FG-module is faithful. 2. but neither v nor r is 0. and has a natural multiplication de®ned on it. and write c €n for the element iˆ1 g i of CG. If V is an FG-module. The group algebra FG of G over F consists of all linear combinations of elements of G. . .Group algebras Summary of Chapter 6 59 1. Deduce that zr ˆ rz for all r in CG. does rs ˆ 0 imply that r ˆ 0 or s ˆ 0? 4. 2. with |G| . Work out matrices for the regular representation of C2 3 C2 over F. . . prove from the de®nition that 0r ˆ 0 for all r P FG. Show that for every ®nite group G. where B is the basis g1 . Let G ˆ C2 . (a) Prove that ch ˆ hc ˆ c for all h in G. yx and x 2 . 3. g n }. say G ˆ { g1 . 1. Exercises for Chapter 6 1. y ˆ b ‡ ab À a2 X Calculate xy. (c) Let W: CG 3 CG be the linear transformation sending v to vc for all v in CG. . . bÀ1 ab ˆ aÀ1 l. with the natural multiplication v g (v P FG. there exists an FG-module V and elements v P V. (b) Let z ˆ b ‡ a2 b. and v0 ˆ 0 for all v P V . . g P G) is the regular FG-module.

is an irreducible CG-submodule of the regular CG-module. Suppose that G ˆ D6 ˆ ka.60 Representations and characters of groups 6. bÀ1 ab ˆ aÀ1 l. and let ù ˆ e2ðia3 . b ‡ ù2 ab ‡ ùa2 b). . Prove that the 2-dimensional subspace W of CG. b: a3 ˆ b2 ˆ 1. de®ned by W ˆ sp (1 ‡ ù2 a ‡ ùa2 .

61 . then for all v P V and r ˆ gPG ë g g P FG. The analogous functions for FG-modules are called FG-homomorphisms. respectively. 7.1 De®nition Let V and W be FG-modules. submodule of W. FG-homomorphisms 7. we have (vr)W ˆ (vW)r since (vr)W ˆ ˆ gPG ë g (v g)W ˆ ˆ gPG ë g (vW) g ˆ (vW)rX The next result shows that FG-homomorphisms give rise to FGsubmodules in a natural way.7 FG-homomorphisms For groups and vector spaces. and we introduce these in this chapter. if W sends v to w then it sends v g to wg. g P GX In other words. Then Ker W is an FG-submodule of V and Im W is an FG. A function W: V 3 W is said to be an FG-homomorphism if W is a linear transformation and (v g)W ˆ (vW) g for all v P V .2 Proposition Let V and W be FG-modules and let W: V 3 W be an FG-homomorphism. group homomorphisms and linear transformations. the `structure-preserving' functions are. Note that if G is a ®nite group and W: V 3 W is an FG-homomorph€ ism.

(2) Let ë P F. and de®ne W: V 3 V by vW ˆ ëv for all v P V. Then W is an FG-homomorphism. For all g P G. and for all € v ˆ ë i v i P V and all g P G.62 Representations and characters of groups Proof First note that Ker W is a subspace of V and Im W is a subspace of W. and 2 3 2 3 ˆ ˆ (vW) g ˆ ë i wg ˆ ë i wX Therefore W is an FG-homomorphism. wg ˆ (vW) g ˆ (v g)W P Im W. (3) Suppose that G is a subgroup of S n . . Let v P Ker W and g P G. and Ker W ˆ V. . De®ne 2 n 3 n ˆ ˆ W: ëivi 3 ë i w (ë i P F)X iˆ1 iˆ1 Thus v i W ˆ w for all i. Here.3 Examples (1) If W: V 3 W is de®ned by vW ˆ 0 for all v P V. Then (v g)W ˆ (vW) g ˆ 0 g ˆ 0. and so Im W is an FG-submodule of W. we have 2 3 2 3 ˆ ˆ (v g)W ˆ ë i v ig W ˆ ë i w. Let V ˆ sp (v1 . so vg P Ker W. @ n A n ˆ ˆ Ker W ˆ ëi vi : ëi ˆ 0 . j 7. We construct an FG-homomorphism W from V to W. Therefore Ker W is an FG-submodule of V. Provided ë Tˆ 0. iˆ1 iˆ1 Im W ˆ W X . . so that w ˆ vW for some v P V. Now let w P Im W.8).10). Im W ˆ V. we have Ker W ˆ {0}. since W is a linear transformation. v n ) be the permutation module for G over F (see De®nition 4. . Then W is a linear transformation. then W is an FGhomomorphism. and let W ˆ sp (w) be the trivial FG-module (see De®nition 4. Im W ˆ {0}.

. then the inverse WÀ1 : W 3 V is also an FG-isomorphism. then we say that V and W are isomorphic FG-modules and write V  W.FG-homomorphisms 63 By Proposition 7. . (3) V contains a trivial FG-submodule if and only if W contains a trivial FG-submodule (since X is a trivial FG-submodule of V if and only if XW is a trivial FG-submodule of W). For w P W and g P G. v n W is a basis of W).4 De®nition Let V and W be FG-modules. . If there is such an FG-isomorphism. we check that if V  W then W  V. We list some examples below: (1) dim V ˆ dim W (since v1 . as required.2. We call a function W: V 3 W an FGisomorphism if W is an FG-homomorphism and W is invertible. 7. Then we may use W and WÀ1 to translate back and forth between the isomorphic FGmodules V and W. . ((wWÀ1 ) g)W ˆ ((wWÀ1 )W) g ˆ wg ˆ ((wg)WÀ1 )WX Hence (wWÀ1 ) g ˆ (wg)WÀ1 . In the next result. . . v n is a basis of V if and only if v1 W. Isomorphic FG-modules 7. Ker W is an FG-submodule of the permutation module V. . (2) V is irreducible if and only if W is irreducible (since X is an FGsubmodule of V if and only if XW is an FG-submodule of W). .5 Proposition If W: V 3 W is an FG-isomorphism. Proof Certainly WÀ1 is an invertible linear transformation. j as W is an FG-homomorphism Suppose that W: V 3 W is an FG-isomorphism. . so we need only show that WÀ1 is an FG-homomorphism. and prove that V and W share the same structural properties.

7). suppose that v1 . it follows that [ g]B 1 ˆ [ g]B 2 . . De®ne a representation ö of G by ö: g 3 [ g]B 1 . w n is a basis B 2 of W such that [ g]B 1 ˆ [ g]B 2 for all g P G.12(1).64 Representations and characters of groups Just as we often regard isomorphic groups as being identical. . . . . . Then by Theorem 4. v n is a basis B 1 of V and w1. Let g P G. v n W is a basis B 2 of W. we show that isomorphic FG-modules correspond to equivalent representations. 7. For the moment. Now assume that V and W are isomorphic FG-modules. and let v1 .12(2). . This completes the proof of (7. then v1 W. we frequently disdain to distinguish between isomorphic FG-modules. we continue simply to emphasize the similarity between isomorphic FG-modules. Then by Theorem 4. . . Let g P G. we deduce that (v i g)W ˆ (v i W) g for all i. Proof We ®rst establish the following fact: (7. suppose ®rst that W is an FG-isomorphism from V to W. and hence W is an FG-isomorphism. .7) The FG-modules V and W are isomorphic if and only if there are a basis B 1 of V and a basis B 2 of W such that [ g]B 1 ˆ [ g]B 2 for all g P GX To see this. Conversely. Since (v i g)W ˆ (v i W) g for each i. . . though. and W is an FGmodule with basis B 9. there are a basis B 1 of V and a basis B 2 of W such that [ g]B 1 ˆ [ g]B 2 for all g P G. Let W be the invertible linear transformation from V to W for which v i W ˆ w i for all i. . ö is equivalent to both r and ó. Conversely. .6 Theorem Suppose that V is an FG-module with basis B . By (7. suppose that r and ó are equivalent. .7). Hence r and ó are equivalent. Then V and W are isomorphic if and only if the representations r: g 3 [ g]B and ó : g 3 [ g]B 9 are equivalent. Since [ g]B 1 ˆ [ g]B 2 . v n be a basis B 1 of V. there is a basis B 0 of V such that gó ˆ [ g]B 0 for all g P G. In the next result. . .

with basis v1 .8 Example Let G ˆ ka: a3 ˆ 1l. d d e e 0 0 1 1 0 0 H 0 0 0 1 1 0 I f [a2 ]B 9 ˆ f 1 d 0 g 0 gX e Compare the FG-module V de®ned in Example 4. v2 a ˆ v3 .FG-homomorphisms 65 that is. Therefore V and W are isomorphic FG-modules. 7.4(1) we encountered two equivalent representations r and ó of G. a cyclic group of order 3. br ˆ À1 0 0 À1 and aó ˆ  i 0   0 0 . call it B 9. We have H H I I 1 0 0 0 1 0 f f g g [1]B 9 ˆ f 0 1 0 g. v2 .7). bÀ1 ab ˆ aÀ1 l. a2 is a basis of W.9 Example Let G ˆ D8 ˆ ka. v2 . [ g]B 9 ˆ [ g]B 0 for all g P G. a. by (7. bó ˆ Ài 1  1 X 0 (ë i P F) . Indeed. j 7. In Example 3. and let W denote the regular FG-module.11. Then 1. where     0 1 1 0 ar ˆ . the FG-modules V and W are therefore isomorphic. b: a4 ˆ b2 ˆ 1.7). the function W: ë1 v1 ‡ ë2 v2 ‡ ë3 v3 3 ë1 1 ‡ ë2 a ‡ ë3 a2 is an FG-isomorphism from V to W. we have [ g]B ˆ [ g]B 9 for all g P GX According to (7. [a]B 9 ˆ f 0 0 1 g. v3 a ˆ v1 X Writing B for the basis v1 . v3 such that v1 a ˆ v2 . v3 of V.

if we write B for the basis v1 . w1. . . w n is a basis B of V. . a direct sum of FG-submodules Ui. . w2 a ˆ Àiw2 . È Ur. 2. w n be a basis B 2 of W. v2 for which v1 a ˆ v2 . .9). and. w2 of W. and for g P G. the CG-modules V and W are isomorphic. and hence W is a CG-isomorphism from V to W. Then by (2. . u1 . Let u1 . . then for all g P G we have r: g 3 [ g]B and ó : g 3 [ g]B 9 X According to Theorem 7. and we show that these give rise to FG-homomorphisms. . . . . w2 for which w1 a ˆ iw1 . . and w1. if V ˆ U1 È . H I 0 [ g]B 1 eX [ g]B ˆ d 0 [ g]B 2 More generally. . To verify this directly.) Direct sums We conclude the chapter with a discussion of direct sums of FGmodules. where U and W are FG-submodules of V. .66 Representations and characters of groups Let V be the CG-module with basis v1 . w2 b ˆ w1 Thus. . and suppose that V ˆ U È W. w1 b ˆ w2 . . . u m be a basis B 1 of U. let W be the CG-module with basis w1. v2 b ˆ Àv2 (see Example 4. . u m . . . Let V be an FG-module. B r to .5(1)). in a similar way. . v2 of V and B 9 for the basis w1. v2 3 iw1 À iw2 X Then (v j a)W ˆ (v j W)a and (v j b)W ˆ (v j W)b for j ˆ 1.4(1). then we can amalgamate B 1 .6. and B i is a basis of Ui. let W: V 3 W be the invertible linear transformation such that W: v1 3 w1 ‡ w2 . since r and ó are equivalent. v1 b ˆ v1 . . (Compare Example 3. v2 a ˆ Àv1 .

v ˆ u1 ‡ . sum of some of the FG-submodules Ui. and we de®ne ð i : V 3 V (1 < i < r) by setting vð i ˆ ui X Then each ð i is an FG-homomorphism. Proof Clearly ð i is a linear transformation. i We now present a technical result concerning sums of irreducible FG-modules which will be used at a later stage. and is also a projection of V . and ð i is an FG-homomorphism. and suppose that V ˆ U1 ‡ X X X ‡ Ur . . vð 2 ˆ ui ð i ˆ ui ˆ vð i . and suppose that V ˆ U1 È X X X È U r where each Ui is an FG-submodule of V. j . since for v P V with v ˆ u1 ‡ . we have (v g)ð i ˆ (u1 g ‡ X X X ‡ ur g)ð i ˆ ui g ˆ (vð i ) gX Also. Thus ð i is a projection (see De®nition 2. . 7. i so ð2 ˆ ð i . ‡ u r (u j P U j for all j). . 7. . ‡ ur for unique vectors ui P Ui. For v P V we have . and for g P G. where each Ui is an irreducible FG-submodule of V Then V is a direct .12 Proposition Let V be an FG-module. H [ g]B 1 f FF [ g]B ˆ d (7X10) F 67 I g eX 0 [ g]B r 0 The next result shows that direct sums give rise naturally to FGhomomorphisms.FG-homomorphisms obtain a basis B of V.30). and g P G.11 Proposition Let V be an FG-module.

.68 Representations and characters of groups Proof The idea is to choose as many as we can of the FG-submodules U1. therefore W ’ Ui ˆ Ui. but W 1 ‡ X X X ‡ W s ‡ U i is not direct. . Ur } which has the properties that W 1 ‡ X X X ‡ W s is direct (iXeX equal to W 1 È X X X È W s ). Kernels and images of FG-homomorphisms are FG-modules. V r are FG-modules. X X X . Ur so that the sum of our chosen FG-submodules is direct. we have V ˆ W ˆ W 1 È F F F ÈW s . then W is an FG-homomorphism. . Let U. and let W: U 3 V and ö: V 3 W be FG-homomorphisms. . we remark that if V1 . choose a subset Y ˆ {W1. Exercises for Chapter 7 1. . g P G. so W ’ Ui Tˆ {0}. X X X . . v r g) for all v i P Vi (1 < i < r) and all g P G. j Finally. 2. and so Ui # W. But W ’ Ui a is an FG-submodule of Ui. . . If Ui P Y this is clear. so assume that Ui P Y. . V and W be FG-modules. To this end. if Ui P Y X a Let W ˆ W1 ‡ X X X ‡ WsX We claim that Ui # W for all i. v r ) g ˆ (v1 g. as required. as claimed. Since U i  W for all i with 1 < i < r. then we can make the external direct sum V1 È F F F ÈVr (see Chapter 2) into an FGmodule by de®ning (v1 . Ws } of {U1. . . Then W ‡ Ui is not a direct sum. Isomorphic FG-modules correspond to equivalent representations. 3. Prove that Wö: U 3 W is an FG-homomorphism. . If V and W are FG-modules and W: V 3 W is a linear transformation which satis®es (v g)W ˆ (vW) g for all v P V. . F F F . Summary of Chapter 7 1. and Ui is irreducible.

Suppose that V and W are isomorphic FG-modules. Prove that V0 and W0 are isomorphic FG-modules. 3. Show that the function ˆ W: v 3 v g (v P V ) gPG is an FG-homomorphism from V to V0 . Prove that the subset V0 ˆ fv P V : v g ˆ v for all g P Gg is an FG-submodule of V. Is the permutation module for G over F isomorphic to the regular FG-module? 6. Let G be the subgroup of S4 which is generated by (1 2) and (3 4). Prove that the permutation module for G over F is isomorphic to the regular FG-module. Let G ˆ C2 ˆ kx: x 2 ˆ 1l. De®ne the FGsubmodules V0 and W0 of V and W as in Exercise 3. Let G be the subgroup of S5 which is generated by (1 2 3 4 5). â P F) is an FG-homomorphism from the regular FG-module to itself. Assume that V is an FG-module. Is it necessarily surjective? 4. (a) Show that the function W: á1 ‡ âx 3 (á À â)(1 À x) (á. 5. (c) Find a basis B of FG such that   2 0 [W]B ˆ X 0 0 .FG-homomorphisms 69 2. (b) Prove that W2 ˆ 2W.

) This essentially reduces representation theory to the study of irreducible FG-modules. namely Maschke's Theorem. then there is an FG-submodule W of V such that V ˆ U È WX Before proving Maschke's Theorem.1 Maschke's Theorem Let G be a ®nite group. There are many subspaces W of V such that V ˆ U È W. and let V be an FG-module.2(2) below. A consequence of this theorem is that every FGmodule is a direct sum of irreducible FG-submodules. But there is.10). Maschke's Theorem 8. Put u ˆ v1 ‡ v2 ‡ v3 and U ˆ sp (u)X Then U is an FG-submodule of V. 8. for instance sp (v2 . since ug ˆ u for all g P G.2 Examples (1) Let G ˆ S3 and let V ˆ sp (v1 . If U is an FG-submodule of V. where as usual F ˆ R or C. v3 ) be the permutation module for G over F (see De®nition 4. We shall ®nd this W in an 70 .8 Maschke's Theorem We now come to our ®rst major result in representation theory. we illustrate it with some examples. only one FGsubmodule W of V with V ˆ U È W. in fact. v2 . v2 À 2v3 ). v3 ) and sp (v1 . (The assumption on F is important ± see Example 8. let F be R or C.

X X X . . v n of V. . we have v ˆ u ‡ w for unique vectors u P U and w P W 0. v n ). For v P V and x P G. . . We aim to modify the projection ö to create an FG-homomorphism from V to V with image U. . let G ˆ C p ˆ ka: a p ˆ 1l. (2) The conclusion of Maschke's Theorem can fail if F is not R or C. let p be a prime number.Maschke's Theorem 71 example after proving Maschke's Theorem (but you may like to look for it yourself now). . Check that the function   1 0 j a 3 ( j ˆ 0. as can easily be seen. 1. . v m of U. . where. p À 1) j 1 is a representation from G to GL (2. The corresponding FG-module is V ˆ sp (v1 . . Proof of Maschke's Theorem 8. extend it to a basis v1 . . F).1 We are given U. and let W0 ˆ sp (v m‡1 . v2 ). an FG-submodule of the FG-module V. v2 a j ˆ jv1 ‡ v2 X Clearly. and take F to be the ®eld of integers modulo p. By Proposition 2. de®ne W: V 3 V by (8X3) vW ˆ 1 ˆ v gö g À1 jGj gPG (v P V )X It is clear that W is an endomorphism of V and Im W # U. since U is the only 1-dimensional FG-submodule of V. . for 0 < j < p À 1.29. We show ®rst that W is an FG-homomorphism. For example. and we de®ne ö: V 3 V by setting vö ˆ u. . v1 a j ˆ v1 . Choose any subspace W0 of V such that V ˆ U È W0X (There are many choices for W0 ± simply take a basis v1 . To this end. U ˆ sp (v1 ) is an FG-submodule of V. But there is no FGsubmodule W such that V ˆ U È W.) For v P V. (vx)W ˆ 1 ˆ (vx) gö g À1 X jGj gPG . ö is a projection of V with kernel W0 and image U.

3) is W: v i 3 1(v1 ‡ v2 ‡ v3 ) (i ˆ 1. Hence 1 ˆ (vx)W ˆ vhöhÀ1 x jGj hPG 2 3 1 ˆ ˆ vhöhÀ1 x jGj hPG ˆ (vW)xX Thus W is an FG-homomorphism.2(1). First. Consequently W2 ˆ W. so by (8. v2 . we prove that W2 ˆ W. we have ug P U. Then W is an FG-submodule of V by Proposition 7.) . and so (ug)ö ˆ ug.32. Moreover.3(3). We use the proof of Maschke's Theorem to ®nd an FG-submodule W of V such that V ˆ U È W. Then vW P U. the FG-submodule constructed in Example 7. v3 ) be the permutation module. First note that for u P U. Using this.2. The projection ö onto U is given by ö: v1 3 0. This completes the proof of Maschke's Theorem. j 8. as claimed. v3 3 v1 ‡ v2 ‡ v3 X Check now that the FG-homomorphism W given by (8. and V ˆ U È W by Proposition 2. let W0 ˆ sp (v1 .5 Example Let G ˆ S3 and let V ˆ sp (v1 . as in Example 8.4) we have (vW)W ˆ vW. 1 ˆ 1 ˆ 1 ˆ uW ˆ (8X4) ugö g À1 ˆ (ug) g À1 ˆ u ˆ uX jGj gPG jGj gPG jGj gPG Now let v P V. (8. We have now established that W: V 3 V is a projection and an FGhomomorphism. Then V ˆ U È W0 (but of course W0 is not an FG-submodule). v2 3 0. v2 ). 2. with submodule U ˆ sp (v1 ‡ v2 ‡ v3 ). Next. Let W ˆ Ker W. so W ˆ sp (v1 À v2 . W ˆ ë i v i : ë i ˆ 0 . so does h ˆ xg. g P G. 3)X 3 The required FG-submodule W is then Ker W. v2 À v3 )X Ȁ É € (In fact.4) shows that Im W ˆ U.72 Representations and characters of groups As g runs over the elements of G.

g 3d Yg Zg for some matrices X g .4)).Maschke's Theorem 73 Note that if B is the basis v1 ‡ v2 ‡ v3 . v1 À v2 . If instead we use v1 ‡ v2 ‡ v3 . v2 À v3 ) is also an FG-submodule of V. This example illustrates the matrix version of Maschke's Theorem: for an arbitrary ®nite group G. suppose that r is a reducible representation of a ®nite group G over F of degree n. To put this another way. Z g . v2 of V. v2 À v3 as a basis B 9. then for all g P G. v1 . then we can ®nd a basis B 9 such that [ g]B 9 has the form H I à 0 d e 0 à for all g P G. 0 j j because sp (v1 À v2 . Then we know that r is equivalent to a representation of the form H I Xg 0 e ( g P G). n. . if we can choose a basis B of an FGmodule V such that [ g]B has the form H I à 0 d e à à for all g P G (see (5. k . where X g is k 3 k with 0 . Yg.4)). then we get H I j 0 0 [ g]B 9 ˆ d 0 j j e. the matrix [ g]B has the form H I j 0 0 [ g]B ˆ d j j j eX j j j The zeros re¯ect the fact that U is an FG-submodule of V (see (5.

8. U ˆ U1 È X X X È Ur . Then by (2. where each Ui and W j is an irreducible FG-module. so suppose that V is reducible. then every non-zero FG-module is completely reducible. The result is true if dim V ˆ 1. we simply mean an FG-submodule which is an irreducible FG-module. W ˆ W 1 È X X X È W s . since V is irreducible in this case. Proof Let V be a non-zero FG-module. dim V and dim W . . where each Ui is an irreducible FG-submodule of V. dim V. If V is irreducible then the result holds. j Another useful consequence of Maschke's Theorem is the next proposition. by induction.) 8. The proof goes by induction on dim V.10).7 Theorem If G is a ®nite group and F ˆ R or C. By Maschke's Theorem. we have. Consequences of Maschke's Theorem We now use Maschke's Theorem to show that every non-zero FGmodule is a direct sum of irreducible FG-submodules.74 Representations and characters of groups Maschke's Theorem asserts further that r is equivalent to a representation of the form H I Ag 0 e. a direct sum of irreducible FG-modules. g 3d 0 Bg where A g is also a k 3 k matrix. Then V has an FG-submodule U not equal to {0} or V. V ˆ U1 È X X X È Ur È W 1 È X X X È W s . there is an FG-submodule W such that V ˆ U È W. Since dim U .6 De®nition An FG-module V is said to be completely reducible if V ˆ U1 È F F F ÈU r . (By an irreducible FG-submodule.

where F ˆ R or C and G is a ®nite group. Summary of Chapter 8 Assume that G is a ®nite group and F ˆ R or C. Maschke's Theorem says that for every FG-submodule U of an FGmodule V. in order to understand FG-modules. Thus. there is an FG-submodule W of V such that V ˆ U È W. there is an FG-submodule W with V ˆ U È WX 2. we may concentrate upon the irreducible FG-modules. Proof By Maschke's Theorem. v2 . . Theorem 8. We begin our study of these in the next chapter.11. where v1 x ˆ v2 . v2 x ˆ Àv1 À v2 X (This is a CG-module. Let G ˆ kx: x 3 ˆ 1l  C3 .8 Proposition Let V be an FG-module. by Exercise 3. Then the function ð: V 3 U which is de®ned by ð: u ‡ w 3 u (u P U . If G ˆ C2 3 C2 . 1. express the group algebra RG as a direct sum of 1-dimensional RG-submodules.2. Then there exists a surjective FG-homomorphism from V onto U.7 tells us that every non-zero FG-module is a direct sum of irreducible FG-modules. by Proposition 7.) Express V as a direct sum of irreducible CG-submodules.Maschke's Theorem 75 8. w P W ) j is an FG-homomorphism onto U. Every non-zero FG-module V is a direct sum of irreducible FGmodules: V ˆ U1 È X X X È Ur X Exercises for Chapter 8 1. Suppose that U is an FG-submodule of V. 2. and let V be the 2-dimensional CGmodule with basis v1 .

Suppose that there are elements g. g P G. ] on V by ˆ [u. vx) (u. and de®ne U c ˆ fv P V : [u.6 in the light of this result. h in G such that the matrices gr and hr do not commute.2(2). Suppose that G is the in®nite group &  ' 1 0 : nPZ n 1 and let V be the CG-module C2 . Find a group G. Let V be a CG-module with basis v1 .5(2) and Exercises 5. ì jv j ˆ ëi ìiX iˆ1 jˆ1 iˆ1 De®ne another complex inner product [ . 5.76 Representations and characters of groups 3. v P V )X xPG (1) Verify that [ .) 6. . v P V and g P GX (2) Suppose that U is a CG-submodule of V. . (You may care to revisit Example 5. with the natural multiplication by elements of G (so that for v P V. the vector vg is just the product of the row vector v with the matrix g). v n and suppose that U is a CG-submodule of V. vg] ˆ [u. . v] for all u. 4. 2 n 3 n n ˆ ˆ ˆ ëi vi . Show that V is not completely reducible. . De®ne a complex inner product ( . v] ˆ (ux. ) on V as follows (see (14. v] ˆ 0 for all u P U gX Show that U c is a CG-submodule of V. Prove that r is irreducible. a CG-module V and a CG-homomorphism W: V 3 V such that V Tˆ Ker W È Im W.1.2) for the de®nition of a complex inner product): for ë i . Let G be a ®nite group and let r: G 3 GL (2. ì j P C.3. (This shows that Maschke's Theorem fails for in®nite groups ± compare Example 8. ] is a complex inner product. C) be a representation of G. . 6. 5. which satis®es [ug.4. An alternative proof of Maschke's Theorem for CG-modules.) 5.

. there exists a faithful irreducible CG-module. (Hint: it is a standard property of complex inner products that V ˆ U È U c for all subspaces U of V.Maschke's Theorem 77 (3) Deduce Maschke's Theorem.) 7. Prove that for every ®nite simple group G.

Then Im W Tˆ {0}. Thus W is invertible.26). Schur's Lemma is fundamental to representation theory. Also by Proposition 7. Schur's Lemma 9. Though simple in both statement and proof. (1) If W: V 3 W is a CG-homomorphism. Ker W ˆ {0}. Ker W is a CG-submodule of V. G denotes a ®nite group. As Im W is a CG-submodule of W by Proposition 7. or vW ˆ 0 for all v P V. and since much of the ensuing theory depends on it. as Ker W Tˆ V and V is irreducible.2. Schur's Lemma concerns CG-modules rather than RG-modules.2.1 Schur's Lemma Let V and W be irreducible CG-modules. Thus Ker (W À ë1 V ) is a non-zero CG-submodule 78 . the endomorphism W has an eigenvalue ë P C. Proof (1) Suppose that vW Tˆ 0 for some v P V.9 Schur's Lemma Schur's Lemma is a basic result concerning irreducible modules. we shall deal with CG-modules for the remainder of the book (except in Chapter 23). we have Im W ˆ W. and hence is a CG-isomorphism. and W is irreducible. Throughout. (2) If W: V 3 V is a CG-isomorphism. and we give an immediate application by determining all the irreducible representations of ®nite abelian groups. then W is a scalar multiple of the identity endomorphism 1 V . (2) By (2. and so Ker (W À ë1 V ) Tˆ {0}. then either W is a CGisomorphism.

Schur's Lemma of V. Then r is irreducible if and only if every n 3 n matrix A which satis®es ( gr)A ˆ A( gr) has the form A ˆ ëI n with ë P C. Part (2) of Schur's Lemma has the following converse. Proof Suppose that V is reducible. Let A be an n 3 n matrix over C. Proof As in Theorem 4. g P G. C) be a representation of G. as required. there is a CGsubmodule W of V such that V ˆ U È WX Then the projection ð: V 3 V de®ned by (u ‡ w)ð ˆ u for all u P U. Therefore v(W À ë1 V ) ˆ 0 That is. Then V is irreducible.3 Corollary Let r: G 3 GL (n. which is a contradiction. j We next interpret Schur's Lemma and its converse in terms of representations. for all g P G . The endomorphism v 3 vA of C n is a CG-homomorphism if and only if (v g)A ˆ (vA) g for all v P C n . and is not a scalar multiple of 1 V . and suppose that every CG-homomorphism from V to V is a scalar multiple of 1 V . for all v P V X 79 j 9. w P W is a CG-homomorphism (see Proposition 7. Ker (W À ë1 V ) ˆ V. By Maschke's Theorem. 9.2 Proposition Let V be a non-zero CG-module. so that V has a CG-submodule U not equal to {0} or V. Since V is irreducible. g P G. Hence V is irreducible. W ˆ ë1 V .11). regard C n as a CG-module by de®ning v g ˆ v( gr) for all v P C n .4(1).

4 Examples (1) Let G ˆ C3 ˆ ka: a3 ˆ 1l. Since the matrix  0 À1 1 À1  commutes with all gr ( g P G). Check that there is a representation r: G 3 GL (2.1 and Proposition 9. and then (br)A ˆ A(br) gives á ˆ ä. C) for which  ar ˆ Assume that the matrix  Aˆ á ã â ä  ù 0  0 . ùÀ1  br ˆ 0 1  1 X 0 commutes with both ar and br. j 9. by Corollary 9.2. C) be the representation for which  ar ˆ 0 À1 1 À1  (see Exercise 3. and let r: G 3 GL (2. b: a5 ˆ b2 ˆ 1. The fact that (ar)A ˆ A(ar) forces ⠈ 㠈 0. Corollary 9. . bÀ1 ab ˆ aÀ1 l. (2) Let G ˆ D10 ˆ ka. if and only if ( gr)A ˆ A( gr) for all g P GX The result now follows from Schur's Lemma 9.3.3 implies that r is reducible.80 Representations and characters of groups that is. Hence  Aˆ á 0 0 á  ˆ áIX Consequently r is irreducible. and let ù ˆ e2ðia5 .2).

9. and hence the endomorphism v 3 vx of V is a CG-homomorphism. C) be an irreducible representation of G (ci in ith position)X . g r i. . X X X . We shall not prove it here.1(2). By Theorem 9. and for 1 < i < r. . with g in i ˆ 1 and g i g j ˆ g j g i for all i. we deduce that dim V ˆ 1. and let V be an irreducible CGmodule. Write g i ˆ (1. Fraleigh listed in the Bibliography. say ë x 1 V . Thus vx ˆ ë x v for all v P V X This implies that every subspace of V is a CG-submodule. Let G ˆ C n1 3 .6. B. but refer you to Chapter 9 of the book of J. The next result is a major structure theorem for ®nite abelian groups. . n r are positive integers. this covers the irreducible representations of all ®nite abelian groups. 1) Then G ˆ h g1 . . Since G is abelian. We shall determine the irreducible representations of all direct products C n1 3 C n2 3 X X X 3 C n r where n1 . As V is irreducible. this endomorphism is a scalar multiple of the identity 1 V . X X X .6 Theorem Every ®nite abelian group is isomorphic to a direct product of cyclic groups.5 Proposition If G is a ®nite abelian group. By Schur's Lemma 9. Pick x P G. jX Now let r: G 3 GL (n. then every irreducible CG-module has dimension 1. X X X . . ci . . 3 C n r . v gx ˆ vxg for all g P G. let c i be a generator for C n i .Schur's Lemma Representation theory of ®nite abelian groups 81 Let G be a ®nite abelian group. Thus we have proved 9.

so for 1 < i < r. . . There are n1 n2 . V4 . 3. We have proved the following theorem. . . .7) for all i1 . and no two of them are equivalent. 9. that is. 2. we have ë in i ˆ 1. write r ˆ rë1 .XXX. . . v2 g 1 ˆ v2 . n r such representations. The representations rë1 .XXX. ë r i determine r. . v3 g 1 ˆ Àv3 . .9 Examples (1) Let G ˆ C n ˆ ka: a n ˆ 1l. .5. v3 g 2 ˆ v3 . .ë r X Conversely. . 9. where Vi is a 1-dimensional space with basis v i (i ˆ 1. ë i is an n i th root of unity. V2 . i r . . and put ù ˆ e2ðia n . . we have g ˆ g11 . . v1 g 2 ˆ v2 .8 Theorem Let G be the abelian group C n1 3 . and then (9X7) i i gr ˆ ( g 11 X X X g irr )r ˆ (ë11 X X Xë irr )X For a representation r of G satisfying (9. v4 g 2 ˆ Àv4 X . There are |G| of these representations. given any n i th roots of unity ë i (1 < i < r). Then n ˆ 1 by Proposition 9. where ak rù j ˆ (ù jk ) (0 < k < n À 1)X (2) The four irreducible CG-modules for G ˆ C2 3 C2 ˆ k g1 . g2 l are V1 . the function i i g 11 X X X g irr 3 (ë11 X X Xë irr ) is a representation of G. The n irreducible representations of G over C are rù j (0 < j < n À 1). As g i has order n i . the values ë1 . .82 Representations and characters of groups over C. . there exists ë i P C such that g i r ˆ (ë i ) (where of course (ë i ) is a 1 3 1 matrix). 4) and v1 g 1 ˆ v1 . g irr for some integers i1 . V3 . 3 C n r . Also. v2 g 2 ˆ Àv2 . v4 g 1 ˆ Àv4 . i r . and every irreducible representation of G over C is equivalent to precisely one of them. since for g P G.ë r of G constructed above are irreducible and have degree 1. .

9. a direct sum of irreducible C H-submodules Ui of V. . then there is a basis B of V such that the matrix [ g]B is diagonal.11 Proposition Let G be a ®nite group and V a CG-module. By Theorem 8. it is easy to check that Z(CG) is a subspace of CG. Put ù ˆ e2ðia n . 9. then the entries on the diagonal of [ g]B are nth roots of unity. V ˆ U1 È X X X È Ur . . and let V be a non-zero C H-module. by Proposition 9. Each Ui has dimension 1. let u i be a vector spanning Ui. For abelian groups G. . If g has order n.5.7.Schur's Lemma Diagonalization 83 Let H ˆ k gl be a cyclic group of order n. there exists an integer m i such that ui g ˆ ù m i u i X Thus if B is the basis u1 . . Proof Let H ˆ k gl. then H m1 I ù 0 f g FF (9X10) [ g]B ˆ d eX F 0 ù mr The following useful result is an immediate consequence of this.5). For arbitrary groups G. Then for each i.12 De®nition Let G be a ®nite group. written Z(CG). is de®ned by Z(CG) ˆ fz P CG: zr ˆ rz for all r P CGgX Using (2. we shall see that Z(CG) plays a crucial role in . The centre of the group algebra CG. u r of V. As V is also a C H-module.10). the result follows from (9. j Some further applications of Schur's Lemma Our next application concerns an important subspace of the group algebra CG. If g P G. the centre Z(CG) is the whole group algebra.

Consequently zr ˆ rz for all r P CG. its dimension is equal to the number of irreducible representations of G ± see Chapter 15). We shall see later that these elements in fact form a basis of Z(CG). this CG-homomorphism is equal to ë1 V for some ë P C. 9. which we now de®ne.14 Proposition Let V be an irreducible CG-module. and the result follows. and hence the function v 3 vz is a CG-homomorphism from V to V. . Then there exists ë P C such that vz ˆ ëv for all v P V X Proof For all r P CG and v P V. so the elements 1. We use Schur's Lemma to prove the following important property of the elements of Z(CG). and let z P Z(CG). we have vrz ˆ vzr. Then for all g P G.1(2). Indeed. For example. b: a3 ˆ b2 ˆ 1.13 Example € The elements 1 and gPG g lie in Z(CG). kal and G are normal subgroups of G. 9. hP H hP H hP H € hP H and so zg ˆ gz. if G ˆ D6 ˆ ka. j Some elements of the centre of CG are provided by the centre of G. By Schur's Lemma 9.84 Representations and characters of groups the study of representations of G (for example. bÀ1 ab ˆ aÀ1 l. then ˆ h P Z(CG)X To see this. if H is any normal subgroup of G. write z ˆ h. then {1}. 1 ‡ a ‡ a2 and 1 ‡ a ‡ a2 ‡ b ‡ ab ‡ a2 b lie in Z(CG). ˆ ˆ g À1 zg ˆ g À1 hg ˆ h ˆ z.

14. and is a subset of Z(CG).15 De®nition The centre of G. we give an example of a group G such that Z(G) is cyclic but there exists no faithful irreducible CG-module. Therefore Z(G)  {ë z : z P Z(G)}. then Z(G) is cyclic. the following result shows that the existence of a faithful irreducible CG-module imposes a strong restriction on the structure of G. they . there is no faithful irreducible CG-module unless G is cyclic.16 is false. The irreducible representations of non-abelian groups are more dif®cult to construct than those of abelian groups. which. C2 3 C2 has no faithful irreducible representation (compare Example 9.6. is cyclic (see Exercise 1. j We remark that the converse of Proposition 9. and so by Proposition 9. If z P Z(G) then z lies in Z(CG). it is not necessarily the case that there is a faithful irreducible CG-module. written Z(G). and hence by Proposition 9. For example. there exists ë z P C such that vz ˆ ë z v for all v P V X Since V is faithful. Proof Let V be a faithful irreducible CG-module. 9. being a ®nite subgroup of Cà . since in Exercise 25. Although we have seen in Proposition 6.17 Example If G is an abelian group.Schur's Lemma 9. is de®ned by Z(G) ˆ fz P G: zg ˆ gz for all g P GgX 85 Clearly Z(G) is a normal subgroup of G.16.9(2)). then G ˆ Z(G).16 Proposition If there exists a faithful irreducible CG-module. In particular. Indeed. the function z 3 ëz (z P Z(G)) is an injective homomorphism from Z(G) into the multiplicative group Cà of non-zero complex numbers. 9.7).6 that for every ®nite group G there is a faithful CG-module.

Also. Schur's Lemma states that every CG-homomorphism between irreducible CG-modules is either zero or a CG-isomorphism. Then dim Vi ˆ 1 for all i.5. y P G. Hence G is abelian. All irreducible CG-modules for a ®nite abelian group G have dimension 1. j Summary of Chapter 9 1. let v i be a vector spanning V i . . as is shown by the following converse to Proposition 9. . we can write CG ˆ V1 È X X X È Vn . . Write down the irreducible representations over C of the groups C2 .7. the matrices [x]B and [ y]B are diagonal. C3 and C2 3 C2 . For all x. Then G is abelian. . and hence they commute. we deduce that x and y commute. Exercises for Chapter 9 1. Since the representation g 3 [ g]B ( g P G) of G is faithful (see Proposition 6. 2. where each Vi is an irreducible CG-submodule of the regular CGmodule CG.18 Proposition Suppose that G is a ®nite group such that every irreducible CG-module has dimension 1. as required. 9. v n is a basis of CG. the only CG-homomorphisms from an irreducible CG-module to itself are scalar multiples of the identity. The elements of Z(CG) act as scalar multiples of the identity on all irreducible CGmodules. Proof By Theorem 8. Then v1 . The centre Z(CG) of the group algebra CG consists of those elements which commute with all elements of CG. . and there are precisely |G| of them. For 1 < i < n.6).86 Representations and characters of groups do not all have degree 1. 3. call it B . since we are assuming that all irreducible CG-modules have dimension 1.

b: a3 ˆ b2 ˆ 1. ó of G. bÀ1 ab ˆ aÀ1 l. (b) Find ë P C such that w(a ‡ aÀ1 ) ˆ ëw for all w P W. irreducible. (Compare Proposition 9. (a) Find a non-trivial irreducible representation r of G such that g2 r ˆ (1) for all g P G. Let G ˆ D6 ˆ ka. and let W be the irreducible CG-submodule of the regular CGmodule de®ned by W ˆ sp (1 ‡ ù2 a ‡ ùa2 . then there exists ë P C such that 2ˆ 3 v g ˆ ëv for all v P V X gPG 6.6). .Schur's Lemma 87 2. Let G be the ®nite abelian group C n1 3 .14.) . Let G ˆ C4 3 C4 . bÀ1 ab ˆ aÀ1 l. Write ù ˆ e2ðia3 . Check that there is a representation r of G over C such that     À7 10 À5 6 ar ˆ . 3. (a) Show that a ‡ aÀ1 P Z(CG). Show that if V is an irreducible CG-module. b ‡ ù2 ab ‡ ùa2 b) (see Exercise 6. br ˆ X À5 7 À4 5 Find all 2 3 2 matrices M such that Hence determine whether or not r is Do the same for the representation   5 À6 aó ˆ . 3 C n r . bó 4 À5 M( gr) ˆ ( gr)M for all g P G. where   À5 6 ˆ X À4 5 5. Suppose that G ˆ D8 ˆ ka. Can G have a faithful representation of degree less than r? 4. Prove that G has a faithful representation of degree r. (b) Prove that there is no irreducible representation ó of G such that gó ˆ (À1) for all elements g of order 2 in G. b: a4 ˆ b2 ˆ 1. .

(d) C3 3 D8.88 Representations and characters of groups 7. (b) D8 . . Which of the following groups have a faithful irreducible representation? (a) C n (n a positive integer). (c) C2 3 D8 .

in theory.7.1 Proposition Let V and W be CG-modules and let W: V 3 W be a CG-homomorphism. unless G is a small group. Then there is a CG-submodule U of V such that V ˆ Ker W È U and U  Im W. . we can write CG ˆ U1 È X X X È Ur where each Ui is an irreducible CG-module. Ur. Consider CG as the regular CG-module. 10. De®ne a function W: U 3 Im W by uW ˆ uW (u P U )X 89 .10 Irreducible modules and the group algebra Let G be a ®nite group and CG be the group algebra of G over C. By Theorem 8. it is suf®cient to decompose CG as a direct sum of irreducible CG-submodules. this is not really a practical way of ®nding the irreducible CG-modules.8). . . Irreducible submodules of CG We begin with another consequence of Maschke's Theorem. We shall show in this chapter that every irreducible CG-module is isomorphic to one of the CG-modules U1. Also.2. there are only ®nitely many non-isomorphic irreducible CG-modules (a result which has already been established for abelian groups in Theorem 9. Proof Since Ker W is a CG-submodule of V by Proposition 7. However. there is by Maschke's Theorem a CG-submodule U of V such that V ˆ Ker W È U . . As a consequence. to ®nd all irreducible CG-modules.

since W is a CG-homomorphism.4 De®nitions (1) If V is a CG-module and U is an irreducible CG-module. Now let w P Im W. as the following example shows.2 Proposition Let V be a CG-module. so w ˆ vW for some v P V. As U and Ui are irreducible.11). we have ð i Tˆ 0. . Thus U  Im W. and ð i Tˆ 0.3 Example Let G be any group and let V be a 2-dimensional CG-module. with basis v1 . . . If u P Ker W then u P Ker W ’ U ˆ {0}. then U  Ui for some i. Clearly W is a CG-homomorphism. Schur's Lemma 9. . 10. and write V ˆ U1 È X X X È Us . such that v g ˆ v for all v P V and g P G. a direct sum of irreducible CG-submodules Ui. Now ð i is a CG-homomorphism (see Proposition 7. where U1 ˆ sp (v1 ) and U2 ˆ sp (v2 ) are irreducible CG-submodules. If U is any irreducible CG-submodule of V. È Us (each Ui irreducible) without U being equal to any Ui. j Of course it can happen that U is an irreducible CG-submodule of U1 È . De®ne ð i : U 3 Ui by setting uð i ˆ ui . However. Proof For u P U. 10.90 Representations and characters of groups We show that W is a CG-isomorphism from U to Im W. We have now established that W: U 3 Im W is an invertible CG-homomorphism. as required. Then V ˆ U1 È U2 . Write v ˆ k ‡ u with k P Ker W. Therefore U  Ui. then we say that U is a composition factor of V if V has a CG-submodule which is isomorphic to U. u P U. 10. ‡ us for unique vectors ui P Ui (1 < i < s). v2 . . U ˆ sp (v1 ‡ v2 ) is an irreducible CG-submodule which is not equal to U1 or U2.1(1) implies that ð i is a CG-isomorphism. we have u ˆ u1 ‡ . Choosing i such that ui Tˆ 0 for some u P U. hence Ker W ˆ {0}. Then w ˆ vW ˆ kW ‡ uW ˆ uW ˆ uWX Therefore Im W ˆ Im W.

a direct sum of irreducible CG-submodules. Observe that {wr: r P CG} is a CG-submodule of W. then there are only ®nitely many non-isomorphic irreducible CG-modules. By Proposition 10. since for r. since W is irreducible. and Im W ˆ W by (10. W is a CG-homomorphism. and write CG ˆ U 1 È X X X È U r .5 shows that there is a ®nite set of irreducible CGmodules such that every irreducible CG-module is isomorphic to one of them. Proof Let W be an irreducible CG-module. and choose a non-zero vector w P W. Moreover. Then every irreducible CG-module is isomorphic to one of the CG-modules Ui. then W  Ui. and the result is proved. j Theorem 10. s P CG. which shows that every irreducible CG-module is a composition factor of the regular CG-module.5 Theorem Regard CG as the regular CG-module. there is a CG-submodule U of CG such that CG ˆ U È Ker W and U  Im W ˆ W X As W is irreducible.2 we have U  Ui for some i. 10.1. We record this fact in the following corollary. so is U. 10.Irreducible modules and the group algebra 91 (2) Two CG-modules V and W are said to have a common composition factor if there is an irreducible CG-module which is a composition factor of both V and W.6). it follows that (10X6) W ˆ fwr: r P CGgX Now de®ne W: CG 3 W by rW ˆ wr (r P CG)X Clearly W is a linear transformation. We now come to the main result of the chapter. . (rs)W ˆ w(rs) ˆ (wr)s ˆ (rW)sX By Proposition 10.7 Corollary If G is a ®nite group.

U1 or U2. As in (1) above. v1 . v1 ˆ 1 ‡ ù2 a ‡ ùa2 . v1 . The irreducible representation of G corresponding to Ui is the representation rù i of Example 9. 2X Hence Ui is a CG-submodule of CG for i ˆ 0.8 Examples (1) Let G ˆ C3 ˆ ka: a3 ˆ 1l. 1. and so sp (v i ) and sp (wi ) are Ckal-modules. v2 b ˆ w1 . and write ù ˆ e2ðia3 . v2 is a basis of CG. to ®nd all the irreducible CG-modules we need only decompose the regular CG-module as a direct sum of irreducible CG-submodules. By Theorem 10. w0 b ˆ v0 . a direct sum of irreducible CG-submodules Ui. w1 b ˆ v2 . 2. We decompose CG as a direct sum of irreducible CG-submodules.5. Next. 1. w2 b ˆ v1 X .5. bÀ1 ab ˆ aÀ1 l. and similarly vi a ˆ ùi vi for i ˆ 0. We now do this for a couple of examples. note that v0 b ˆ w 0 . v2 ˆ 1 ‡ ùa ‡ ù2 a2 . v1 b ˆ w2 . Let ù ˆ e2ðia3 and de®ne v0 ˆ 1 ‡ a ‡ a2 . Then v1 a ˆ a ‡ ù2 a2 ‡ ù1 ˆ ùv1 .92 Representations and characters of groups According to Theorem 10. 2. 1. however. (2) Let G ˆ D6 ˆ ka. v1 ˆ 1 ‡ ù2 a ‡ ùa2 . and hence CG ˆ U0 È U 1 È U2 . 2. w0 ˆ bv0 w1 ˆ bv1 . w2 ˆ bv2 X (ˆ b ‡ ba ‡ ba2 ).9(1). v2 ˆ 1 ‡ ùa ‡ ù2 a2 . It is easy to check that v0 . v i a ˆ ù i v i for i ˆ 0. b: a3 ˆ b2 ˆ 1. this is not a practical method for studying CG-modules in general. v2 P CG by v0 ˆ 1 ‡ a ‡ a2 . 10. De®ne v0 . every irreducible CG-module is isomorphic to U0. and let Ui ˆ sp (v i ) for i ˆ 0. 1.

Find a CG-submodule of CG which is isomorphic to the trivial CG-module. 2.5 that there are exactly three nonisomorphic irreducible CG-modules.5(2). and hence CG ˆ U1 È U2 È U3 È U4 . By the argument in Example 5. v1 . Now v0. the CG-submodules U3 ˆ sp (v1 . v2 . the other 1-dimensional Ui. as U 1 ˆ sp(v0 ‡ w0 ) and U2 ˆ sp(v0 À w0 ) are CG-submodules. Is there only one such CGsubmodule? 2. We conclude from Theorem 10. and U1 is not isomorphic to U2.b3 X 1 0 0 ùÀ1 Summary of Chapter 10 1. Correspondingly. U2 and U3. sp(v0 . w0 ) is reducible. sp (v0 . w2 ) and sp (v2 . w0 ). But U3  U4 (there is a CG-isomorphism sending v1 3 w1 . w1. However.     0 1 ù 0 r3 : a 3 . Let G ˆ C4 . every irreducible representation of D6 over C is equivalent to precisely one of the following: r1 : a 3 (1).8(1). w2 is a basis of CG. b 3 (1). Note that U1 is the trivial CG-module. r2 : a 3 (1). namely U1.Irreducible modules and the group algebra 93 Therefore. (Hint: copy the method of Example 10. Express CG as a direct sum of irreducible CGsubmodules. w0. sp (v1 . Every irreducible CG-module occurs as a composition factor of the regular CG-module. b 3 (À1). Exercises for Chapter 10 1. w1 ) are Ckbl-modules.) . Let G be a ®nite group. w2 3 v2 ). There are only ®nitely many non-isomorphic irreducible CGmodules. a direct sum of irreducible CG-submodules. w2 ) and U4 ˆ sp (v2 . w1 ) are irreducible. and hence are CG-submodules of CG.

b: a4 ˆ b2 ˆ 1. v2 a ˆ Àiv2 . v2 and v1 a ˆ iv1 .94 Representations and characters of groups 3. Show that there is a V which is not equal to U1 or U2. b: a4 ˆ 1. and let V be the CG-module given in Example 4. Let G ˆ Q8 ˆ ka. and u3 b ˆ Àu3 X 4. Suppose that V is a where U1 and U2 are CG-submodule U of isomorphic to both of non-zero CG-module such that V ˆ U1 È U2. v2 b ˆ Àv1 X Show that V is irreducible. sp (u2 ) and sp (u3 ). Thus V has basis v1 . isomorphic CG-modules. bÀ1 ab ˆ aÀ1 l. v1 b ˆ v2 . but is them. Let G ˆ D8 ˆ ka. 6. sp (u1 ) say. u2 b ˆ u2 . Use the method of Example 10. b2 ˆ a2 .5(2). u3 a ˆ Àu3 . . bÀ1 ab ˆ aÀ1 l.8(2) to ®nd all the irreducible representations of D8 over C. of CG such that u1 a ˆ u1 . Find a 1-dimensional CG-submodule. and ®nd a CG-submodule of CG which is isomorphic to V. 5. such that u2 a ˆ Àu2 . u1 b ˆ Àu1 X Find also 1-dimensional CG-submodules.

9). ö P HomCG (V . W ). In Theorem 10. The question arises: how many of the Ui are isomorphic to U ? There is an elegant and signi®cant answer to this question: the number is precisely dim U (see Theorem 11. a direct sum of irreducible CG-modules Ui. we write CG ˆ U 1 È X X X È U r . W) with an easy consequence of Schur's Lemma.11 More on the group algebra We now go further into the structure of the group algebra CG of a ®nite group G.5 we proved that every irreducible CG-module U is isomorphic to one of the Ui. W ) for the set of all CG-homomorphisms from V to W. We write HomCG (V . W ) and ë P C.9 is based on a study of the vector space of CG-homomorphisms from one CG-module to another. The space of CG-homomorphisms 11. 95 . As in Chapter 10. De®ne addition and scalar multiplication on HomCG (V . v(ëW) ˆ ë(vW) for all v P V . ëW P HomCG (V . W ) is a vector space over C. W ) as follows: for W. it is easily checked that HomCG (V. Then W ‡ ö. de®ne W ‡ ö and ëW by v(W ‡ ö) ˆ vW ‡ vö. Our proof of Theorem 11. We begin our study of the vector space HomCG (V.1 De®nition Let V and W be CG-modules. With these de®nitions.

W ) ˆ fëW: ë P Cg. Then (1) dim (HomCG (V. . j For the next result.1(2). Let X be an irreducible CG-submodule of U. and let W: V 3 W be a CG-isomorphism.1(1) implies that XW  X. W )). W )) ˆ 0. Then V and W have a common composition factor. if V T W X Proof If V T W then this is immediate from Schur's Lemma 9. by Maschke's Theorem. so by Schur's Lemma 9.3 Proposition Let V and W be CG-modules. 11. then öWÀ1 is a CG-isomorphism from V to V. W ). Proof Let W be a non-zero element of HomCG (V . W 2 be CG-modules. Then V ˆ Ker W È U for some non-zero CG-module U. If ö P HomCG (V . 11. (2) dim (HomCG (V1 È V2 . Therefore X is a common composition factor of V and W. W)) ‡ dim (HomCG (V2 . V2 and W . W2 )). The key step is the following proposition. and suppose that HomCG (V . Since XW Tˆ {0}. a 1-dimensional space. W 1 . W ) Tˆ f0g. W ). W1 È W2 )) ˆ dim (HomCG (V. there exists ë P C such that öWÀ1 ˆ ë1 V X Then ö ˆ ëW. Then & 1. j The next few results show how to calculate the dimension of HomCG (V . if V  W .1(1).96 Representations and characters of groups 11. W1 )) ‡ dim (HomCG (V. and so HomCG (V . V1 . W ) in general. recall the de®nition of a composition factor of a CG-module from 10.4. dim (HomCG (V . Now suppose that V  W.2 Proposition Suppose that V and W are irreducible CG-modules. Schur's Lemma 9. W )) ˆ dim (HomCG (V1 .4 Proposition Let V .

We show that f is invertible. W Vi is the function v i W Vi ˆ v i W (v i P Vi )X to Then W Vi P HomCG (V i .1). W1 ) and HomCG (V. and the image of ö under f is (ö1 . If W P HomCG (V . Now let h be the function from HomCG (V1 È V2 . ö2 ) under h. 2. ð1 and ð2 are CG-homomorphisms. W 1 È W 2 ))X Clearly f is a linear transformation. (2) For W P HomCG (V1 È V2 .11. We have established that f is an invertible linear transformation from HomCG (V. W2 ). If W P Ker f. We now de®ne a function f from HomCG (V.More on the group algebra Proof (1) De®ne W 1 È W 2 3 W 2 by the functions ð1 : W1 È W2 3 W1 (w1 ‡ w2 )ð2 ˆ w2 and 97 ð2 : (w1 ‡ w2 )ð1 ˆ w1 . Hence h is surjective. Therefore W ˆ 0. W2 ) by f : W 3 (Wð1 . de®ne W Vi : Vi 3 W (i ˆ 1. Wi ) (i ˆ 1. W1 È W2 ) to HomCG (V. the function ö: v 3 vö1 ‡ vö2 (v P V ) lies in HomCG (V. W 1 ) and Wð2 P HomCG (V. W1 ) È HomCG (V. W ) for i ˆ 1. W ) which is given by h: W 3 (W V1 . Consequently these two vector spaces have equal dimensions. and (1) follows. j . We have shown that h is an invertible linear transformation. W ) È HomCG (V2 . then vWð1 ˆ 0 and vWð2 ˆ 0 for all v P V. 2) lies in HomCG (V1 È V2 . W2 ) (see Exercise 7. 2) to be the restriction of W to V i . then Wð1 P HomCG (V . 2). W1 È W2 ). so Ker f ˆ {0} and f is injective. W V2 ) (W P HomCG (V1 È V2 . Given ö i P HomCG (V. and (2) follows. Hence f is surjective. W ). w2 P W 2 . the function ö: v1 ‡ v2 3 v1 ö1 ‡ v2 ö2 öi P (v i P Vi for i ˆ 1. W ) and has image (ö1 . 2). By Proposition 7. Wð2 ) (W P HomCG (V . W ) HomCG (V1 . W ))X transformation. W 1 È W 2 ). Given Clearly h is an injective linear HomCG (Vi . that is. W1 È W2 ) to the (external) direct sum of HomCG (V. ö2 ). W ) (i ˆ 1. for all w1 P W 1 . so vW ˆ vW(ð1 ‡ ð2 ) ˆ 0.

4. . È Ws )) ˆ s ˆ jˆ1 dim (HomCG (V. In the following corollary we single out the case where one of the CGmodules is irreducible. By an obvious induction using Proposition 11. U i ))X .98 Representations and characters of groups Now suppose that we have CG-modules V. Wj (1 < i < r. 1 < j < s). È Ws )) ˆ r s ˆˆ iˆ1 jˆ1 dim (HomCG (Vi. Let W be any irreducible CG-module. W )) ˆ dim (HomCG (W . Then the dimensions of HomCG (V . where each Ui is an irreducible CG-module.6 Corollary Let V be a CG-module with V ˆ U1 È X X X È Us . and dim (HomCG (W . . V ) are both equal to the number of CG-modules Ui such that Ui  W. . Wj )).5). W1 È . W )) in general. 11. W )) ˆ r ˆ iˆ1 dim (HomCG (Vi.2. . These in turn imply (3) dim (HomCG (V1 È . Wj )). dim (HomCG (V . By applying (3) when all Vi and Wj are irreducible. W1 È . V )) ˆ s ˆ iˆ1 s ˆ iˆ1 dim (HomCG (Ui . W ) and HomCG (W . W)). W. .5) (1) dim (HomCG (V. Proof By (11. we have (11. (2) dim(HomCG (V1 È F F F ÈV r . and using Proposition 11. we can ®nd dim (HomCG (V. . Vi. È Vr . W )).

11. we have dim (HomCG (CG. U ). then dim (HomCG (CG. with U3  U4 but U3 not isomorphic to U1 or U2. ö d is a basis of HomCG (CG. 0.8 Proposition If U is a CG-module. U3 )) ˆ dim (HomCG (U3 . . ud of U. Since ö is a CG-homomorphism. Choose a basis u1 . for all r P CG we have .6. When combined with Corollary 11. .7 Example For G ˆ D6. CG)) ˆ 2X You are asked in Exercise 11.2. The next proposition investigates the space of CG-homomorphisms from the regular CG-module to any other CG-module. W )) ˆ dim (HomCG (W . we saw in Example 10.5 to ®nd bases for these two vector spaces of CG-homomorphisms. . . s P CG. U ). if U i T W X j a direct sum of irreducible CG-submodules. 11. Then 1ö ˆ ë1 u1 ‡ X X X ‡ ë d u d for some ë i P C.8(2) that CG ˆ U1 È U2 È U3 È U4 . de®ne ö i : CG 3 U by rö i ˆ ui r (r P CG)X Then ö i P HomCG (CG.More on the group algebra And by Proposition 11. Thus by Corollary 11. it will give the main result of this chapter. U )) ˆ dim U X Proof Let d ˆ dim U. . . . (rs)ö i ˆ ui (rs) ˆ (ui r)s ˆ (rö i )sX We shall prove that ö1 . & 99 1. Ui )) ˆ The result follows. U) since for all r. if U i  W . Suppose that ö P HomCG (CG. For 1 < i < d. . dim (HomCG (Ui .6.

and . ö d is a basis of HomCG (CG. Hence ö1 . Proof By Proposition 11. . Now assume that Therefore ö1 . If U is any irreducible CG-module. U2 are non-isomorphic 1-dimensional CG-modules. HomCG (CG. ‡ ë d ö d . we have 0 ˆ 1(ë1 ö1 ‡ X X X ‡ ë d ö d ) ˆ ë1 u1 ‡ X X X ‡ ë d u d . . a direct sum of irreducible CG-submodules.8. .100 Representations and characters of groups rö ˆ (1r)ö ˆ (1ö)r ˆ ë1 u1 r ‡ X X X ‡ ë d u d r ˆ r(ë1 ö1 ‡ X X X ‡ ë d ö d )X Hence ö ˆ ë1 ö1 ‡ . .9 Theorem Suppose that CG ˆ U1 È X X X È Ur . which therefore has dimension d. . . then the number of CG-modules Ui with Ui  U is equal to dim U. . ö d span ë1 ö1 ‡ X X X ‡ ë d ö d ˆ 0 (ë i P C)X Evaluating both sides at the identity 1. . U )). . j We now come to the main theorem of the chapter. dim U ˆ dim (HomCG (CG.10 Example Recall again from Example 10.6. j 11. and by Corollary 11. which forces ë i ˆ 0 for all i. .8(2) that if G ˆ D6 then CG ˆ U1 È U 2 È U3 È U 4 . U ). this is equal to the number of Ui with Ui  U. which tells us how often each irreducible CG-module occurs in the regular CGmodule. U ). 11. where U1.

.12. the result follows. dim U2 ˆ 1.11 De®nition We say that the irreducible CG-modules V1 . Then k ˆ (dim Vi )2 ˆ jGjX iˆ1 Proof Let CG ˆ U1 È . U2 occurs once. (By Corollary 10. for each i. the number of CG-modules Uj with Uj  Vi is equal to di . U3 occurs twice. write d i ˆ dim V i . U4 are isomorphic irreducible 2-dimensional CG-modules. By Theorem 11. and let d1. Therefore dim CG ˆ dim U 1 ‡ X X X ‡ dim Ur ˆ k ˆ iˆ1 k ˆ iˆ1 d i (dim Vi ) ˆ d2X i j As dim CG ˆ |G|. V k form a complete set of non-isomorphic irreducible CG-modules. X X X . dk be the dimensions of all the irreducible CG-modules. . V k are isomorphic. V k form a complete set of non-isomorphic irreducible CG-modules if every irreducible CG-module is isomorphic to some V i .) 11.12 Theorem Let V1 . a direct sum of irreducible CG-submodules. This illustrates Theorem 11. . . k ˆ iˆ1 d 2 ˆ 8X i . X X X . dim U1 ˆ 1. X X X .7. dim U3 ˆ 2X We conclude the chapter with a signi®cant consequence of Theorem 11.9. By Theorem 11. 11. . For 1 < i < k. È Ur.More on the group algebra 101 U3. for any ®nite group G there exists a complete set of non-isomorphic irreducible CG-modules.9: U1 occurs once. and no two of V1 . 11. .13 Example Let G be a group of order 8.9 concerning the dimensions of all irreducible CG-modules.

) 3.102 Representations and characters of groups Observe that the trivial CG-module is irreducible of dimension 1. . Let CG ˆ U1 È . and this fact. 3. If V1 . combined with Theorem 11. 1. W j ))X 2. dk are 1.3. 1. Then the number of Ui with Ui  U is equal to dim U. . We shall see later that dim Vi divides |G| for all i. . 1 1. 1. 1. Summary of Chapter 11 1. Let G be a ®nite group. . U )) ˆ dim U . dim (HomCG (CG. ®nd the dimensions of all the irreducible CG-modules. . . If G is a non-abelian group of order 6. 2. . 1. 1. X X X . 2X Both these possibilities do occur: the ®rst holds when G is an abelian group (see Proposition 9. is quite a powerful tool in ®nding the dimensions of irreducible CG-modules. Find a basis for HomCG (CG. and the second when G ˆ D8 (see Exercise 10. a direct sum of irreducible CG-modules. what are the possible degrees of all the irreducible representations of G? Find the degrees of the irreducible representations of D12. W 1 È X X X È W s )) ˆ r s ˆˆ iˆ1 jˆ1 and dim (HomCG (Vi . È Ur. CG). 1. V k is a complete set of non-isomorphic irreducible CGmodules. and so di ˆ 1 for some i.5). 1. 4. and let U be any irreducible CG-module. (Hint: use Exercise 5. dim (HomCG (V1 È X X X È V r .12. then k ˆ (dim Vi )2 ˆ jGjX iˆ1 Exercises for Chapter 11 1. If G is a group of order 12. Hence the possibilities for d1.4). 1.

Suppose that G ˆ Sn and V is the n-dimensional permutation module for G over C. 5. U3 ) and a basis for HomCG (U3. W be arbitrary CG-modules. W )) ˆ iˆ1 d i ei .More on the group algebra 103 4. . Vi ))X €k Show that dim (HomCG (V . Find a basis for HomCG (CG. show that HomCG (V. and let V.10. 6.8(2). Vi )) and ei ˆ dim (HomCG (W . If U is the trivial CGmodule. Assume that for 1 < i < k. as in Example 10. a direct sum of irreducible CG-modules. V k be a complete set of non-isomorphic irreducible CG-modules. Let G ˆ D6 and let CG ˆ U1 È U2 È U3 È U4. Let V1 . X X X . as de®ned in 4. CG). U) has dimension 1. d i ˆ dim (HomCG (V .

We say that x is conjugate to y in G if y ˆ g À1 xg for some g P GX The set of all elements conjugate to x in G is x G ˆ f g À1 xg: g P Gg. G is a ®nite group. After de®ning conjugacy classes. Our ®rst result shows that two distinct conjugacy classes have no elements in common. Conjugacy classes 12. Proof Suppose that x G ’ y G is not empty. symmetric and alternating groups. Then there exist g. y P G. we develop enough theory to determine the conjugacy classes of dihedral. h P G such that z ˆ gÀ1 xg ˆ hÀ1 yhX 104 .12 Conjugacy classes We take a break from representation theory to discuss some topics in group theory which will be relevant in our further study of representations.1 De®nition Let x. y P G. then either x G ˆ y G or x G ’ y G is empty. and is called the conjugacy class of x in G. Throughout the chapter. 12. At the end of the chapter we prove a result linking the conjugacy classes of a group to the structure of its group algebra. and pick z P x G ’ y G.2 Proposition If x.

. Another way of seeing this Corollary 12.3 is to observe that conjugacy is an equivalence relation. we have aG ˆ fa. a2 . .Conjugacy classes Hence x ˆ ghÀ1 yhgÀ1 ˆ kÀ1 yk. fa. 1 G ˆ {1} is a conjugacy class of G. Since gÀ1 ag is a or a2 for every g P G. a2 gX Also. and distinct conjugacy classes are disjoint. a2 b. 12. .5 Examples (1) For every group G. ab. G is the union of its conjugacy classes and so we deduce immediately 12. x G are l l distinct. . Similarly y G # x G (using y ˆ kxkÀ1 ). a2 g. xl representatives of the conjugacy classes of G. The elements of G are 1. j Since every element x of G lies in the conjugacy class x G (as x ˆ 1À1 x1 with 1 P G). b: a3 ˆ b2 ˆ 1. aÀi ba i ˆ aÀ2i b for all integers i. (2) Let G ˆ D6 ˆ ka. and that the conjugacy classes are the equivalence classes. and bÀ1 ab ˆ a2 . a2 bgX Thus the conjugacy classes of G are f1g. a. fb. ‘ x G . ab. . 12. where the conjugacy classes x1 . . then we call x1 . So a P x G A a ˆ bÀ1 xb A a ˆ cÀ1 yc AaP y X G 105 for some b P G where c ˆ kb A a ˆ bÀ1 k À1 ykb Therefore x G # y G. bÀ1 ab ˆ aÀ1 l. .3 Corollary Every group is a union of conjugacy classes. b. . a2 bgX . . where k ˆ hgÀ1 . and so xG ˆ yG .4 De®nition G G If G ˆ x1 ‘ . so bG ˆ fb. . ab.

and x and y have the same order. Let x have order m. 12. y P G. CG (x) ˆ f g P G: xg ˆ gxgX (So also CG (x) ˆ { g P G: gÀ1 xg ˆ x}. The centralizer of x in G. Observe that x P CG (x) and indeed. is the set of elements of G which commute with x.106 Representations and characters of groups (3) If G is abelian then gÀ1 xg ˆ x for all x. written CG (x). we have gÀ1 abg ˆ ( g À1 ag)( g À1 bg)X Hence gÀ1 x n g ˆ ( gÀ1 xg) n . Then y m ˆ gÀ1 x m g ˆ 1.) It is easy to check that CG (x) is a subgroup of G (Exercise 12. m. j Conjugacy class sizes The next theorem determines the sizes of the conjugacy classes in G in terms of certain subgroups which we now de®ne. |x G | divides |G|. and for 0 . . that is. 12. 12. r . so that y ˆ gÀ1 xg for some g P G. Hence every conjugacy class of G consists of just one element. Then y n ˆ gÀ1 x n g and therefore x n is conjugate to y n in G. b P G.6 Proposition Let x. Proof Observe that for a. so y also has order m. Then the size of the conjugacy class x G is given by jx G j ˆ jG: CG (x)j ˆ jGjajCG (x)jX In particular. y r ˆ gÀ1 x r g Tˆ 1. g P G. kxl # CG (x) for all x P G. If x is conjugate to y in G.7 De®nition Let x P G. and so x G ˆ {x}.1). then x n is conjugate to y n in G for every integer n. The next proposition is often useful when calculating conjugacy classes. Suppose that x is conjugate to y in G.8 Theorem Let x P G.

Since CG (ai ) contains kal.15.8 by ®nding the conjugacy classes of all dihedral groups. (1) n odd First consider ai (1 < i < n À 1). we make the observation that (12X9) jx G j ˆ 1 D g À1 xg ˆ x D x P Z(G). and both | Z(G)| and |x G | divide |G|. we may de®ne an injective function f from x G to the set of right cosets of CG (x) in G by f : g À1 xg 3 CG (x) g ( g P G)X Clearly f is surjective. jG: CG (ai )j < jG: haij ˆ 2X . h P G.Conjugacy classes Proof Observe ®rst that for g. . proving that |x G | ˆ |G:CG (x)|. xl be representatives of the conjugacy classes of G. Then ˆ jx G j. as de®ned in 9. j Before summarizing our results on conjugacy classes. Thus G ˆ ha. where Z(G) is the centre of G. b: an ˆ b2 ˆ 1. jGj ˆ j Z(G)j ‡ i xiP Z(G) a for all g P G where |x G | i ˆ |G:CG (xi )|. we have g À1 xg ˆ hÀ1 xh D hg À1 x ˆ xhg À1 D hg À1 P CG (x) D CG (x) g ˆ CG (x)hX 107 By dint of this. . Let G ˆ D2 n. it is convenient to consider separately the cases where n is odd and where n is even.10 The Class Equation Let x1 . . the dihedral group of order 2n. bÀ1 ab ˆ aÀ1 iX In ®nding the conjugacy classes of G. 12. Hence f is a bijection. . We have now proved all parts of the following result. i Conjugacy classes of dihedral groups We illustrate the use of Theorem 12.

As n is odd. and so |(ai ) G | > 2.108 Representations and characters of groups Also bÀ1 ai b ˆ aÀi . (2) n even Write n ˆ 2m. aÀ1 }. n ˆ 2m) has precisely m ‡ 3 conjugacy classes: {1}. ab. {b. {a mÀ1 . aÀi gX Next. a nÀ1 b}. . Thus CG (b) ˆ f1. As in case (1). That is. . As bÀ1 am b ˆ aÀ m ˆ am . For every integer j. {a2 j‡1 b: 0 < j < m À 1}. {a m }. and as bÀ1 ai b ˆ aÀi . aÀ1 }.11) The dihedral group D2 n (n odd) has precisely 1(n ‡ 3) con2 jugacy classes: {1}. . bG ˆ fb. bG must consist of the remaining n elements of G. the centralizer of am in G contains both a and b. aÀi } for 1 < i < m À 1. . .8. . aÀi } # (ai ) G . {a( nÀ1)a2 . |bG | ˆ n. and hence CG (am ) ˆ G. no element ai or ai b (with 1 < i < n À 1) commutes with b. Therefore the conjugacy class of am in G is just fam g. (ai ) G ˆ {ai . bgX Therefore by Theorem 12. so {ai . . a j baÀ j ˆ a2 j b. . Since all the elements ai have been accounted for. a j (ab)aÀ j ˆ a2 j‡1 bX It follows that bG ˆ fa2 j b: 0 < j < m À 1g. ab. {a. we have 2 > jG: CG (ai )j ˆ j(ai ) G j > 2X Hence equality holds here. Using Theorem 12. a nÀ1 bgX We have shown (12. and CG (ai ) ˆ hai.12) The dihedral group D2 n (n even. ai Tˆ aÀi . (ab) G ˆ fa2 j‡1 b: 0 < j < m À 1gX Hence (12. X X X . CG (b) contains {1.8. (ai ) G ˆ fai . {a. aÀ( nÀ1)a2 }. . . b}. . {a2 j b: 0 < j < m À 1}. aÀ m‡1 }. . .

ck s 3 c9 s . . . and let g P Sn . and so by (12. . ik ) g ˆ (i1 g i2 g . and note that x and gÀ1 xg have the same cycle-shape. . By Proposition 12. exists g P Sn sending j . .14) g À1 xg ˆ g À1 (a1 X X X a k 1 ) gg À1 (b1 X X X b k 2 ) g X X X g À1 (c1 X X X ck s ) g ˆ (a1 g X X X a k 1 g)(b1 g X X X b k 2 g) X X X (c1 g X X X c k s g)X We call (k1 . Then gÀ1 xg is the k-cycle (i1 g i2 g . . 1 k g À1 xg ˆ yX We have proved the following result. say x ˆ (a1 X X X a k 1 ) X X X (c1 X X X ck s ). 1 k 1 k (products of disjoint cycles). with k1 > k2 > .Conjugacy classes Conjugacy classes of S n 109 We shall later need to know the conjugacy classes of the symmetric group Sn . ir g( g À1 xg) ˆ i r xg ˆ i r‡1 g (or i1 g if r ˆ k)X Also.14). there a1 3 a9 . Now consider an arbitrary permutation x P Sn . . . . . For ir P A. ik g). ik }. . given any two permutations x. for g P Sn we have (12. Our ®rst observation is simple but crucial. . On the other hand. a product of disjoint cycles. a ig( gÀ1 xg) ˆ ixg ˆ igX Hence gÀ1 (i1 i2 . for 1 < i < n and i P A. . y ˆ (a9 X X X a9 1 ) X X X (c9 X X X c9 s ). > ks . . . . . ik ) in Sn . .13. . . Proof Write A ˆ {i1 . ks ) the cycle-shape of x. y of the same cycle-shape. ik g). 12. . Write x ˆ (a1 X X X ak 1 )(b1 X X X bk 2 ) X X X (c1 X X X ck s ). as required.13 Proposition Let x be a k-cycle (i1 i2 . .

the conjugacy class x Sn of x in Sn consists of all permutations in Sn which have the same cycle-shape as x. (1 3)(2 4).8) are as follows: Representative Class size g | gG | |CG ( g)| 1 1 24 (1 2) 6 4 (1 2 3) (1 2)(3 4) 8 3 3 8 (1 2 3 4) 6 4 We check our arithmetic by noting that jS4 j ˆ 1 ‡ 6 ‡ 8 ‡ 3 ‡ 6X . we simply count the number of 2-cycles. (1 3).) The number of 3-cycles is 4 3 2 (4 for the choice of ®xed point and 2 because there are two 3-cycles ®xing a given point). Similarly. (1 2). (1 2 3 4)X To calculate the sizes of the conjugacy classes. (The notation … n † means the binomial coef®cient 2 r n3a(r3(n À r)3). 2) and is f(1 2)(3 4). with representatives (see De®nition 12. (1 3 2)} Cycle-shape (1) (2) (3) (2) The conjugacy class of (1 2)(3 4) in S4 consists of all the elements of cycle-shape (2. the conjugacy class sizes | gG | and the centralizer orders |CG ( g)| (obtained using Theorem 12. and so on. 2. the conjugacy class representatives g. (1 4)(2 3)gX (3) There are precisely ®ve conjugacy classes of S4 .16 Examples (1) The conjugacy classes of S3 are Class {1} {(1 2). there are three elements of cycle-shape (2. 12. Thus for G ˆ S4 . (2 3)} {(1 2 3).15 Theorem For x P Sn . 2) and there are six 4-cycles.110 Representations and characters of groups 12.4): 1. The number of 2-cycles is equal to the number of pairs that can be chosen from {1. ÀÁ which is 4 ˆ 6. 3. 4}. (1 2 3). (1 2)(3 4). 3-cycles.

g 1 (1 2) | gG | 1 10 |CG ( g)| 120 12 (1 2 3) (1 2)(3 4) 20 15 6 8 (1 2 3 4) 30 4 111 (1 2 3)(4 5) (1 2 3 4 5) 20 24 6 5 Conjugacy classes of A n Given an even permutation x P A n .Conjugacy classes (4) Similarly.17 Proposition Let x P An with n . and what happens when equality fails. (1) If x commutes with some odd permutation in Sn . is of course contained in x S n . The conjugacy class x A n of x in A n . For an easy example where equality does not hold. so again y P x An . 12. and if h is odd then gh P An and y ˆ hÀ1 xh ˆ hÀ1 g À1 xgh ˆ ( gh)À1 x( gh). given by x A n ˆ f g À1 xg: g P An g. then x Sn ˆ x An . x A n might not be equal to x S n . If h is even then y P x An . with representatives x and (1 2)À1 x(1 2). we have seen in Theorem 12. (2) Assume that x does not commute with any odd permutation. Let y P x S n . however. (2) If x does not commute with any odd permutation in Sn then x Sn splits into two conjugacy classes in An of equal size. here x A3 ˆ fxg. and so x Sn ˆ x An . Then CSn (x) ˆ CAn (x)X . while x S3 ˆ {x. The next result determines precisely when x An and x Sn are equal. so that y ˆ hÀ1 xh for some h P Sn .15 that the conjugacy class x S n consists of all permutations in S n which have the same cycle-shape as x. 1. Proof (1) Assume that x commutes with an odd permutation g. the corresponding table for G ˆ S5 is Rep. Thus x Sn # x An . consider x ˆ (1 2 3) P A3 . x À1 }.

Proposition 12.17 implies that (1 2)(3 4) A4 ˆ (1 2)(3 4) S4 ˆ f(1 2)(3 4). 2) and (5). together with the permutations of cycle-shapes (2. 2) and (3). j 12. but (1 2 3 4 5) commutes with no odd permutation. The elements (1 2 3) and (2 3)(4 5) commute with the odd permutation (4 5). Since (1 2)(3 4) commutes with the odd permutation (1 2). the conjugacy classes x An and ((1 2)À1 x(1 2)) An must 2 be disjoint and of equal size. Now x Sn ˆ fhÀ1 xh: h is eveng ‘ fhÀ1 xh: h is oddg ˆ x An ‘ ((1 2)À1 x(1 2)) An X Since |x An | ˆ 1|x Sn |. (1 3)(2 4).112 Representations and characters of groups Hence by Theorem 12.17. Hence by Proposition 12. Thus the conjugacy classes of A4 are Representative Class size Centralizer order 1 1 12 (1 2)(3 4) 3 4 (1 2 3) 4 3 (1 3 2) 4 3 (as jAn j ˆ 1jSn j) 2 (2) We ®nd the conjugacy classes of A5 .17. an even permutation. (1 4)(2 3)gX However. as we wished to show. so g is 1. the 3-cycle (1 2 3) commutes with no odd permutation: for if gÀ1 (1 2 3) g ˆ (1 2 3) then (1 2 3) ˆ (1 g 2 g 3 g) by Proposition 12. we observe that fhÀ1 xh: h is oddg ˆ ((1 2)À1 x(1 2)) An since every odd permutation has the form (1 2)a for some a P An . The elements of A4 are the identity. jx An j ˆ jAn : CAn (x)j ˆ 1jSn : CAn (x)j 2 ˆ 1jSn : CSn (x)j ˆ 1jx Sn jX 2 2 Next. (1 2 3) S4 splits into two conjugacy classes in A4 of size 4. (Check this by using the argument in (1) above. with representatives (1 2 3) and (1 2)À1 (1 2 3)(1 2) ˆ (1 3 2).13.8. the .18 Examples (1) We ®nd the conjugacy classes of A4 . The non-identity even permutations in S5 are those of cycle-shapes (3). (1 2 3) or (1 3 2).) Hence by Proposition 12. (2.

these conjugacy classes have sizes 1. Conversely. we have gÀ1 hg P H.19. g P G A g À1 hg P H.19 Proposition Let H be a subgroup of G.16(3). H is a union of conjugacy classes of S4 . and so hG # H. (1 2)(3 4). there are just four possibilities: j Hj ˆ 1. j 12. if H v G then for all h P H. and 1 P H. so gÀ1 Hg # H.17(2).Conjugacy classes 113 conjugacy classes of A5 are represented by 1. Using Proposition 12. 6. Since j Hj divides 24 by Lagrange's Theorem. 3. hP H and so H is a union of conjugacy classes of G. g P G. 12.20 Example We ®nd all the normal subgroups of S4 . Then H v G if and only if H is a union of conjugacy classes of G. Thus H v G. As we saw in Example 12. we see that the class sizes and centralizer orders are as follows: Representative Class size Centralizer order 1 1 60 (1 2 3) 20 3 (1 2)(3 4) 15 4 (1 2 3 4 5) 12 5 (1 3 4 5 2) 12 5 Normal subgroups Normal subgroups are related to conjugacy classes by the following elementary result. (1 2 3 4 5) and (1 2)À1 (1 2 3 4 5)(1 2) ˆ (1 3 4 5 2). (1 2 3). then h P H. 6. Proof If H is a union of conjugacy classes. Then by Proposition 12. 8. 1 ‡ 3. Let H v S4 . 1 ‡ 8 ‡ 3 or 1 ‡ 6 ‡ 8 ‡ 3 ‡ 6X . Therefore ‘ Hˆ hG .

and in the third case H ˆ A4 . we write it as V4 (V stands for `Viergruppe'. 12. Proof First we show that each C i belongs to Z(CG). . Let Ci consist of the r distinct conjugates yÀ1 gy1. hÀ1 C i h ˆ r ˆ jˆ1 r ˆ jˆ1 yÀ1 gyj X j hÀ1 yÀ1 gyj hX j . (1 4)(2 3)gX This is easily checked to be a subgroup of S4 . meaning `four-group'). Recall from De®nition 9. . C l form a basis of Z(CG).22 Proposition The class sums C1 . Cl be the distinct conjugacy classes of G. (1 3)(2 4). In the case where j Hj ˆ 1 ‡ 3. There is a convenient basis for this subspace which can be described in terms of the conjugacy classes of G. C l of CG are called class sums. . . we have H ˆ 1 S4 ‘ (1 2)(3 4) S4 ˆ f1. .114 Representations and characters of groups In the ®rst case H ˆ {1}. . . (1 2)(3 4). . S4 . A4 and V4 ˆ f1. . in the last case H ˆ S4 . . . We have now shown that S4 has exactly four normal subgroups: f1g. (1 2)(3 4). . . 12. (1 3)(2 4). so 1 r Ci ˆ For all h P G. yÀ1 gyr of an element g. For 1 < i < l. . .21 De®nition Let C1 . . (1 4)(2 3)gX The centre of a group algebra In this ®nal section we link the conjugacy classes of the group G to the centre of the group algebra CG.12 that the centre of CG is Z(CG) ˆ fz P CG: zr ˆ rz for all r P CGgX We know that Z(CG) is a subspace of the vector space CG. de®ne ˆ Ci ˆ g P CGX gPCi The elements C1 .

the coef®cient ë g of g is equal to the coef®cient ë hÀ1 gh of hÀ1 gh. Every group is a union of conjugacy classes. . . For an element x of a group G. . . . b ‡ a2 b. a ‡ a3 . we have rh ˆ hr. observe that C1 . and so C i P Z(CG). For h P G. the centralizer CG (x) is the set of . a basis for Z(CD8 ) is 1. the function g 3 ë g is constant on €l conjugacy classes of G. It follows that r ˆ iˆ1 ë i C i where ë i is the j coef®cient ë gi for some gi P Ci . then all ë i ˆ 0 as the classes C1 . 2. Let r ˆ € À1 gPG ë g g P Z(CG). This completes the proof. That is to say. a2 . and distinct conjugacy classes are disjoint. C l are linearly independent: for if €l iˆ1 ë i C i ˆ 0 (ë i P C). That is. C l span Z(CG). . That is. j and so h C i h ˆ C i . so h rh ˆ r. (1 2) ‡ (1 3) ‡ (2 3). ab ‡ a3 bX Summary of Chapter 12 1. .12). hence with all € hPG ë h h P CG.3. . . Next. 12.23 Examples (1) From Example 12. .Conjugacy classes 115 As j runs from 1 to r. since j hÀ1 yÀ1 gyj h ˆ hÀ1 yÀ1 gyk h D yÀ1 gyj ˆ yÀ1 gyk X j k j k Hence r ˆ jˆ1 À1 hÀ1 yÀ1 gyj h ˆ C i . . (1 2 3) ‡ (1 3 2)X (2) From (12. the elements hÀ1 yÀ1 gyj h run through Ci . Cl are pairwise disjoint by Corollary 12. C i h ˆ hC i X Therefore each C i commutes with all h P G. It remains to show that C1 . ˆ ˆ ë g hÀ1 gh ˆ ë g gX gPG gPG So for every h P G. a basis for Z(CS3 ) is 1. .16(1).

(b) Suppose that n > 3 and that | Z(G)| ˆ p. The class sums in CG form a basis for the centre of CG. 4. The conjugacy classes of Sn correspond to the cycle-shapes of permutations in Sn .8. Let G ˆ Sn . and let n be a positive integer. and the number of elements in the conjugacy class x G is equal to |G:CG (x)|. Show that j(1 2 3)(4 5 6) G j ˆ 40 and j(1 2)(3 4)(5 6) G j ˆ 15. Let p be a prime number. Find the conjugacy classes of the quaternion group Q8. .10 to show that Z(G) Tˆ {1}. and ®nd the sizes of the other conjugacy classes of S6 . 5. What are the cycle-shapes of those permutations x P A6 for which x A6 Tˆ x S6 ? 5. 3.20. Give a basis of the centre of the group algebra CQ8. It is a subgroup of G. (Hint: use the method of Example 12. 3. Verify that your solution satis®es Theorem 12. n n (b) Show that j(1 2 3) G j ˆ 2…3 † and j(1 2)(3 4) G j ˆ 3…4 †. (a) Use the Class Equation 12. Suppose that G is a group of order pn . If G is a group and x P G. Show that A5 is a simple group. Prove that G has a conjugacy class of size p. (c) Now let n ˆ 6. Let G be a ®nite group and suppose that g P G and z P Z(G). n (a) Prove that j(1 2) G j ˆ …2 † and ®nd CG ((1 2)).) 6. show that CG (x) is a subgroup of G which contains Z(G).116 Representations and characters of groups elements of G which commute with x. 7. Exercises for Chapter 12 1. 2. Prove that the conjugacy classes gG and ( gz) G have the same size.) 4. If x P An then x Sn ˆ x An if and only if x commutes with some odd permutation in Sn . (There are 11 conjugacy classes in all.

C). it appears that we must keep track of all the n2 entries in each matrix gr. basic problems. such as deciding whether or not a given representation is irreducible. The function ÷: G 3 C is called the character of the representation r.1 De®nition If A ˆ (aij ) is an n 3 n matrix. Moreover. Characters of representations have many remarkable properties. With each n 3 n matrix gr ( g P G) we associate the complex number given by adding all the diagonal entries of the matrix. In this chapter we present some basic properties and examples. and they are the fundamental tools for performing calculations in representation theory. since from the de®nition of a representation r: G 3 GL(n. The theory of characters will occupy a considerable portion of the rest of the book. we shall show later that two representations have the same character if and only if they are equivalent. the trace of A is the sum of the diagonal entries of A. then the trace of A. For example. can be resolved by doing some easy arithmetic with the character of the representation. C) is a representation of the ®nite group G.13 Characters Suppose that r: G 3 GL(n. whereas the character records just one number for each matrix. and call this number ÷( g). is given by tr A ˆ n ˆ iˆ1 aii X That is. These facts are surprising. written tr A. 117 . The trace of a matrix 13.

aij bji ˆ n n ˆ ˆ jˆ1 iˆ1 bji aij ˆ tr (BA)X Notice that. Then tr (A ‡ B) ˆ tr A ‡ tr B. Therefore tr (A ‡ B) ˆ and tr (AB) ˆ For the last part. tr (T À1 AT ) ˆ tr ((T À1 A)T ) ˆ tr (T (T À1 A)) ˆ tr AX (by the second part ) j n n ˆ ˆ iˆ1 jˆ1 n ˆ iˆ1 (aii ‡ bii ) ˆ n ˆ iˆ1 aii ‡ n ˆ iˆ1 bii ˆ tr A ‡ tr B. Then the character of V is the function ÷: G 3 C de®ned by ÷( g) ˆ tr [ g]B ( g P G)X The character of V does not depend on the basis B . then [ g]B 9 ˆ T À1 [ g]B T . and the ii-entry of AB is €n jˆ1 aij bji. the trace function is not multiplicative. Characters 13.2 Proposition Let A ˆ (aij ) and B ˆ (bij ) be n 3 n matrices.118 Representations and characters of groups 13. and tr (AB) ˆ tr (BA)X Moreover.3 De®nition Suppose that V is a CG-module with a basis B . if T is an invertible n 3 n matrix. since if B and B 9 are bases of V. tr (AB) need not equal (tr A)(tr B). unlike the determinant function. that is. then tr (T À1 AT ) ˆ tr AX Proof The ii-entry of A ‡ B is aii ‡ bii .

Then by (7. there are a basis B 1 of V and a basis B 2 of W such that [ g]B 1 ˆ [ g]B 2 for all g P GX Consequently tr [ g]B 1 ˆ tr [ g]B 2 for all g P G. Then [x]B ˆ [ g À1 yg]B ˆ [ g]À1 [ y]B [ g]B X B Hence by Proposition 13. we de®ne the character of a representation r: G 3 GL(n.7). and ÷ is reducible if it is the character of a reducible CG-module. j The result corresponding to Proposition 13. so that x ˆ gÀ1 yg for some g P G. That is. and so V and W have the same character.5(1) for representations is that equivalent representations have the same character. Let V be a CG-module. Proof (1) Suppose that V and W are isomorphic CG-modules. and so by Proposition 13. then ÷(x) ˆ ÷( y) for all characters ÷ of G. Therefore ÷(x) ˆ ÷( y). . C) to be the character ÷ of the corresponding CGmodule C n . tr [ g]B 9 ˆ tr [ g]B for all g P GX Naturally enough.Characters 119 for some invertible matrix T (see (2. namely ÷( g) ˆ tr ( gr) ( g P G)X 13. and let B be a basis of V.2. ÷ is an irreducible character of G if ÷ is the character of an irreducible CG-module. (2) If x and y are conjugate elements of the group G. 13.5 Proposition (1) Isomorphic CG-modules have the same character. we have tr [x]B ˆ tr [ y]B . You will have noticed that we are writing characters as functions on the left. where ÷ is the character of V.24)).4 De®nition We say that ÷ is a character of G if ÷ is the character of some CGmodule. Further. we write ÷( g) and not g÷.2. (2) Assume that x and y are conjugate elements of G.

2(1)). then they are isomorphic. thus B is the basis v1 . bÀ1 ab ˆ aÀ1 l. br ˆ ar ˆ 0 À1 À1 0 (see Example 3.1. together with the character ÷ of V. v3 . The following table records g. g [ g]B ÷( g) 1 (1 2) (1 3) H 1 d0 0 0 0 1 0e 0 1 3 I H 0 d1 0 1 0 0 0e 0 1 1 I H 0 0 d0 1 1 0 1 I 1 0e 0 .5(1): if two CG-modules have the same character. we shall prove an astonishing converse of Proposition 13. and let r: G 3 GL(2. (We obtain ÷( g) by adding the two entries on the diagonal of gr.) g gr ÷( g) 1 1 0 0 1 2 a 0 1 À1 0 0 a2 À1 0 0 À1 À2 a3 0 À1 1 0 0         g gr ÷( g)  b 1 0 0 À1 0   ab 0 À1 0 À1 0   a2 b À1 0 0 1 0   a3 b 0 1 0 1 0  (2) Let G ˆS3 . 13. v2 .6 Examples (1) Let G ˆ D8 ˆ ka.120 Representations and characters of groups Later. gr and ÷( g) as g runs through G. Let B be the natural basis of V. C) be the representation for which     1 0 0 1 .10). The matrices [ g]B ( g P G) are given by Exercise 4. where v i g ˆ v ig for 1 < i < 3 and all g P G. We record these matrices. and take V to be the 3-dimensional permutation module for G over C (see De®nition 4. b: a4 ˆ b2 ˆ 1. Let ÷ be the character of this representation.

G has just three irreducible characters ÷1 . it is much quicker to write down the single complex number ÷( g) for the group element g than to record the matrix which corresponds to g. then the irreducible characters of G are ÷1 . (4) Let G ˆ D6 ˆ ka. U3. In Example 10. we found a complete set of non-isomorphic irreducible CG-modules U1.8(2). Nevertheless.8(2). ÷2 . U2. r2 . Thus if ÷ i is the character of Ui for 1 < i < 3. By Theorem 9. the character encapsulates a great deal of information about the representation. ÷2 and ÷3 . Moreover. the characters given take few distinct values. . ÷3 .8. b: a3 ˆ b2 ˆ 1. This re¯ects the fact that by Proposition 13. bÀ1 ab ˆ aÀ1 l (so G  S3 ). with values g ÷1 ( g) ÷2 ( g) ÷3 ( g) 1 1 1 1 a 1 ù ù2 a2 1 ù2 ù where ù ˆ e2ðia3 . r3 given in Example 10. This will become clear as the theory of characters develops. every character is constant on conjugacy classes of G. The values of these characters on the elements of G can be calculated from the corresponding representations r1 .5(2). and they are as follows: g ÷1 ( g) ÷2 ( g) ÷3 ( g) 1 1 1 2 a 1 1 À1 a2 1 1 À1 b 1 À1 0 ab 1 À1 0 a2 b 1 À1 0 Notice that in all the above examples.Characters g [ g]B ÷( g) H (2 3) I H (1 2 3) I H (1 3 2) I 0 1 0 0e 1 0 0 121 1 0 0 d0 0 1e 0 1 0 1 0 1 d0 0 1 0 0 0 1e 0 0 d1 0 (3) Let G ˆ C3 ˆ ha: a3 ˆ 1 i.

6(4) we saw that the irreducible characters of D6 have degrees 1. Then for all v P V X . Finding all the irreducible characters is usually dif®cult. Suppose that g P G and g has order m. in particular.9 Proposition Let ÷ be the character of a CG-module V.6(1) we gave a character of D8 of degree 2. The values of a character The next result gives information about the complex numbers ÷( g). Characters of degree 1 are called linear characters. called the trivial character of G.6(2) we gave a character of S3 of degree 3. these are the only non-zero characters of G which are homomorphisms (see Exercise 13.8(1)) is a linear character. namely the trivial character.8 Examples (1) In Example 13.122 Representations and characters of groups 13. In fact. they are.4). 13. they are all linear characters. (3) The character of the trivial CG-module (see De®nition 4. where ÷ is a character of G and g P G. 1 and 2. 13. (2) If V is any 1-dimensional CG-module. then for each g P G there is a complex number ë g such that v g ˆ ë gv The character ÷ of V is given by ÷( g) ˆ ë g ( g P G) and ÷ has degree 1.7 De®nition If ÷ is the character of the CG-module V. Observe that Theorem 9. Thus 1 G : g 3 1 for all g P GX Given any group G. then the dimension of V is called the degree of ÷.8 gives all the irreducible characters of ®nite abelian groups. We denote it by 1 G . of course. Every linear character of G is a homomorphism from G to the multiplicative group of non-zero complex numbers. irreducible characters. and in 13. we therefore know at least one of the irreducible characters of G. in 13.

and so ÷( g) ˆ ÷( g). that is. (3) We have H f [ gÀ1 ]B ˆ d ùÀ1 1 0 FF 0 F ùÀ1 n I g e and so ÷( gÀ1 ) ˆ ùÀ1 ‡ . ÷( g) is a real number if g is conjugate to gÀ1 . When the element g of G has order 2. we can be much more speci®c about the possibilities for ÷( g): . ÷( gÀ1 ) ˆ ÷( g). . and so ÷(1) ˆ dim V. Therefore ÷( g À1 ) ˆ ù1 ‡ X X X ‡ ù n ˆ ÷( g)X (4) If g is conjugate to gÀ1 then ÷( g) ˆ ÷( gÀ1 ) by Proposition 13.Characters (1) (2) (3) (4) ÷(1) ˆ dim V. ÷( g) is a sum of mth roots of unity. and let B be a basis of V.5(2). 123 Proof (1) Let n ˆ dim V. the n 3 n identity matrix.11 there is a basis B of V such that H I ù1 0 f g FF [ g]B ˆ d e F ùn 0 where each ù i is an mth root of unity. Consequently ÷(1) ˆ tr [1]B ˆ tr I n ˆ n. Every complex mth root of unity ù 1 n satis®es ùÀ1 ˆ ù. . ÷( g) is j real. which is the complex conjugate of eiW . a sum of mth roots of unity. Therefore ÷( g) ˆ ù1 ‡ X X X ‡ ù n . Also ÷( gÀ1 ) ˆ ÷( g) by (3). (eiW )À1 ˆ eÀiW . ‡ ùÀ1 . Then the matrix [1]B of the identity element 1 relative to B is equal to In . (2) By Proposition 9. since for all real W.

and ÷( g)  ÷(1) mod 2X Proof By Proposition 13. showing that we can determine the kernel of a representation just from knowledge of its character.10 Corollary Let ÷ be a character of G. and ÷(1) ˆ r ‡ sX Certainly then. By Proposition 9. and since r À s ˆ r ‡ s À 2s  r ‡ s mod 2.9. Then (13X12) j÷( g)j ˆ jù1 ‡ X X X ‡ ù n j ˆ ÷(1) ˆ nX for some ë P CX . Then ÷( g) is an integer. and suppose that g has order m. we have ÷( g) ˆ ù1 ‡ X X X ‡ ù n .124 Representations and characters of groups 13. and let ÷ be the character of r. and s are À1. and let g be an element of order 2 in G. so |÷( g)| ˆ n ˆ ÷(1). C) be a representation of G. j Our next result gives the ®rst inkling of the importance of characters. Conversely.11. Proof (1) Let g P G. Then each ù i is ‡1 or À1.11 Theorem Let r: G 3 GL(n. suppose that |÷( g)| ˆ ÷(1). If gr ˆ ëIn with ë P C. j÷( g)j ˆ ÷(1) D gr ˆ ëI n (2) Ker r ˆ { g P G: ÷( g) ˆ ÷(1)}. Suppose r of them are ‡1. there is a basis B of C n such that H I ù1 0 f g FF [ g]B ˆ d e F ùn 0 where each ù i is an mth root of unity. where n ˆ ÷(1) and each ù i is a square root of unity. ÷( g) P Z. and ÷( g) ˆ në. 13. we have ÷( g)  ÷(1) mod 2. then ë is an mth root of unity. (1) For g P G. so that ÷( g) ˆ r À s.

Conversely. 13.6(4). if r is a representation of G with character ÷. and so g P Ker r. then the kernel of ÷. Part (2) follows.Characters Note now that for any complex numbers z1 . . whence ë ˆ 1. j. zn . Thus H I ù1 0 f g FF [ g]B ˆ d e ˆ ù1 I n X F ù1 0 Hence for all bases B 9 of C n we have [ g]B 9 ˆ ù1 In . we have jz1 ‡ X X X ‡ zn j < jz1 j ‡ X X X ‡ jzn j. with the following values: . In particular. We call ÷ a faithful character if Ker ÷ ˆ {1}. . we de®ne the kernel of a character as follows.) Since |ù i | ˆ 1 for all i. is de®ned by Ker ÷ ˆ f g P G: ÷( g) ˆ ÷(1)gX By Theorem 13. 125 with equality if and only if the arguments of z1 . . (2) If g P Ker r then gr ˆ In . Therefore gr ˆ In . This implies that ÷( g) ˆ ë÷(1). This completes the proof of (1).12) that ù i ˆ ù j for all i. .14 Examples (1) According to Example 13. consider the picture in the Argand diagram. written Ker ÷. . and so ÷( g) ˆ n ˆ ÷(1). . Ker ÷ v G. and so gr ˆ ù1 In . Then by (1). 13. the irreducible characters of the group G ˆ D6 ˆ ka. . . bÀ1 ab ˆ aÀ1 l are ÷1 .11(2). we have gr ˆ ëIn for some ë P C. j Motivated by Theorem 13. (To see this. suppose that ÷( g) ˆ ÷(1). b: a3 ˆ b2 ˆ 1.11(2). ÷3 . zn are all equal.13 De®nition If ÷ is a character of G. ÷2 . we deduce from (13. then Ker r ˆ Ker ÷.

Theorem 13.6(1): g ÷( g) 1 2 a 0 a2 À2 a3 0 b 0 ab 0 a2 b 0 a3 b 0 Then Ker ÷ ˆ {1}. and let ÷ be the character of G given in Example 13. For a character ÷ of G. And since |÷(a2 )| ˆ |À2| ˆ ÷(1).11(1) implies that if r: G 3 GL(2. C). 13. so ÷ is a faithful character. Thus ÷( g) ˆ tr ( gr) ( g P G)X If A ˆ (aij ) is an n 3 n matrix over C. Then ÷ is a character of G. then a2 r ˆ ÀI. then so is ÷. We next prove a result which is sometimes useful for constructing a new character from a given one. Observe that if A ˆ (aij ) and B ˆ (bij ) are n 3 n matrices over C. (2) Let G ˆ D8 ˆ ka. b: a4 ˆ b2 ˆ 1. If ÷ is irreducible. then we de®ne A to be the n 3 n matrix (a ij ). ÷3 is a faithful irreducible character of D6. de®ne ÷: G 3 C by ÷( g) ˆ ÷( g) ( g P G)X Thus the values of ÷ are the complex conjugates of the values of ÷. . Proof Suppose that ÷ is the character of a representation r: G 3 GL(n. bÀ1 ab ˆ aÀ1 l. Ker ÷2 ˆ kal and Ker ÷3 ˆ {1}.15 Proposition Let ÷ be a character of G.126 g ÷1 ( g) ÷2 ( g) ÷3 ( g) 1 1 1 2 Representations and characters of groups a 1 1 À1 a2 1 1 À1 b 1 À1 0 ab 1 À1 0 a2 b 1 À1 0 Then Ker ÷1 ˆ G. C) is a representation with character ÷. then (13X16) (AB) ˆ A B. In particular.

Hence ÷ is irreducible if and only if ÷ is irreducible. k let ÷ i be the character of Vi and di ˆ ÷ i (1). . . . Proof This is immediate from (7. It is clear that if r is reducible then r is reducible. the ij-entry of AB. j The regular character 13.11). a direct sum of irreducible CG-modules Ui. and for i ˆ 1. We write the regular character as ÷reg . . It follows from (13.10). . First we need a preliminary result. . Then the character of V is equal to the sum of the characters of the CG-modules U1. V k be a complete set of non-isomorphic irreducible CG-modules (see De®nition 11. j ( g P G) 13. In Theorem 13. C) de®ned by gr ˆ ( gr) is a representation of G.18 Proposition Let V be a CG-module.Characters since the ij-entry of A B is n ˆ kˆ1 127 a ik b kj . X X X . €n which is equal to the complex conjugate of kˆ1 aik bkj . we shall express the regular character in terms of the irreducible characters of G. . Since tr ( gr) ˆ tr ( gr) ˆ tr ( gr) ˆ ÷( g) ( g P G).17 De®nition The regular character of G is the character of the regular CG-module.19 Theorem Let V1 . Ur. . 13.16) that the function r: G 3 GL(n. and suppose that V ˆ U1 È X X X È Ur . Then ÷reg ˆ d 1 ÷1 ‡ X X X ‡ d k ÷ k X . the character of the representation r is ÷.19.

then ÷reg (1) ˆ jGj.19 and Proposition 13. CG  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ). Now let g P G with g Tˆ 1. Therefore the ith row of the matrix [ g]B has zeros in every place except column j.18. in particular. and let B be the basis g1 . It follows that ÷reg ( g) ˆ tr [ g]B ˆ 0X j 13.20 for the group G ˆ D6 . ÷ 3 : g ÷1 ( g) ÷2 ( g) ÷3 ( g) 1 1 1 2 a 1 1 À1 a2 1 1 À1 b 1 À1 0 ab 1 À1 0 a2 b 1 À1 0 We calculate ÷1 ‡ ÷2 ‡ 2÷3 : (÷1 ‡ ÷2 ‡ 2÷3 )( g) 6 0 0 0 0 0 . Then for 1 < i < n. we have gi g ˆ gj for some j with j Tˆ i.21 Example We illustrate Theorem 13. By Example 13. gn be the elements of G. ÷reg (1) ˆ dim CG ˆ |G|. . . . .9.128 Representations and characters of groups Proof By Theorem 11. where for each i there are di factors V i .20 Proposition If ÷reg is the regular character of G. and are given in the next result.6(4). Now the result follows from Proposition 13. the ii-entry is zero for all i. . gn of CG. . ÷2 . 13. By Proposition 13. . . j The values of ÷reg on the elements of G are easily described.9(1). and ÷reg ( g) ˆ 0 if g Tˆ 1X Proof Let g1 . the irreducible characters of G are ÷1 .

v i g ˆ v ig (1 < i < n) (see De®nition 4. (1 2). . and the value 0 on all non-identity elements of G. where for all g P G.Characters 129 This is the regular character of G. Let B denote the basis v1 .19. Suppose that G is a subgroup of Sn .20. Then by Example 12. by Theorem 13. . so that G is a group of permutations of {1. The permutation module V for G over C has basis v1 . and we now describe this.23 Example Let G ˆ S4 . there is an easy construction using the permutation module which produces a character of degree n. . . . . Therefore the character ð of the permutation module V is given by ð( g) ˆ (the number of i such that ig ˆ i)X For g P G. illustrating Proposition 13. (1 2)(3 4).10). 13. . with representatives 1. (1 2 3). n}. . and it takes the value |G| on 1. . Then the iientry in the matrix [ g]B is 0 if ig Tˆ i. v n . . let fix ( g) ˆ fi: 1 < i < n and ig ˆ igX Then (13X22) ð( g) ˆ jfix( g)j ( g P G)X We call ð the permutation character of G.16(3). v n . and is 1 if ig ˆ i. Permutation characters In the case where G is a subgroup of the symmetric group Sn . . (1 2 3 4)X The permutation character ð takes the values gi ð( gi ) 1 4 (1 2) 2 (1 2 3) 1 (1 2)(3 4) 0 (1 2 3 4) 0 . . G has ®ve conjugacy classes.

so U is a CG-submodule of V. so the character of U is the trivial character 1 G (see Example 13.18(1).130 Representations and characters of groups 13. . Indeed. (1 2 3). U is isomorphic to the trivial CG-module.25 Example Let G ˆ A4 . (1 2)(3 4). (1 3 2)X The values of the character í of G are gi í( gi ) 1 3 (1 2)(3 4) À1 (1 2 3) 0 (1 3 2) 0 Summary of Chapter 13 1. Then the function í: G 3 C de®ned by í( g) ˆ jfix ( g)j À 1 is a character of G. Then ð ˆ 1 G ‡ í. By Maschke's Theorem 8. and let u ˆ v1 ‡ X X X ‡ v n . v n be the basis of the permutation module V as above. Characters are constant on conjugacy classes.24 Proposition Let G be a subgroup of Sn . . .8(3)). .1. A character is obtained from a representation by taking the trace of each matrix. Proof Let v1 . By Example 12. a subgroup of S4 . so |®x( g)| ˆ 1 ‡ í( g) for all g P G. the conjugacy classes of G are represented by 1. there is a CG-submodule W of V such that V ˆ U È WX Let í be the character of W. and U ˆ sp (u)X Observe that ug ˆ u for all g P G. . 2. and therefore í( g) ˆ jfix( g)j À 1 ( g P G)X j ( g P G) 13.

Exercises for Chapter 13 1. and let r1 . 6. the complex number ÷( g) is a sum of roots of unity. Prove that there exists an mth root of unity ë P C such that for all g P G. then Z(G) ˆ { g P G: |÷( g)| ˆ ÷(1)}. Let G ˆ D12 ˆ ka. and ÷( gÀ1 ) ˆ ÷( g). br2 ˆ 1 0  0 X À1 ( g P G) Find the characters of r1 and r2 . Assume that ÷ is an irreducible character of G. 2. 3.Characters 3. check that your answers are consistent with Theorem 13. br1 ˆ (where ù ˆ e2ðia3 ) 0 ùÀ1 1 0 and ar2 ˆ  À1 0   0 1 . If G is a subgroup of Sn . 7. The character of a representation determines the kernel of the representation. Find all the irreducible characters of C4 . Write the regular character of C4 as a linear combination of these. and all g P G. Isomorphic CG-modules have the same character. 5. Suppose that z P Z(G) and that z has order m. 4. then the function í which is given by í( g) ˆ jfix( g)j À 1 is a character of G. 5. Prove that if ÷ is a faithful irreducible character of the group G. Prove that the only non-zero characters of G which are homomorphisms are the linear characters. For all characters ÷ of G. Find also Ker r1 and Ker r2 . b: a6 ˆ b2 ˆ 1. r2 be the representations of G for which     ù 0 0 1 ar1 ˆ . Let ÷ be the character of the 7-dimensional permutation module for S7 . 131 4. ÷(zg) ˆ ë÷( g)X 6. The regular character ÷reg of G takes the value |G| on the identity and the value 0 on all other elements of G. bÀ1 ab ˆ aÀ1 l.11. Find ÷(x) for x ˆ (1 2) and for x ˆ (1 6)(2 3 5). .

Let g be a group of order 2k. where k is an odd integer. Hint: use Exercise 7. By considering the regular representation of G. (Compare Corollary 13. 9. or (2) G has a normal subgroup of index 2.) 10. then ÷(x) Tˆ ÷(1) for some irreducible character ÷ of G. Show that either (1) ÷( g)  ÷(1) mod 4. Let r be a representation of the group G over C. (a) Show that ä: g 3 det ( gr) ( g P G) is a linear character of G.10. show that G has a normal subgroup of index 2.132 Representations and characters of groups 7. Let ÷ be a character of a group G. 8. (b) Prove that G/Ker ä is abelian. . Show that G has a normal subgroup of index 2. Prove that if x is a non-identity element of the group G. (c) Assume that ä( g) ˆ À1 for some g P G. and let g be an element of order 2 in G.

if W. Also. The proofs rely on an inner product involving the characters of a group. and we describe this ®rst.) 14. then we de®ne W ‡ ö: G 3 C by (W ‡ ö)( g) ˆ W( g) ‡ ö( g) and we de®ne ëW: G 3 C by ëW( g) ˆ ë(W( g)) ( g P G)X ( g P G) (We write these functions on the left to agree with our notation for characters. and ë P C. we describe a method for decomposing a given CG-module as a direct sum of CG-submodules. ö are functions from G to C.1 Example Let G ˆ C3 ˆ ka: a3 ˆ 1l. using characters. if we adopt the natural rules for adding functions and multiplying functions by complex numbers. Inner products The characters of a ®nite group G are certain functions from G to C.14 Inner products of characters We establish some signi®cant properties of characters in this chapter. The set of all functions from G to C forms a vector space over C. and in particular we prove the striking result (Theorem 14. and suppose that W: G 3 C and ö: G 3 C are given by 133 .21) that if two CG-modules have the same character then they are isomorphic. That is.

W(a2 ) ˆ À1 and ö(1) ˆ ö(a) ˆ ö(a2 ) ˆ 1.2) (a) kW. öl for all ë1 . Notice that condition (a) implies that kW. Wi for all W. ë2 P C and all vectors W1 . W(a) ˆ i. (c) kW. 0 if W Tˆ 0. öi ˆ W( g)ö( g)X jGj gPG . öl ‡ ë2 kW2 . (b) kë1 W1 ‡ ë2 W2 . Then W ‡ ö and 3W are given by 1 W ‡ö 3W 3 6 a 1‡i 3i a2 0 À3 We shall often think of functions from G to C as row vectors. Wl . ë2 P C and all vectors W1 . W2 . We now introduce an inner product on the vector space of all functions from G to C. ö in the vector space. ö. ö. as in this example. 14. and that conditions (a) and (b) give hö. The vector space of all functions from G to C can be equipped with an inner product in a way which we shall describe shortly. ë1 è1 ‡ ë2 W2 i ˆ ë1 hö. W2 i for all ë1 . ö. De®ne 1 ˆ hW.134 Representations and characters of groups 1 W ö 2 1 a i 1 a2 À1 1 This means that W(1) ˆ 2. Wl is always real. öl which satis®es the following conditions: (14. there is associated a complex number kW. W2 . öl ˆ ë1 kW1 .3 De®nition Suppose that W and ö are functions from G to C. This will be of basic importance in our study of characters. The de®nition of an inner product on a vector space over C runs as follows. W1 i ‡ ë2 hö. öl ˆ hö. With every ordered pair of vectors W.

by Proposition 13. 3 3 hè. Let ÷ and ø be characters of G. . 1 ‡ 1 . 1 À 1 . 2 ‡ i . öi ˆ 1(1 . .4 Example As in Example 14. Therefore 1 ˆ h÷.Inner products of characters 135 It is transparent that the conditions of (14. and this is a real numberX jGj gPG (2) h÷. ÷i ˆ ÷( g)ø( g À1 ). 1) ˆ 1(1 ‡ i). (À1)) ˆ 2. èi ˆ 1(2 . with representatives g1 . øi ˆ l ˆ ÷( g i )ø( g i ) iˆ1 jCG ( g i )j X Proof (1) We have ø( g) ˆ ø( gÀ1 ) for all g P G. 3 hö. .2) hold. gl . . l is an inner product on the vector space of functions from G to C. 1) ˆ 1X 3 Inner products of characters We can exploit the fact that characters are constant on conjugacy classes to simplify slightly the calculation of the inner product of two characters. 1 ‡ i . i ‡ (À1) .9(3). øi ˆ hø. 1 ‡ 1 . öi ˆ 1(2 .5 Proposition Assume that G has exactly l conjugacy classes. so k . øi ˆ ÷( g)ø( g À1 )X jGj gPG .1. 14. 14. suppose that G ˆ C3 ˆ ka: a3 ˆ 1l and that W and ö are given by 1 W ö 2 1 a i 1 a2 À1 1 Then hè. 1 ˆ (1) h÷.

3 and Theorem 12. g 2 ˆ (1 2)(3 4). i i by Corollary 12. g 3 ˆ (1 2 3). we also have 1 ˆ h÷. ÷iX jGj gPG Since kø. Since characters are constant on conjugacy classes. with representatives g 1 ˆ 1.) (2) Recall that g G denotes the conjugacy class of G which contains i gi . ˆ ÷( g)ø( g) ˆ j g G j÷( g i )ø( g i )X i gP g G i Now Gˆ l ‘ iˆ1 g G and j g G j ˆ jGjajCG ( g i )j. g 4 ˆ (1 3 2) (see Example 12. øi ˆ l 1 ˆ 1 ˆˆ ÷( g)ø( g) ˆ ÷( g)ø( g) jGj gPG jGj iˆ1 G gP g i ˆ ˆ l ˆ j gGj iˆ1 l ˆ iˆ1 jGj i ÷( g i )ø( g i ) j 1 ÷( g i )ø( g i )X jCG ( g i )j 14.136 Representations and characters of groups Since { gÀ1 : g P G} ˆ G. øi is real. it follows that h÷. in fact. øi is. We shall see in Chapter 18 that there are characters ÷ and ø of A4 which take the following values on the representatives gi : gi |CG ( gi )| ÷ ø g1 12 1 4 g2 4 1 0 g3 3 ù ù2 g4 3 ù2 ù .8. øi. Hence h÷. ÷l ˆ h÷. øi ˆ ÷( g À1 )ø( g) ˆ hø.6 Example The alternating group A4 has four conjugacy classes. (We shall prove later that h÷. an integer.18(1)).

øi ˆ ‡ ‡ ˆ 2X ‡ 12 4 3 3 h÷. øi ˆ 0. say CG ˆ U 1 È X X X È U r . and that every irreducible CG-module is isomorphic to one of the CGmodules U1. we have 1 . ‡ 12 4 3 3 4 . where W1 and W2 are CG-submodules which have no common composition factor.4). Recall from Chapter 10 that the regular CG-module is a direct sum of irreducible CG-submodules. for distinct irreducible characters ÷ and ø of G. øi ˆ 137 We advise you to check also that k÷.12) that the irreducible characters of G form an orthonormal set of vectors in the vector space of functions from G to C. We are now going to pave the way to proving the key fact (Theorem 14. . For example. and then let W2 be the sum of the remaining CG-modules U i . ÷i ˆ 1 and h÷. 4 0 . . ù hø.5. we have h÷. We shall investigate some consequences of writing CG like this. we temporarily adopt the following Hypothesis: 14. 4 1 . . 0 ù 2 . Using part (2) of Proposition 14. ù2 ù2 . . . ù2 ù . therefore. Write 1 ˆ e1 ‡ e2 where e1 P W1 and e2 P W2. and to ®nd the inner products of ÷ and ø with the trivial character (which takes the value 1 on all elements of A4 ). ÷l ˆ 1.Inner products of characters (where ù ˆ e2ðia3 ).7 Hypothesis Let CG ˆ W 1 È W 2 . Ur. that is. we may take W1 to be the sum of those irreducible CG-submodules Ui which are isomorphic to a given irreducible CG-module. 0 ù . Among other results. We ®rst look at the effect of applying the elements e1 and e2 of CG to W1 and W 2 . we shall derive a formula for e1 in terms of the character of W 1 . ù ‡ ‡ ˆ 0. There are several ways of choosing CG-submodules W1 and W2 of CG such that CG ˆ W1 È W2 and W1 and W2 have no common composition factor (see De®nition 10.

We shall calculate the trace of W in two ways. w1 e2 ˆ 0. so every CG-homomorphism from W2 to W1 is zero. we evaluate e1 .138 Representations and characters of groups 14. Then 1 ˆ e1 ˆ ÷( g À1 ) gX jGj gPG Proof Let x P G. we have w1 e1 ˆ w1 .8. and this completes the proof. But W2 and W1 have no common composition factor. w2 P W 2 . 14.3.10. e2 ˆ e2 and e1 e2 ˆ e2 e1 ˆ 0X 1 2 Proof In Proposition 14.9 Corollary For the elements e1 and e2 of CG which appear in Hypothesis 14.7.7. Therefore w1 w2 ˆ 0 for all w1 P W 1 . Proposition Let ÷ be the character of the CG-module W1 which appears in Hypothesis 14. w1 e2 ˆ w2 e1 ˆ 0. we have e2 ˆ e1 . w2 e 2 ˆ w2 X Proof If w1 P W1 then the function w2 3 w1 w2 (w2 P W2 ) is clearly a CG-homomorphism from W2 to W 1 . and w2 ˆ w2 1 ˆ w2 (e1 ‡ e2 ) ˆ w2 e2 . Next. take w1 ˆ e1 and w2 ˆ e2 . j . by Proposition 11. Now w1 ˆ w1 1 ˆ w1 (e1 ‡ e2 ) ˆ w1 e1 . The function W: w 3 we1 x À1 (w P CG) is an endomorphism of CG. Similarly w2 w1 ˆ 0. In particular.8 Proposition For all w1 P W1 and w2 P W2. w2 e1 ˆ 0. j 14.

11 Corollary Let ÷ be the character of the CG-module W1 which appears in Hypothesis 14. we deduce from Proposition 14. w2 W ˆ w2 e1 x À1 ˆ 0X Thus W acts on W1 by w1 3 w1 x À1 and on W2 by w2 3 0.Inner products of characters 139 First. e1 P CG. and of course the endomorphism w2 3 0 of W2 has trace 0.8. the endomorphism w1 3 w1 x À1 of W1 has trace equal to ÷(x À1 ). we see that for all x P G. so e1 ˆ ˆ gPG ëg g for some ë g P C. By the de®nition of the character ÷ of W 1 .7. By Proposition 13.3 of the multiplication in CG. w1 W ˆ w1 e1 x À1 ˆ w1 x À1 . ÷iX ÷( g À1 )÷( g) ˆ jGj2 gPG jGj . ë x ˆ ÷(x À1 )ajGjX Therefore e1 ˆ 1 ˆ ÷( g À1 ) gX jGj gPG j 14. as W: w 3 w gPG ë g gx À1 .20. for w1 P W1 and w2 P W2 we have. ÷i ˆ ÷(1)X Proof Using the de®nition 6. the endomorphism w 3 wgx À1 (w P CG) of CG has trace 0 if g Tˆ x and has trace |G| if g ˆ x. Then h÷. € Hence. we have tr W ˆ ë x jGjX Comparing our two expressions for tr W.10 that the coef®cient of 1 in e2 is 1 ˆ 1 1 h÷. Therefore tr W ˆ ÷(x À1 )X Secondly. in view of Proposition 14.

and obtain hm÷. say CG ˆ U1 È X X X È Ur . as required. 14. and the 1 coef®cient of 1 in e1 is ÷(1)/|G|. Next. Then CG ˆ Y È Z. ÷i ˆ 1. with characters ÷ and ø.9 that CG is a direct sum of irreducible CG-submodules. We now apply Corollary 14. øi ˆ 0X Proof Recall from Theorem 11. Let m ˆ dim U. Hence k÷. we know from Corollary 14. m÷i ˆ m÷(1)X As ÷(1) ˆ dim U ˆ m. where the number of CG-submodules Ui which are isomorphic to U is dim U.12 Theorem Let U and V be non-isomorphic irreducible CG-modules. and no composition factor of X is isomorphic to U. each of which has character ÷. The character of W is m÷. since W is the direct sum of m CG-submodules. j We can now prove the main theorem concerning the inner product k . let X be the sum of the remaining CG-submodules Ui. and de®ne W to be the sum of the m irreducible CG-submodules Ui which are isomorphic to U. this yields h÷. every composition factor of W is isomorphic to U. In particular. let Y be the sum of those CG-submodules Ui of CG which are isomorphic to either U or V. respectively. Then CG ˆ W È X X Moreover. Then h÷. ÷l ˆ ÷(1).11 to the character of W. ÷i ˆ 1. and h÷. . W and X have no common composition factor. and let Z be the sum of the remaining CG-submodules Ui.140 Representations and characters of groups On the other hand.9 that e2 ˆ e1 . l.

÷ i i ˆ h÷ i .5(1). then by Theorem 14. . j Applications of Theorem 14. øi ‡ hø. ÷i)X Now h÷. øi ˆ 1. m÷(1) ‡ nø(1) ˆ hm÷ ‡ nø.11. j. k÷. ä ij is 1 if i ˆ j and is 0 if i Tˆ j). so there are non-negative integers d1. and ÷(1) ˆ m.12. øi ˆ d i for 1 < i < k. and hø. øi ‡ hø. In particular. V k be a complete set of nonisomorphic irreducible CG-modules (see De®nition 11. we have (14X13) h÷ i . By Corollary 14. . øi ˆ d 2 ‡ X X X ‡ d 2 X 1 k Summarizing. and hence k÷.13). . we obtain from this (14X16) hø. where n ˆ dim V. The character of Y is m÷ ‡ nø.Inner products of characters 141 and Y and Z have no common composition factor. there are d i factors Vi X Therefore the character ø of V is given by (14X15) ø ˆ d 1 ÷1 ‡ X X X ‡ d k ÷ k X Using (14. ÷i ‡ n2 hø. we have . where ä ij is the Kronecker delta function (that is.7. . Therefore h÷. V is equal to a direct sum of irreducible CG-submodules. ÷i ˆ 0X By Proposition 14. By Theorem 8. øl ˆ 0. by the part of the theorem which we have already proved. m÷ ‡ nøi ˆ m2 h÷. øl ˆ kø. ÷ j i ˆ ä ij for all i. ø(1) ˆ n. ÷l. .12 Let G be a ®nite group. where for each i. . ÷i ˆ hø. . Now let V be a CG-module. Each of these is isomorphic to some V i . .11). and let V1 . dk such that (14X14) V  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ). ÷ k are all distinct. X X X . If ÷ i is the character of Vi (1 < i < k). this implies that the irreducible characters ÷1 . øi ‡ mn(h÷.

÷ i i hø.7. Moreover. .142 Representations and characters of groups 14. (1 2). If ø is any character of G. hø. . and d2X i 14. .17. ø ˆ ÷1 ‡ ÷3 X (This can of course be checked immediately by comparing the values of ø and ÷1 ‡ ÷3 on each conjugacy class representative. . ÷2 . ÷1 i ˆ 3.6(4) that the irreducible characters of S3  D6 are ÷1 . d i ˆ hø. ÷3 l ˆ 1. we know that ø(1) ˆ 3. ÷ k be the irreducible characters of G. then ø ˆ d 1 ÷1 ‡ X X X ‡ d k ÷ k for some non-negative integers d1. By Example 13. by Proposition 14. ø(1 2) ˆ 1. dk . ø(1 2 3) ˆ 0X Therefore. . taking the following values on the conjugacy class representatives 1. . . ÷3 . øi ˆ k ˆ iˆ1 for 1 < i < k. ÷2 l ˆ 0 and kø.6(2). kø. .) A more substantial calculation along these lines is given in Example 15.17 Theorem Let ÷1 . Thus by Theorem 14. (1 2 3): Now let ø be the character of the 3-dimensional permutation module gi |CS3 ( g i )| ÷1 ÷2 ÷3 1 6 1 1 2 (1 2) 2 1 À1 0 (1 2 3) 3 1 1 À1 for S3 .1 1.5(2).18 Example Recall from Example 13.1 ‡ ‡ 0 ˆ 1X 6 2 Similarly. .

÷l Tˆ 0. It is the converse which is the substantial part of this theorem. 14. Conversely. 14. j We are now in a position to prove the remarkable result that `a CGmodule is determined by its character'. We say that ÷ is a constituent of ø if kø. It is this fact which motivates our study of characters in much of the rest of the book. . The next result is another signi®cant consequence of Theorem 14. V  Vi for some i.12.19 De®nition Suppose that ø is a character of G. 1 ˆ hø. Then by (14. respectively. Proof In Proposition 13. assume that kø. with characters ÷ and ø. Thus. Then V and W are isomorphic if and only if ÷ ˆ ø. for it means that many questions about CG-modules can be answered using character theory. .21 Theorem Suppose that V and W are CG-modules.5 we proved the elementary fact that if V  W then ÷ ˆ ø.17. and so V is irreducible. . Proof If V is irreducible then kø. 14.Inner products of characters 143 We shall see many more applications of the important Theorem 14. øl ˆ 1. It gives us a quick and effective method of determining whether or not a given CG-module is irreducible. Then V is irreducible if and only if kø. øl ˆ 1. the constituents of ø are the irreducible characters ÷ i of G for which the integer di in the expression ø ˆ d1 ÷1 ‡ . We have ø ˆ d 1 ÷1 ‡ X X X ‡ d k ÷ k for some non-negative integers di. øl ˆ 1 by Theorem 14. and by (14.14). and that ÷ is an irreducible character of G. øi ˆ d 2 ‡ X X X ‡ d 2 X 1 k It follows that one of the integers di is 1 and the rest are zero. ‡ dk ÷ k is non-zero.20 Theorem Let V be a CG-module with character ø.16).12.

suppose that ÷ ˆ ø. and let r1 . . r4 be the representations of G over C for which 2 3 2 3 ù 0 ù 0 .144 Representations and characters of groups Thus. d i ˆ hø. ar2 ˆ 0 ù 0 ùÀ1 2 3 2 3 0 1 1 ùÀ1 ar3 ˆ . X X X . ci ˆ h÷.21. r3 . 2. it follows that ci ˆ di for all i.14) that there are non-negative integers ci .22 Example Let G ˆ C3 ˆ ka: a3 ˆ 1l. j 14. di (1 < i < k) such that V  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ) with ci factors Vi for each i.12. Vk be a complete set of non-isomorphic irreducible CG-modules with characters ÷1 . . ÷ i i (1 < i < k)X Since ÷ ˆ ø. r2 . r3 and r4 . ÷ i i. and W  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ) with di factors Vi for each i. Again let V1 . ar4 ˆ À1 À1 0 ù (ù ˆ e2ðia3 ). . We know by (14. the representations r2 and r3 are equivalent. The next theorem is another consequence of Theorem 14. By (14. r2 . The characters ø i of the representations r i (i ˆ 1. ÷ k . . . 4) are 1 ø1 ø2 ø3 ø4 2 2 2 2 a 2ù À1 À1 1‡ù a2 2ù2 À1 À1 1 ‡ ù2 Hence by Theorem 14.16). and hence V  W. ar1 ˆ . but there are no other equivalences among r1 . 3.

÷ k be the irreducible characters of G. ÷ i i ˆ ë i X Therefore ÷1 . By Proposition 11. .5)(3) we see that dim (HomCG (V .24 Theorem Let V and W be CG-modules with characters ÷ and ø.2. 14. .14) that there are non-negative integers ci . W )) ˆ h÷. di (1 < i < k) such that V  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ) with ci factors Vi for each i. respectively. ÷ k are linearly independent. øiX Proof We know from (14.13) we have 0 ˆ hë1 ÷1 ‡ X X X ‡ ë k ÷ k . and W  (V1 È X X X È V1 ) È (V2 È X X X È V2 ) È X X X È (Vk È X X X È Vk ) with di factors Vi for each i. using (11. . . .Inner products of characters 145 14. Proof Assume that ë1 ÷1 ‡ X X X ‡ ë k ÷ k ˆ 0 (ë i P C)X Then for all i. j we have dim (HomCG (Vi . Then dim (HomCG (V . using (14. W )) ˆ k ˆ iˆ1 ci d i X . for any i.23 Theorem Let ÷1 . . Vj )) ˆ ä ij X Hence. Then ÷1 . ÷ k are linearly independent vectors in the vector space of all functions from G to C. . . . . j We now relate inner products of characters to the spaces of CGhomomorphisms which we constructed in Chapter 11. . .

we deduce the stated results just as in the proof of Proposition 14. Decomposing CG-modules It is sometimes of practical importance to be able to decompose a given CG-module into a direct sum of CG-submodules.8. and we now describe a process for doing this.25 Proposition With the above notation. then 2ˆ 3 À1 V ÷( g ) g gPG . Let V be any CG-module. e2 P W 2 . and V is any CG-module.13) implies that h÷. for all v1 P V1 and v2 P V2 we have v1 e1 ˆ v1 . v2 e1 ˆ 0. Since W2 and V1 have no common composition factor. v 2 e2 ˆ v 2 X k ˆ iˆ1 j ci d i X Proof If v1 P V1 then the function w2 3 v1 w2 (w2 P W2 ) is clearly a CG-homomorphism from W2 to V1. ÷ˆ ci ÷ i and ø ˆ di÷i and so (14. We can write V ˆ V1 È V2 . 14.7: CG ˆ W 1 È W 2 . v1 e2 ˆ 0. øi ˆ The result follows. and 1 ˆ e1 ‡ e2 with e1 P W 1 . Once more we adopt Hypothesis 14. where every composition factor of V1 is a composition factor of W1 and every composition factor of V2 is a composition factor of W2. where the CG-modules W1 and W2 have no common composition factor.146 Representations and characters of groups k ˆ iˆ1 k ˆ iˆ1 On the other hand.26 Proposition If ÷ is an irreducible character of G. j 14.

(3) Then V is the direct sum of the CG-modules V÷ as ÷ runs over the irreducible characters of G. by Theorem 11.7. . a direct sum of irreducible CG-submodules Ui. Let W1 be the sum of those CG-submodules Ui which have character ÷. 14. and by Proposition 14. The character of V÷ is a multiple of ÷. . and let W2 be the sum of the remaining CG-submodules Ui. Proposition 14. . Taking ÷ to be the trivial character of G in Proposition 14.26.28 Examples (1) Let G be any ®nite group and let V be any non-zero CG-module.10.9. Also W1 and W2 satisfy Hypothesis 14. the element e1 of W1 is given by m ˆ e1 ˆ ÷( g À1 ) gX jGj gPG Let V1 be the sum of those CG-submodules of V which have character ÷.Inner products of characters 147 is equal to the sum of those CG-submodules of V which have character ÷ (where for r P CG. . Then Proposition 14. Proof Write CG ˆ U 1 È X X X È U r . We illustrate this method with a couple of simple examples. Clearly we may omit the constant multiplier m/|G|.26 provides a useful practical tool for ®nding CG-submodules of a given CG-module V. we see that . Then the character of W1 is m÷ where m ˆ ÷(1). we de®ne Vr ˆ fvr: v P V g). so 2ˆ 3 À1 V1 ˆ V ÷( g ) g X gPG j Once the irreducible characters of our group G are known. The procedure is as follows: (14. Some more complicated uses of the method can be found in Chapter 32.27) (1) Choose a basis v1 . calculate the vectors € v i ( gPG ÷( gÀ1 ) g) for 1 < i < n. and let V÷ be the subspace of V spanned by these vectors. (2) For each irreducible character ÷ of G. v n of V.25 shows that Ve1 ˆ V1.

. Ve5 ˆ sp (v1 À v3 . v4 such that v i g ˆ v ig for all i and all g P G. v3 .148 Representations and characters of groups 2ˆ 3 V g gPG is the sum of all the trivial CG-submodules of V. . let G ˆ Sn and let V be the permutation module. with basis v1 .5). Then 2ˆ 3 V g ˆ sp (v1 ‡ X X X ‡ v n )X gPG Hence V has a unique trivial CG-submodule. v n such that v i g ˆ v ig for all i and all g P G. . let ÷ i (1) ˆ ei ˆ ÷ i ( g À1 ) gX 8 gPG For example. For example. v2 À v4 )X . ÷5 of D8 (see Example 16. Then 2 Ve1 ˆ sp (v1 ‡ v2 ‡ v3 ‡ v4 ). Ve2 ˆ 0. (2) Let G be the subgroup of S4 which is generated by a ˆ (1 2 3 4) and b ˆ (1 2)(3 4)X Then G  D8 (compare Example 1. . e5 ˆ 1(1 À a2 ). . . Ve4 ˆ sp (v1 À v2 ‡ v3 À v4 ). v2 . .3(3)): 1 ÷1 ÷2 ÷3 ÷4 ÷5 1 1 1 1 2 a 1 1 À1 À1 0 a2 1 1 1 1 À2 a3 1 1 À1 À1 0 b 1 À1 1 À1 0 ab 1 À1 À1 1 0 a2 b 1 À1 1 À1 0 a3 b 1 À1 À1 1 0 Let V be the permutation module for G. For 1 < i < 5. with basis v1 . Here is a list of the irreducible characters ÷1 . . Ve3 ˆ 0.

j. øl ˆ 1. e2 ˆ ei for 1 < i < 5. You might like to check that e1 ‡ X X X ‡ e5 ˆ 1. . The inner product of two functions W. ÷ k of G form an orthonormal set. Note that the procedure described in (14. h÷ i . . . The irreducible characters ÷1 . . Every CG-module is determined by its character. respectively.9. . then ø ˆ d 1 ÷1 ‡ X X X ‡ d k ÷ k where d i ˆ hø. ÷ j i ˆ ä ij for all i. ÷4 and ÷5 . and ø is any character.27) does not in general enable us to write a given CG-module as a direct sum of irreducible CG-submodules (since V÷ is not in general irreducible). 4. . öi ˆ W( g)ö( g)X jGj gPG 2.Inner products of characters We have V ˆ Ve1 È Ve4 È Ve5 . 149 and so we have expressed V as a direct sum of irreducible CGsubmodules whose characters are ÷1 . that is. . i ei ej ˆ 0 for i Tˆ jX Compare these results with Corollary 14. If ÷1 . ÷ k are the irreducible characters of G. . . ö from G to C is given by 1 ˆ hW. 3. ø is irreducible if and only if kø. Also. Summary of Chapter 14 1. ÷ i iX Each di is a non-negative integer.

show that h÷reg . i 0 1 0 2 3 2 3 À1 0 1 0 ar3 ˆ . b2 ˆ a2 . r3 be the representations of G over C for which 2 3 2 3 i 0 0 1 ar1 ˆ . br2 ˆ . Let G ˆ S4 . r2 . Let G ˆ Q8 ˆ ka. bÀ1 ab ˆ aÀ1 l. gó ˆ T À1 ( gr)TX 4.150 Representations and characters of groups Exercises for Chapter 14 1. br1 ˆ . and let r1 . non-trivial character of G. Prove that ÷ is reducible. and that for each g in G there is an invertible matrix Tg such that gó ˆ T À1 ( gr)Tg X g Prove that there is an invertible matrix T such that for all g in G. b: a4 ˆ 1. Which of ÷ and ø is irreducible? 2. If ÷ is a character of G. øi and hø. br3 ˆ X 0 1 0 À1 Show that r1 and r2 are equivalent. 3. and that ÷( g) is a non-negative real number for all g in G. We shall see in Chapter 18 that G has characters ÷ and ø which take the following values on the conjugacy classes: Class representative |CG ( gi )| ÷ ø 1 24 3 3 (1 2) 4 À1 1 (1 2 3) 3 0 0 (1 2)(3 4) 8 3 À1 (1 2 3 4) 4 À1 À1 Calculate h÷. 0 Ài À1 0 2 3 2 3 0 i 0 À1 ar2 ˆ . ÷i ˆ ÷(1)X . 5. ÷i. h÷. øi. but r3 is not equivalent to r1 or r2. Suppose that r and ó are representations of G. Suppose that ÷ is a non-zero.

What can you say about the integers di in the cases kø. .4 relevant. øl ˆ 1. If ð is the permutation character of Sn . Let ÷1 .) 151 7. Suppose that ÷ is a character of G and that for every g P G. and suppose that ø ˆ d 1 ÷1 ‡ X X Xd k ÷ k is a character of G. Does it follow that ÷ ˆ 2ö for some character ö? . 2. ÷ k be the irreducible characters of the group G.Inner products of characters 6. . 1 S n i ˆ 1X (Hint: you may ®nd Exercise 11. 3 or 4? 8. . ÷( g) is an even integer. . prove that hð.

1 De®nition A class function on G is a function ø: G 3 C such that ø(x) ˆ ø( y) whenever x and y are conjugate elements of G (that is. Thus. Throughout. the characters of G are class functions on G.5(2).3 Theorem The number of irreducible characters of G is equal to the number of conjugacy classes of G. The set C of all class functions on G is a subspace of the vector space of all functions from G to C. G is as usual a ®nite group. and to some consequences of this theorem. then (15X2) dim C ˆ lX 15. Together with the material from Chapter 14. if l is the number of conjugacy classes of G.15 The number of irreducible characters We devote this chapter to the theorem which states that the number of irreducible characters of a ®nite group is equal to the number of conjugacy classes of the group. A basis of C is given by those functions which take the value 1 on precisely one conjugacy class and zero on all other classes. By Proposition 13. ø is constant on conjugacy classes). 152 . the theorem provides machinery for investigating characters which is used in the remainder of the book. Class functions 15.

and in particular. ÷ k be the irreducible characters of G. . In order to prove the reverse inequality l < k.4 Corollary The irreducible characters ÷1 . the centre of CG. . we consider the regular CG-module. if ø is a class function.The number of irreducible characters 153 Proof Let ÷1 .22. f k . . . Proof Since ÷1 . X X X . ÷ i l for 1 < i < k. which is equal to k by k ˆ iˆ1 ëi÷i . Since CG contains the identity element 1. vz ˆ ë i vX Hence wz ˆ ë i w for all w P W i . This completes the proof that k ˆ l. and let l be the number of conjugacy classes of G. Indeed. . they span a subspace of C of dimension k. for each i there exists ë i P C such that for all v P Vi. then øˆ where ë i ˆ kø. . dim C ˆ l. .2) implies that k < l. By Proposition 9.2). . we deduce that l < k. where for each i. . By Theorem 14. f iz ˆ ëi f i It follows that z ˆ 1z ˆ ( f 1 ‡ X X X ‡ f k )z ˆ f 1 z ‡ X X X ‡ f k z ˆ ë1 f 1 ‡ X X X ‡ ë k f k X This shows that Z(CG) is contained in the subspace of CG spanned by f 1 . ÷ k are linearly independent. ÷ k of G form a basis of the vector space of all class functions on G.23.7 that CG ˆ W 1 È X X X È W k . .14. If V1 . W i is isomorphic to a direct sum of copies of V i . X X X . Since Z(CG) has dimension l by Proposition 12. V k is a complete set of non-isomorphic irreducible CG-modules. . we know from Theorem 8. so (15. ÷1 . By (15. j (1 < i < k)X 15. ÷ k are linearly independent elements of C. we can write 1 ˆ f1 ‡ X X X ‡ fk with f i P W i for 1 < i < k. Now let z P Z(CG). X X X . .

. 15. j Corollary 15.5(2). . ÷ k span C. The last part follows. and six irreducible characters ÷1 . Proof Since ÷( g) is real if and only if ÷( g) ˆ ÷( gÀ1 ) (see Proposition j 13.6 Corollary Suppose that g P G. We conclude the chapter with an example illustrating some practical methods of expressing characters and class functions of a group as combinations of irreducible characters. h P G. this is true for the class function ø which takes the value 1 on the conjugacy class of g and takes the value 0 elsewhere. As in previous examples.4. . suppose that ÷( g) ˆ ÷(h) for all characters ÷. j 15. Hence ÷1 . and so g is conjugate to h. . .3. .154 Representations and characters of groups Theorem 15. ÷6 given as follows: . ø( g) ˆ ø(h) for all class functions ø on G. whose k entries are the values of ÷ on the k conjugacy classes of G. Then g is conjugate to h if and only if ÷( g) ˆ ÷(h) for all characters ÷ of G. . Then g is conjugate to gÀ1 if and only if ÷( g) is real for all characters ÷ of G. Then by Corollary 15. g6 (where g1 ˆ 1). using (14. In particular. by Proposition 13.5 Proposition Suppose that g. . Conversely.5.4 has the following useful consequence.13). and so they form a basis of C. . . Then ø( g) ˆ ø(h) ˆ 1. 15. .9(3)). Proof If g is conjugate to h then ÷( g) ˆ ÷(h) for all characters ÷ of G. the result follows immediately from Proposition 15.4 that there is a certain group G of order 12 which has exactly six conjugacy classes with representatives g1 . we regard a character ÷ of G as a row vector. .7 Example We shall see in Section 18.

and the entries in the column corresponding to g1 ˆ 1 are positive integers (indeed. it is not hard to use tactical guesswork to express ö as a combination of the irreducible characters. we see that ÷ must be a combination of ÷2 . . The correct answer now comes quickly to mind. given any character ö of G whose degree is not large compared with the degrees of the ÷ i . ÷6 : g1 ë ì 2 4 g2 À2 4 g3 À2 1 g4 2 1 g5 0 0 g6 0 0 . the second entry in the row vector for ÷ is equal to minus the ®rst entry. they are the degrees of the ÷ i ). The reason for this is that the required coef®cients are known to be non-negative integers. . ø ˆ ÷1 ‡ ÷2 ‡ ÷3 ‡ ÷4 X For example.The number of irreducible characters gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g1 12 1 1 1 1 2 2 g2 12 1 À1 1 À1 2 À2 g3 6 1 À1 1 À1 À1 1 g4 6 1 1 1 1 À1 À1 g5 4 1 i À1 Ài 0 0 155 g6 4 1 Ài À1 i 0 0 Suppose we are given characters ÷ and ø of G as follows: g1 ÷ ø 3 4 g2 À3 0 g3 0 0 g4 0 4 g5 i 0 g6 Ài 0 Then it is easy to spot that ÷ ˆ ÷2 ‡ ÷6 . In fact. ì of G as combinations of ÷1 . We suggest that you use the `guesswork method' to express the following characters ë. . ÷4 and ÷6 . Inspecting the values of the irreducible characters ÷ i . .

1 (À1 ‡ 2i) . 1 (À1 ‡ 2i) . ÷6 l ˆ 0. ÷3 i ˆ ‡ ‡ ‡ ‡ 12 12 6 6 4 (À1 À 2i) . ÷4 l ˆ 1. 1 3 . 4 hö. i ˆ 3. 1 ‡ ‡ ‡ ‡ 12 12 6 6 4 (À1 À 2i) . Therefore ö ˆ ÷1 ‡ 3÷2 ‡ 2÷3 ‡ ÷4 ‡ 2÷5 X Summary of Chapter 15 1. (À1) ‡ ˆ 2. 1 3 . 1 5 . 1 ‡ ˆ 1. like the following one? g1 ö 11 g2 3 g3 À3 g4 5 g5 À1 ‡ 2i g6 À1 À 2i The answer is to use the inner product k . (À1) hö. ‡ 4 11 . 1 3 . 1 (À1 ‡ 2i) . We know from Corollary 15. kö. ÷1 i ˆ and similarly kö. (À1) 5 . The number of irreducible characters of a group is equal to the number of conjugacy classes of the group. 1 À3 . (Ài) hö.156 Representations and characters of groups How do we cope with a class function or with a more dif®cult character. ÷5 l ˆ 2 and kö.4 that the coef®cients ë i in the expression ö ˆ ë1 ÷1 ‡ X X X ‡ ë6 ÷6 are given by ë i ˆ hö. ÷ i i (1 < i < 6)X Using Proposition 14. 1 5 . 4 11 . (À1) (À3) . we calculate these inner products: 11 . 1 À3 . l. . ÷2 i ˆ ‡ ‡ ‡ ‡ 12 12 6 6 4 (À1 À 2i) .5(2).

÷2 . ø2 and ø3 be the class functions on S3 taking the following values: 1 ø1 ø2 ø3 1 0 0 (1 2) 0 1 0 (1 2 3) 0 0 1 Express ø1 . Let ø1 . ÷ k of G form a basis of the vector space of all class functions on G. The three irreducible characters of S3 are ÷1 . . Is ÷ a character of S3 ? 2. ÷2 and ÷3 . then øˆ k ˆ iˆ1 ë i ÷ i where ë i ˆ hø. The irreducible characters ÷1 . ÷2 and ÷3 of S3 . .The number of irreducible characters 157 2. If ø is a class function. . ÷ i iX Exercises for Chapter 15 1. . ÷3 : 1 ÷1 ÷2 ÷3 1 1 2 (1 2) 1 À1 0 (1 2 3) 1 1 À1 Let ÷ be the class function on S3 with the following values: 1 ÷ 19 (1 2) À1 (1 2 3) À2 Express ÷ as a linear combination of ÷1 . . ø2 and ø3 as linear combinations of the irreducible characters ÷1 .

prove that G has 4. . Is ø a character of G? 4.7.2. Suppose that G is the group of order 12 in Example 15.) (b) Using the solution to Exercise 11. 6 or 12 conjugacy classes. Find groups G in which each of these possibilities is realized. . . . . ÷6 . g6 and irreducible characters ÷1 . Let G be a group of order 12. . ÷6 as in that example. . . . . . (a) Show that G cannot have exactly 9 conjugacy classes. with conjugacy class representatives g1 . . Let ø be the class function on G taking the following values: g1 ø 6 g2 0 g3 3 g4 À3 g5 À1 À i g6 À1 ‡ i Express ø as a linear combination of ÷1 . (Hint: show that Z(G) cannot have order 6.158 Representations and characters of groups 3. .

X X X . Much of the later material in the book will be devoted to understanding character tables.4). the identity element of G. Note that in the character table. Character tables 16. X X X . The motivation for this is Theorem 14. the rows are indexed by the irreducible characters of G and the columns are indexed by the conjugacy classes (or. This matrix is called the character table of G. and g1 ˆ 1. the numbering is arbitrary. The entries in a character table are related to each other in subtle ways. Beyond this. 1 < j < k) is called the character table of G. ÷ k be the irreducible characters of G and let g 1 . which tells us that every CG-module is determined by its character. by conjugacy class representatives). j with 1 < i < k. Thus. many problems in representation theory can be solved by considering characters.1 De®nition Let ÷1 . g k be representatives of the conjugacy classes of G. It is usual to number the irreducible characters and conjugacy classes of G so that ÷1 ˆ 1 G . It is therefore convenient to record all the values of all the irreducible characters of G in a square matrix. and the number of them is equal to the number of conjugacy classes of G.16 Character tables and orthogonality relations The irreducible characters of a ®nite group G are class functions.21. in practice. The k 3 k matrix whose ij-entry is ÷ i ( g j ) (for all i. 159 . many of which are encapsulated in the orthogonality relations (Theorem 16. the trivial character.

j 16. are linearly independent (Theorem 14. b as representatives of the conjugacy classes of G.6(4).4. and then the character table of G is 1 ÷1 ÷2 ÷3 1 1 2 a 1 1 À1 b 1 À1 0 (2) We can write down the character table of any ®nite abelian group using Theorem 9.2 Proposition The character table of G is an invertible matrix. and hence also the rows of the character table.23).3 Examples (1) Let G ˆ D6 ˆ ka. b: a3 ˆ b2 ˆ 1. For example. b: a4 ˆ b2 ˆ 1. We take 1. You found all the irreducible representations of G in Exercise 10. bÀ1 ab ˆ aÀ1 l. The conjugacy classes .8. the character table of C2 ˆ ha: a2 ˆ 1i is 1 ÷1 ÷2 1 1 a 1 À1 and the character table of C3 ˆ ka: a3 ˆ 1l is 1 ÷1 ÷2 ÷3 1 1 1 a 1 ù ù2 a2 1 ù2 ù (ù ˆ e2ðia3 ) (3) Let G ˆ D8 ˆ ka. bÀ1 ab ˆ aÀ1 l. a. The irreducible characters of G are given in Example 13. Proof This follows immediately from the fact that the irreducible characters of G.160 Representations and characters of groups 16.

. a2 . X X X . These relations can be expressed in terms of the rows of the character table.12). s P {1. g k be representatives of the conjugacy classes of G. ÷ s i ˆ ä rs . and let g 1 . k}. Hence the character table of G is 1 ÷1 ÷2 ÷3 ÷4 ÷5 1 1 1 1 2 a2 1 1 1 1 À2 a 1 1 À1 À1 0 b 1 À1 1 À1 0 ab 1 À1 À1 1 0 The character tables of all dihedral groups will be found in Chapter 18. 16. . a.13). Then the following relations hold for any r. . ab. ÷ k be the irreducible characters of G. .4 Theorem Let ÷1 . . . . The orthogonality relations We have already seen many uses for the relations (14. h÷ r . and these are given by part (2) of our next result. Similar relations exist between the columns of the character table. . by writing them as k ˆ ÷ r ( g i )÷ s ( g i ) ˆ ä rs jCG ( g i )j iˆ1 (see Proposition 14.Character tables and orthogonality relations 161 of G are given by (12. b. . among the irreducible characters ÷1 . ÷ k of G. . and representatives are 1.5(2)). . . (1) The row orthogonality relations: k ˆ ÷ r ( g i )÷ s ( g i ) iˆ1 jCG ( g i )j ˆ ä rs X (2) The column orthogonality relations: k ˆ iˆ1 ÷ i ( g r )÷ i ( g s ) ˆ ä rs jCG ( g r )jX .

so ë i ˆ hø s . and ø s ( g) ˆ 0 otherwise. ÷ j i ˆ ä ij . and this time we record the order of the centralizer C G ( g i ) next to each conjugacy class representative gi : gi |CG ( g i )| ÷1 ÷2 ÷3 1 6 1 1 2 a 3 1 1 À1 b 2 1 À1 0 .4. . . say øs ˆ ëi ÷i (ë i P C)X We know that h÷ i . also there are jGjajC G ( g s )j elements of G which are conjugate to g s . by Theorem 12.8. (1) Let G ˆ D6. let ø s be the class function which satis®es ø s ( g r ) ˆ ä rs k ˆ iˆ1 (1 < r < k)X By Corollary 15. ø s is a linear combination of ÷1 . .5 Examples We illustrate the column orthogonality relations. ÷ k . They are recorded here merely for comparison with the column relations. j and the column orthogonality relations follow.162 Representations and characters of groups Proof The row orthogonality relations have already been proved.3(1). For 1 < s < k. ÷ i i ˆ 1 ˆ ø s ( g)÷ i ( g)X jGj gPG Now ø s ( g) ˆ 1 if g is conjugate to g s . . 16. Hence ëi ˆ Therefore ä rs ˆ ø s ( g r ) ˆ k ˆ iˆ1 1 ˆ ÷ i ( gs ) ø s ( g)÷ i ( g) ˆ X jGj gP g G jCG ( g s )j s ëi ÷i ( gr ) ˆ k ˆ ÷ i ( g r )÷ i ( g s ) iˆ1 jCG ( g s )j . We copy the character table of G from Example 16.

We shall use the column orthogonality relations to determine the last row of the character table. 1 ‡ 1 . 1 ‡ (À1) . 1 ‡ 1 . 1 ‡ 1 . The entries in the ®rst column of the character table are the degrees of the irreducible characters. By considering the orthogonality relations between the ®rst column and columns 3 and 4. s ˆ 2: s ˆ 2: s ˆ 3: 1 . we obtain the complete character table as . By the column orthogonality relations with r ˆ s ˆ 1.Character tables and orthogonality relations € Consider the sums 3 ÷ i ( g r )÷ i ( g s ) for various cases: iˆ1 r ˆ 1. the sum of the squares of these numbers is 12 (this also follows from Theorem 11. 1 .12). Let x denote the number at the foot of the second column. and is the number at the top of the column (that is. (À1) ‡ 2 . r ˆ 1. we read down columns r and s of the character table. 0 ˆ 0X 163 In each case. 1 ‡ 1 . (À1) ˆ 0. 1 ‡ 1 . The sum of the products is 0 if r Tˆ s. 1 . The column orthogonality relation 4 ˆ iˆ1 ÷ i ( g 1 )÷ i ( g 2 ) ˆ 0 gives 1 . r ˆ 2. 1 ‡ 3x ˆ 0X Therefore x ˆ À1. (À1) ˆ 3. Hence the last entry in the ®rst column is 3. so they are positive integers. the order of the centralizer of g r ) if r ˆ s. 1 ‡ 2 . taking the products of the numbers which appear. (2) Suppose we are given the following part of the character table of a group G of order 12 which has exactly four conjugacy classes: gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 g1 12 1 1 1 g2 4 1 1 1 g3 3 1 ù ù2 g4 3 1 ù2 ù (where ù ˆ e2ðia3 ).

if g ˆ 1. if g ˆ 1. 0 ˆ 0X We shall see later that the character table which we have constructed here is that of A4 . 1 ‡ 1 . . if g Tˆ 1. k ˆ ÷ i (1)÷ i ( g) ˆ X iˆ1 0. although our calculation has used only those relations which involve the ®rst column. we get V ` jGj. ù ‡ 0 . (À1) ˆ 4. since Theorem 13. By taking the complex conjugate of each side of this equation.164 follows: gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 Representations and characters of groups g1 12 1 1 1 3 g2 4 1 1 1 À1 g3 3 1 ù ù2 0 g4 3 1 ù2 ù 0 Notice that the orthogonality relations hold between all pairs of columns. ÷ i ( g 3 )÷ i ( g 3 ) ˆ 1 . ˆ d i ÷ i ( g) ˆ X iˆ1 0. ù2 ‡ ù2 . where d i ˆ ÷ i (1). 1 ‡ ù . if g Tˆ 1. 1 ‡ 1 . 4 ˆ iˆ1 4 ˆ iˆ1 4 ˆ iˆ1 ÷ i ( g 2 )÷ i ( g 2 ) ˆ 1 . ù ‡ ù2 . ÷ i ( g 3 )÷ i ( g 4 ) ˆ 1 .19 and Proposition 13. 1 ‡ ù . For example. 1 ‡ (À1) .20 give V k ` jGj. ù2 ‡ 0 . 0 ˆ 3. Those column orthogonality relations which involve the ®rst column of the character table were proved in Chapter 13.

Rows versus columns Notice that in Example 16. Indeed. by letting the ij-entry of M be ÷ i ( gj ) X jCG ( g j )j1a2 Let M t denote the transpose of the complex conjugate of M. More importantly. the equation M M t ˆ I is just another way of expressing the row orthogonality relations. we calculated the values of the last character one at a time using the column orthogonality relations. We could have used the above argument to deduce the column orthogonality relations from the row ones. so M M t ˆ I. On the other hand. ÷4 i ˆ ä i4 to obtain four equations in the four unknown values ÷4 ( g j ) (1 < j < 4). it is a fact that the column orthogonality relations contain precisely the same information as the row orthogonality relations. Now the rs-entry in M M t is k ˆ ÷ r ( g i )÷ s ( g i ) iˆ1 jCG ( g i )j ˆ ä rs . Since the properties M t M ˆ I and M M t ˆ I of a square matrix M are equivalent to each other. where we were given three of the four irreducible characters of G. as we shall now show. we see that the row and column orthogonality relations are equivalent. and we adjust the entries ÷ i ( g j ) in this matrix to obtain another k 3 k matrix M. jCG ( g r )j1a2 jCG ( g s )j1a2 iˆ1 by the column orthogonality relations. the row and column orthogonality relations encapsulate the same information. so M t M ˆ I. Although the calculation with the column orthogonality relations was easier to perform.5(2). . the rs-entry in M t M is k ˆ 1 ÷ i ( g r )÷ i ( g s ) ˆ ä rs . An alternative approach would have been to use the row orthogonality relations h÷ i .Character tables and orthogonality relations 165 and these are just the column orthogonality relations which involve the ®rst column. by the row orthogonality relations. The character table of G is a k 3 k matrix.

with representatives g1 . . ÷2 as follows: . The row orthogonality relations state that for all r. . . . . gk . we can deduce exactly the same results using either set of relations. . . . ÷4 taking the following values: gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 g1 8 1 1 1 1 g2 8 1 1 1 1 g3 4 1 1 À1 À1 g4 4 1 À1 1 À1 g5 4 1 À1 À1 1 Find the complete character table of G. and four linear characters ÷1 . .166 Representations and characters of groups so when we are working with character tables. k ˆ ÷ r ( g i )÷ s ( g i ) iˆ1 jCG ( g i )j ˆ ä rs X 3. . and has irreducible characters ÷1 . g5 . . . Summary of Chapter 16 Let G be a ®nite group with irreducible characters ÷1 . . 1. . There exists a group G of order 10 which has precisely four conjugacy classes. Write down the character table of C2 3 C2 . 2. . . . . g4 . A certain group G of order 8 is known to have a total of ®ve conjugacy classes. The column orthogonality relations state that for all r. with representatives g1 . The character table of G is the k 3 k matrix with ij-entry ÷ i ( g j ). s. k ˆ iˆ1 ÷ i ( g r )÷ i ( g s ) ˆ ä rs jCG ( g r )jX Exercises for Chapter 16 1. 2. ÷ k and conjugacy class representatives g1 . s. . 3. .

(Hint: ®rst ®nd the values of the remaining irreducible characters on g1 . k ‰ iˆ1 jCG ( g i )jX . (a) Find æ. . ÷ k be the irreducible characters of G. . 5.) 4. Show that det C is either real or purely imaginary. . Show that @ A k ˆ Z(G) ˆ g P G: ÷ i ( g)÷ i ( g) ˆ jGj X iˆ1 6. (b) Find another column of the character table. Find the complete character table of G.Character tables and orthogonality relations gi |CG ( g i )| ÷1 ÷2 g1 10 1 2 g2 5 1 á g3 5 1 â g4 2 1 0 167 p p where á ˆ (À1 ‡ 5)a2 and ⠈ (À1 À 5)a2. then on g4 ± use Corollary 13. A certain group G has two columns of its character table as follows: gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 ÷5 g1 21 1 1 1 3 3 g2 7 1 1 1 æ æ where g1 ˆ 1 and æ P C. g k and character table C. X X X . and that jdet Cj2 ˆ Find Æ(det C) when G ˆ C3 . Let ÷1 . .10. Let G be a ®nite group with conjugacy class representatives g1 .

Lifted characters We begin by constructing a character of G from a character of GaN . 17. In fact. De®ne ÷ ÷: G 3 C by ÷( g) ˆ ~(Ng) ( g P G)X ÷ Then ÷ is a character of G. by a process which is known as lifting. The linear characters of G (i. C) be a representation of GaN with 168 .e. Thus. In the opposite direction. in particular. we can use the characters of GaN to get some of the characters of G. and ÷ and ~ have the same degree. it is easy to tell from the character table whether or not G is simple. it is also true that the character table of G enables us to ®nd the normal subgroups of G. then the factor group GaN is smaller than G.7.17 Normal subgroups and lifted characters If N is a normal subgroup of the ®nite group G. can be used to get new irreducible characters from a given irreducible character. and N Tˆ {1}.1 Proposition Assume that N v G. ÷ ~ Proof Let r: GaN 3 GL (n. and let ~ be a character of GaN . in a way which we shall describe. normal subgroups help us to ®nd characters of G.) The linear characters. (The derived subgroup is de®ned below in De®nition 17. in turn. The characters of GaN should therefore be easier to ®nd than the characters of G. the characters of degree 1) are obtained by lifting the irreducible characters of GaN in the case where N is the derived subgroup of G.

Also. and hence 2 2 ~ g1 r ˆ g2 r. j 17. ÷ 17. h P G we have ((Ng)(Nh))~ ˆ (Ngh)~ ˆ ( gh)r ˆ ( gr)(hr) r r ˆ ((Ng)~)((Nh)~). C) is a representation of G with character ÷. Now let ÷ be a character of G with N < Ker ÷. Irreducible characters of GaN correspond to irreducible characters of G which have N in their kernel.3 Theorem Assume that N v G. then the character ÷ of G ÷ which is given by ÷( g) ˆ ~(Ng) ÷ is called the lift of ~ to G. Thus r is a representation of G. By associating each character of GaN with its lift to G. C) which is given by the ÷ composition g 3 Ng 3 (Ng)~ r ( g P G) is a homomorphism from G to GL (n. g2 P G and Ng1 ˆ Ng2 then g1 gÀ1 P N. so ÷ and ~ have the same ÷ ÷ degree. then ÷ ~(N) ˆ ÷(1). Moreover. so ( g1 gÀ1 )r ˆ I.Normal subgroups and lifted characters 169 character ~. and ÷ is the lift of ~ to G. ÷ ÷ so N < Ker ÷. C) by (Ng)~ ˆ gr r Then for all g.2 De®nition If N v G and ~ is a character of GaN . if k P N then ÷ ÷(k) ˆ ~(Nk) ˆ ~(N ) ˆ ÷(1). The function r: G 3 GL (n. ~ Proof If ÷ is a character of GaN . If g 1 . The character ÷ of r satis®es ÷( g) ˆ tr ( gr) ˆ tr ((Ng)~) ˆ ~(Ng) r ÷ for all g P G. ÷(1) ˆ ~(N). we obtain a bijective correspondence between the set of characters of GaN and the set of characters ÷ of G which satisfy N < Ker ÷. C). Suppose that r: G 3 GL (n. We may therefore de®ne a function r: GaN 3 GL (n. r r ( g P G)X ( g P G) .

(1 3)(2 4). If we put a ˆ N(1 2 3) and b ˆ N(1 2) then GaN ˆ ha. We know from Example 16.20). then Theorem 17. If ~ is the character of r then ÷ ~(Ng) ˆ ÷( g) ( g P G)X ÷ ~ Thus ÷ is the lift of ÷. It remains to show that irreducible characters correspond to irreducible characters.3 enables us to write down as many irreducible characters of G as there are irreducible characters of GaN .4 Example Let G ˆ S4 and N ˆ V4 ˆ f1. and note that u( gr) P U for all u P U D u(Ng)~ P U for all u P U X r Thus. ÷ j If we know the character table of GaN for some normal subgroup N of G. bi and a3 ˆ b2 ˆ N . (1 2)(3 4). To see this.3(1) that the character table of GaN is N ~1 ÷ ~2 ÷ ~3 ÷ 1 1 2 N (1 2) 1 À1 0 N(1 2 3) 1 1 À1 . Hence ÷ is irreducible if and only if ~ is irreducible. so GaN  D6 . bÀ1 ab ˆ aÀ1 . (1 4)(2 3)g. 17. U is a CG-submodule of C n if and only if U is a C(GaN )submodule of C n .170 Representations and characters of groups ~ ~ so r is a representation of GaN . We have now established that the function which sends each character of GaN to its lift to G is a bijection between the set of characters of GaN and the set of characters of G which have N in their kernel. let U be a subspace of C n . The representation r is therefore irreducible if and ~ only if the representation r is irreducible. so that N v G (see Example 12.

we note that ÷ ÷((1 2)(3 4)) ˆ ~(N ) since (1 2)(3 4) P N . once the character table of G is known. ÷2 . which are given by ÷ ÷ ÷ 1 ÷1 ÷2 ÷3 1 1 2 (1 2) 1 À1 0 (1 2 3) 1 1 À1 (1 2)(3 4) 1 1 2 171 (1 2 3 4) 1 À1 0 Then ÷1 . . ~3 are ÷1 . as our next two propositions will demonstrate. ~2 . since ~1 . so g ˆ 1 by Proposition 15. Finding normal subgroups The character table contains accessible information about the structure of a group. ÷ s of G such that Nˆ s ’ iˆ1 Ker ÷ i X Proof If g belongs to the kernel of each irreducible character of G. Recall that we can easily locate the kernel of an irreducible character ÷ from the character table. Of course.Normal subgroups and lifted characters To calculate the lift ÷ of a character ~ of GaN . First we shall show how to ®nd all the normal subgroups of G. 17.5. then ÷( g) ˆ ÷(1) for all characters ÷. ÷3 .5 Proposition If N v G then there exist irreducible characters ÷1 . ~2 . Hence the intersection of the kernels of all the irreducible characters of G is {1}. The following proposition shows that every normal subgroup arises in this way. since Ker ÷ ˆ f g P G: ÷( g) ˆ ÷(1)g (see De®nition 13. ÷3 are irreducible characters of G. X X X . any subgroup which is the intersection of the kernels of irreducible characters is a normal subgroup too. Also Ker ÷ v G. ÷ ÷((1 2 3 4)) ˆ ~(N (1 3)) since N (1 2 3 4) ˆ N (1 3)X ÷ Hence the lifts of ~1 . ÷2 .13). ~3 are ÷ ÷ ÷ irreducible characters of GaN .

6 Proposition The group G is not simple if and only if ÷( g) ˆ ÷(1) for some non-trivial irreducible character ÷ of G. s ’ iˆ1 Ker ~ i ˆ fN gX ÷ For 1 < i < s. Hence Nˆ s ’ iˆ1 Ker ÷ i X j It is particularly easy to tell from the character table of G whether or not G is simple: 17. we have ÷( g) ˆ ÷(1). Then g P Ker ÷. . Thus Ker ÷ is a normal subgroup of G which is not equal to {1} of G. and so g P N. ÷ is non-trivial. If r is a representation of G with character ÷. so that there is a normal subgroup N of G with N Tˆ {1} and N Tˆ G. Then by Proposition 17. so Ker ÷ Tˆ {1}.11(2). Conversely. By the ÷ ÷ above observation. If g P Ker ÷ i then ÷ ~ i (N ) ˆ ÷ i (1) ˆ ÷ i ( g) ˆ ~ i (Ng). . Therefore if g P Ker ÷ i then Ng P Ker ~ i ˆ ÷ ÷ {N}. . ÷ ÷ „ „ and so Ng P Ker ~ i. We shall show how to ®nd all linear characters of any group G. ~ s be the irreducible characters of GaN . since the . j Linear characters Recall that a linear character of a group is a character of degree 1. and taking 1 Tˆ g P Ker ÷. and some nonidentity element g of G. let ÷ i be the lift to G of ~ i . Ker r Tˆ G. hence Ker ÷ Tˆ G. and so G is not simple.5. there is an irreducible character ÷ of G such that Ker ÷ is not {1} or G. As Ker ÷ Tˆ G. then Ker ÷ ˆ Ker r by Theorem 13. . suppose that G is not simple. Since ÷ is non-trivial and irreducible. Proof Suppose there is a non-trivial irreducible character ÷ such that ÷( g) ˆ ÷(1) for some non-identity element g.172 Representations and characters of groups Now let ~1 .

Then ÷ is a homomorphism from G to the multiplicative group of non-zero complex numbers. As a preliminary step. (2) Let G ˆ S3 . h P G. . 17. then G9 < Ker ÷. 17. it is necessary to determine the derived subgroup of G.Normal subgroups and lifted characters 173 ®rst move in constructing the character table of G is often to write down the linear characters. we explore some group-theoretic properties of the derived subgroup.9 Proposition If ÷ is a linear character of G. Thus G9 ˆ h[ g. Clearly [ g. h] ˆ 1 for all g. h] ˆ (1 2 3). let G9 be the subgroup of G which is generated by all elements of the form g À1 hÀ1 gh ( g. so G9 ˆ {1}. j Next. We abbreviate gÀ1 hÀ1 gh as [ g. h P GiX 17. h P G)X Then G9 is called the derived subgroup of G. for all g. We are going to show that G9 v G and that the linear characters of G are the lifts to G of the irreducible characters of GaG9. One step is provided by the following proposition. If g ˆ (1 2) and h ˆ (2 3) then [ g. Therefore.8 Examples (1) If G is abelian then [ g. Hence G9 ˆ h(1 2 3)i ˆ A3 .7 De®nition For a group G. ÷( g À1 hÀ1 gh) ˆ ÷( g)À1 ÷(h)À1 ÷( g)÷(h) ˆ 1X Hence G9 < Ker ÷. which is de®ned in the following way. Proof Let ÷ be a linear character of G. h]: g. h]. h] is always an even permutation. so G9 < A3 . h P G.

(2) Let g. In particular. . b. to prove that G9 v G it is suf®cient by the ®rst sentence to prove that x À1 [ g. GaG9 is abelian.10 that G9 is the smallest normal subgroup of G with abelian factor group. h]x ˆ x À1 g À1 hÀ1 ghx ˆ (x À1 gx)À1 (x À1 hx)À1 (x À1 gx)(x À1 hx) ˆ [x À1 gx. all of degree 1. But x À1 [ g. . and x À1 aÀ1 x ˆ (x À1 ax)À1 X Now G9 consists of products of elements of the form [ g. Proof (1) Note that for all a.8 shows that GaG9 has exactly m irreducible characters ~1 . and so divides |G|. h]x P G9 for all g. h] and their inverses. Since GaG9 is abelian. (2) G9 < N if and only if GaN is abelian. ~ m . Proof Let m ˆ jGaG9j.174 Representations and characters of groups 17. . x P G. ÷ m of these characters to G also have degree 1. x P G. . we deduce that GaG9 is abelian. . Given the derived subgroup G9. (1) G9 v G. j It follows from Proposition 17. . In particular. we have x À1 (ab)x ˆ (x À1 ax)(x À1 bx). Theorem 9.3 they are precisely the irreducible characters ÷ of G . We have ghgÀ1 hÀ1 P N D Ngh ˆ Nhg D (Ng)(Nh) ˆ (Nh)(Ng)X Hence G9 < N if and only if GaN is abelian. we can obtain the linear characters of G by applying the next theorem. h. x À1 hx]X Therefore G9 v G. 17. and by Theorem 17. Since we have proved that G9 v G. . ÷ ÷ The lifts ÷1 .10 Proposition Assume that N v G. . the number of distinct linear characters of G is equal to jGaG9j.11 Theorem The linear characters of G are precisely the lifts to G of the irreducible characters of GaG9. Therefore. h P G.

we have G9 < A n by Proposition 17. where ÷ ÷ n ~1 (An (1 2)) ˆ 1. 2). . G9 contains all 3-cycles and all elements of cycle-shape (2.8(2).13 Example We ®nd the linear characters of S n (n > 2). j 17.9. h] ˆ (1 2 3). Since Sn aS9 ˆ fAn . An (1 2)g  C2 . so G9 ˆ {1} ˆ A n . But every product of two transpositions is equal to the identity. we know that S9 ˆ An . which are given by ÷1 ˆ 1 Sn . a 3-cycle or an element of cycle-shape (2. all the elements in (1 2 3) G and (1 4)(2 3) G belong to G9. Therefore. and A n consists of permutations.Normal subgroups and lifted characters 175 such that G9 < Ker ÷. 17. We have now proved that G9 ˆ A n . by Theorem 12.11. . S n has exactly two linear characters ÷1 . 2). If n ˆ 1 or 2 then S n is abelian. as the next result shows. ÷ ~2 (An (1 2)) ˆ À1X ÷ Therefore by Theorem 17. but they can also be used to construct new irreducible characters from old.12 Example Let G ˆ S n . . [h. If g ˆ (1 2). . ÷ m are therefore all the linear characters of G. ÷2 .15. We shall show that G9 ˆ An . if g P An X a À1. . As S n aA n  C2 . each of which is the product of an even number of transpositions. @ ÷2 ( g) ˆ 1. Not only are the linear characters of G important in being irreducible characters. the characters ÷1 . Therefore A n < G9. We proved that S9 ˆ A3 in Example 3 17. k] ˆ (1 4)(2 3)X Since G9 v G. if g P An . so we assume that n > 4. From the last example. the group n n S n aS9 has two linear characters ~1 and ~2. then [ g. h ˆ (2 3) and k ˆ (1 2)(3 4).10(2). In view of Proposition 17.

The normal subgroups of G can be found from the character table of G. Since r and ë are homomorphisms it follows easily that rë is a homomorphism. ÷iX jGj gPG By Theorem 14. 3. so ë( g)ë( g) ˆ 1. ÷ ÷ 2. The matrix g(rë) has trace ë( g) tr ( gr). the complex number ë( g) is a root of unity. The linear characters of G are precisely the lifts to G of the irreducible characters of GaG9. ÷ëi ˆ ÷( g)ë( g)÷( g)ë( g) jGj gPG ˆ 1 ˆ ÷( g)÷( g) ˆ h÷. C) by g(rë) ˆ ë( g)( gr) ( g P G)X Thus g(rë) is the matrix gr multiplied by the complex number ë( g). j The general case of a product of two characters will be discussed in Chapter 19.14 Proposition Suppose that ÷ is a character of G and ë is a linear character of G. it follows that ÷ë is irreducible if and only if ÷ is irreducible. Moreover. which is ë( g)÷( g). de®ned by ÷ë( g) ˆ ÷( g)ë( g) ( g P G) is a character of G. Proof Let r: G 3 GL (n. Therefore 1 ˆ h÷ë. then so is ÷ë.20. if ÷ is irreducible. Then the product ÷ë. and is given by ÷( g) ˆ ~(Ng) ( g P G). Now for all g P G. Summary of Chapter 17 1. The character of G which corresponds to the character ~ ÷ of GaN is the lift of ~. Characters of GaN correspond to characters ÷ of G for which N < Ker ÷.176 Representations and characters of groups 17. Hence rë is a representation of G with character ÷ë. De®ne rë: G 3 GL (n. . C) be a representation with character ÷.

bÀ1 ab ˆ aÀ1 l. bÀ1 ab ˆ a2 X (a) Show that G has order 21. . Compare your table with the character table of D8 (Example 16.Normal subgroups and lifted characters Exercises for Chapter 17 177 1. 4. (c) Find the character table of G. 4 or 12 linear characters. and hence cannot be simple. The character table of D8 is as shown (see Example 16. a2 b 1 À1 1 À1 0 ab. bl. (c) Complete the character table of G. g6 (where g1 ˆ 1). Show that every group of order 12 has 3. (b) Find G9. b2 ˆ a2 . (b) Find the conjugacy classes of G. 3.14 to complete the character table of G. A certain group G of order 12 has precisely six conjugacy classes. a3 b 1 À1 À1 1 0 . a3 1 1 À1 À1 0 b. b ˆ (2 3 5)(4 7 6)X Let G ˆ ka. and construct all the linear characters of G.3(3)): 1 ÷1 ÷2 ÷3 ÷4 ÷5 1 1 1 1 2 a2 1 1 1 1 À2 a. 2. and has irreducible characters ÷. with representatives g1 . b: a4 ˆ 1. . What are the sizes of the conjugacy classes of G? 5. Let G ˆ Q8 ˆ ka. ö with values as follows: g1 ÷ ö 1 2 g2 Ài 0 g3 i 0 g4 1 À1 g5 À1 À1 g6 À1 2 Use Proposition 17.3(3)). Let a and b be the following permutations in S7 : a ˆ (1 2 3 4 5 6 7). . Check that a7 ˆ b3 ˆ 1. . (a) Find the ®ve conjugacy classes of G.

For n > 1. then there is a representation of U6n over C which sends     0 å ù 0 X a3 . as in Proposition 17. then there is a representation of V8 n over C which sends     å 0 0 1 X a3 . b: a2 n ˆ b4 ˆ 1. the group U6 n ˆ ha.b3 X 0 å À1 ån 0 (b) Find all the irreducible representations of T4 n. The group V8 n ˆ ha. ba ˆ aÀ1 bÀ1 . bÀ1 a ˆ aÀ1 bi has order 8n. 7. then there is a representation of T4 n over C which sends     å 0 0 1 a3 . b: a2 n ˆ b3 ˆ 1.5. bÀ1 ab ˆ aÀ1 i has order 4n. Let n be an odd positive integer. (a) Let ù ˆ e2ðia3 .b3 0 Àå À1 À1 0 (b) Find all the irreducible representations of V8 n . (a) Show that if å is any nth root of unity in C. (It is known as a dicyclic group. an ˆ b2 . aÀ1 ba ˆ bÀ1 i has order 6n. . You are given that the group T4 n ˆ ha. Show that if å is any (2n)th root of unity in C. 6.) (a) Show that if å is any (2n)th root of unity in C. b: a2 n ˆ 1. 8.178 Representations and characters of groups Express each normal subgroup of D8 as an intersection of kernels of irreducible characters.b3 å 0 0 ù2 (b) Find all the irreducible representations of U6 n.

and all dihedral groups. Let ÷4 be the character ÷4 ( g) ˆ jfix ( g)j À 1 ( g P S4 ) which is given in Proposition 13. ÷4 and ÷4 ÷2 are as follows: gi |CG ( gi )| ÷2 ÷4 ÷4 ÷2 1 24 1 3 3 (1 2) 4 À1 1 À1 (1 2 3) 3 1 0 0 (1 2)(3 4) 8 1 À1 À1 (1 2 3 4) 4 À1 À1 1 Note that h÷4 . ÷2 .1 The group S4 In Example 17. 24 4 8 4 179 . we produced three irreducible characters ÷1 . 18. By Proposition 17. including the groups S4 and A4 .14. The values of ÷2 . ÷3 of S4 by lifting characters of the factor group S4 aV4 . We shall now use Proposition 17.4. which deals with the product of a character with a linear character. to complete the character table of S4 .18 Some elementary character tables We now illustrate the techniques we have presented so far by constructing the character tables of several groups. ÷4 i ˆ 9 1 1 1 ‡ ‡ ‡ ˆ 1.24.14. the product ÷4 ÷2 is also a character of S4 .

(1 2 3). we have now found the complete character table of S4 . (1 3 2) (see Example 12. and G has four conjugacy classes. Since S4 has ®ve conjugacy classes. the alternating group of degree 4. Thus jGaG9j ˆ 3 by Theorem 17. Since G has four irreducible characters. It is not dif®cult to con®rm this by showing that . so that í( g) ˆ |®x ( g)| À 1 for all g P A4 .18(1)). Let ÷5 ˆ ÷4 ÷2 . Let í be the character of A4 given by Proposition 13. íi ˆ 9 1 ‡ ˆ 1. The character ÷4 ÷2 is also irreducible. with representatives 1. either by using the same calculation or by quoting the result of Proposition 17. The values of í are as follows: gi |CG ( gi )| í 1 12 3 (1 2)(3 4) 4 À1 (1 2 3) 3 0 (1 3 2) 3 0 Note that hí. and the sum of the squares of their degrees is 12.2 The group A4 Let G ˆ A4 . and we have produced ®ve irreducible characters. Then |G| ˆ 12. (1 2)(3 4).14. 12 4 so í is an irreducible character of G of degree 3.180 Representations and characters of groups so ÷4 is irreducible. there must be exactly three linear characters of G.24. as shown.11. Character table of S4 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 1 24 1 1 2 3 3 (1 2) 4 1 À1 0 1 À1 (1 2 3) 3 1 1 À1 0 0 (1 2)(3 4) 8 1 1 2 À1 À1 (1 2 3 4) 4 1 À1 0 À1 1 18.

(1 4)(2 3)gX 181 Now GaG9 ˆ fG9. ~3 to G. (1 3)(2 4). s P Z).Some elementary character tables G9 ˆ V4 ˆ f1. G9(1 2 3). so that G ˆ ha. and the character table of GaG9 is G9 ~1 ÷ ~2 ÷ ~3 ÷ 1 1 1 G9(1 2 3) 1 ù ù2 G9(1 3 2) 1 ù2 ù ~ ÷ ÷ (where ù ˆ e2ðia3 ). de®ne  j    0 1 å 0 Aj ˆ .3 The dihedral groups Let G be the dihedral group D2 n of order 2n. (1 2)(3 4). . give the complete character table of A4 : Character table of A4 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 1 12 1 1 1 3 (1 2)(3 4) 4 1 1 1 À1 (1 2 3) 3 1 ù ù2 0 (1 3 2) 3 1 ù2 ù 0 18. C) by (ar bs )r j ˆ (Aj ) r (Bj ) s (r. b: an ˆ b2 ˆ 1. G9(1 3 2)g  C3. BÀ1 Aj Bj ˆ AÀ1 X j j j j It follows that by de®ning r j : G 3 GL(2. na2. The lifts of ÷1 . na2. For each integer j with 1 < j . together with the character ÷4 ˆ í. we obtain a representation r j of G for each j with 1 < j . with n > 3. bÀ1 ab ˆ aÀ1 iX We shall derive the character table of G. ~2 . Bj ˆ X 1 0 0 åÀ j Check that A n ˆ B2 ˆ I. Write å ˆ e2ðia n .

so ar i and ar j have different eigenvalues. Let ø j be the character of r j . X X X .182 Representations and characters of groups Each r j is an irreducible representation.5(2) or by applying the result of Exercise 8. and so r i and r j are not equivalent. Therefore there is no matrix T with ar i ˆ T À1 (ar j )T. These characters ÷1 and ÷2 are given by ÷1 ˆ 1 G and & 1 if g ˆ ar for some r. We have now constructed distinct irreducible characters ø j of G. At this point it is convenient to consider separately the cases where n is odd and where n is even. Case 1: n odd By (12. aÀ r g(1 < r < (n À 1)a2). ø( nÀ1)a2 each have degree 2. na2. Since kal v G and Gahai  C2. fas b: 0 < s < n À 1gX Thus there are (n ‡ 3)a2 conjugacy classes. If i and j are distinct integers with 1 < i . The (n À 1)a2 irreducible characters ø1 . then å i Tˆ å j and å i Tˆ å À j . ø2 .11) the conjugacy classes of D2 n (n odd) are f1g. na2.4. one for each j which satis®es 1 < j . ÷2 of G by lifting the irreducible characters of Gahai to G.11. far . there are two more to be found. As G has (n ‡ 3)/2 irreducible characters in all. na2 and 1 < j . we have proved that D9 n ˆ kal for n odd. either by the proof of Example 5. in view of 2 Theorem 17.) The character table of D2 n (n odd) is therefore as follows (where å ˆ e2ðia n ): gi |CG ( gi )| ÷1 ÷2 øj (1 < j < (n À 1)/2) 1 2n 1 1 2 ar (1 < r < (n À 1)/2) n 1 1 ‡ å À jr b 2 1 À1 0 å jr . (Incidentally. we obtain two linear characters ÷1 . ÷2 ( g) ˆ À1 if g ˆ ar b for some rX We have now found all the irreducible characters of D2 n (n odd).

4 Another group of order 12 We shall now describe a non-abelian group G of order 12 which is not isomorphic to either A4 or D12. they are easy to calculate. fas b: s oddgX Hence G has m ‡ 3 irreducible characters. say n ˆ 2m. ÷4 (and G9 ˆ ka2 l). aÀ r g(1 < r < m À 1). ø2 . far . X X X . then the conjugacy classes of D2 n. are f1g. we ®rst note that ha2 i ˆ fa j : j eveng is a normal subgroup of G and Gaha2 i ˆ fha2 i. å ˆ e2ðia n ). ha2 iabg  C2 3 C2 X Therefore G has four linear characters ÷1 .12). fas b: s eveng. ÷2 . as supplied by (12. ha2 ib. It is in fact known that every non-abelian group of order . fam g. ÷3 .Some elementary character tables 183 Case 2: n even If n is even. and their values appear in the following complete character table of D2 n (n even. gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 øj (1 < j < m À 1) 1 2n 1 1 1 1 2 am 2n 1 1 (À1) m (À1) m 2(À1) j a r (1 < r < m À 1) n 1 1 (À1) r (À1) r jr å ‡ å À jr b 4 1 À1 1 À1 0 ab 4 1 À1 À1 1 0 18. and we shall construct the character table of G. ha2 ia. ø mÀ1 X To ®nd the remaining four irreducible characters. of which m À 1 are given by ø1 . Since these linear characters are the lifts of the irreducible characters of G/ka2 l. n ˆ 2m.

but we shall not prove this result here. b ˆ (1 7 4 10)(2 12 5 9)(3 11 6 8). ha2 ia. namely ar . a2 . a4 } v G. a5 b} Representative gi 1 a3 a a2 b ab |CG ( gi )| 12 12 6 6 4 4 Therefore G has six irreducible characters. bÀ1 ab ˆ aÀ1 X It follows from these relations that every element of G has the form ar bs with 0 < r < 5. aÀ1 } {a2 . ha2 iabgX Since ka2 la ˆ ka2 lb2 . aÀ2 } {b. a the group G has at least 12 elements. a3 ˆ b2 . a2 b.184 Representations and characters of groups 12 is isomorphic to A4 . 0 < s < 1 as given above. ÷2 . Observe that ka2 l ˆ {1. b. CG (a3 ) ˆ G. a3 . help us to ®nd the conjugacy classes of G. bl. which are tabulated below: Conjugacy class {1} {a3 } {a. ha2 ib. ar b (0 < r < 5)X Check that a and b satisfy a6 ˆ 1. a4 b} {ab. and similar facts. CG (b) ˆ f1. The relations further imply that CG (a) ˆ hai. ÷4 of G given below: . a3 bgX These. and Gaha2 i ˆ fha2 i. we have Gaha2 i  C4 . and so |G| ˆ 12. D12 or G. ÷3 . a3 b. a subgroup of S12 . Since a has order 6 and b P kal. By lifting the irreducible characters of C4 to G. and let G ˆ ka. we obtain the linear characters ÷1 . Let a and b be the following permutations in S12 : a ˆ (1 2 3 4 5 6)(7 8 9 10 11 12).

so we can solve them for á r and â r .Some elementary character tables gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 1 12 1 1 1 1 á1 â1 a3 12 1 À1 1 À1 á2 â2 a 6 1 À1 1 À1 á3 â3 a2 6 1 1 1 1 á4 â4 b 4 1 i À1 Ài á5 â5 185 ab 4 1 Ài À1 i á6 â6 It remains to ®nd the values á r . 4 ‡ 2á3 À 2â3 ˆ 0. the column orthogonality relations 6 ˆ iˆ1 ÷ i ( g r )÷ i ( g 1 ) ˆ 0 and 6 ˆ iˆ1 ÷ i ( g r )÷ i ( g 2 ) ˆ 0 now give us two equations involving 2á r ‡ 2â r and 2á r À 2â r . so á2 and â2 are integers by Corollary 13. 2á6 À 2â6 ˆ 0X . 2.4(2). we have 4 ‡ á2 ‡ â2 ˆ 12. ÷6 . so they are positive integers. 2 2 á1 á2 ‡ â1 â2 ˆ 0X Since á1 . we may take á2 ˆ 2 and â2 ˆ À2. 2á5 ‡ 2â5 ˆ 0. For this. For r . Observe that á1 . â1 are the degrees of ÷5 . â r taken by the last two irreducible characters ÷5 . ÷6 . 1 1 4 ‡ á2 ‡ â2 ˆ 12. respectively. the ®rst equation gives á1 ˆ â1 ˆ 2. we shall use the column orthogonality relations. The other two equations then imply that á2 ˆ Àâ2 ˆ Æ2. â1 are positive integers. 2á6 ‡ 2â6 ˆ 0. Since we have not yet distinguished between ÷5 and ÷6 . 2á5 À 2â5 ˆ 0. By the column orthogonality relations applied to columns 1 and 2. 4 ‡ 2á4 ‡ 2â4 ˆ 0. also a3 is an element of order 2.10. Explicitly: r ˆ 3: r ˆ 4: r ˆ 5: r ˆ 6: 2á3 ‡ 2â3 ˆ 0. 2á4 À 2â4 ˆ 0. Theorem 16.

without constructing the corresponding CG-modules. 3. as follows. (In fact. . Section 18. â5 ˆ 0. it is not hard to construct the representations of the above group G with characters ÷5 and ÷6 ± see Exercise 17. á6 ˆ 0.3: the dihedral groups. â6 ˆ 0X The complete character table of G is therefore as follows: gi CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 1 12 1 1 1 1 2 2 a3 12 1 À1 1 À1 2 À2 a 6 1 À1 1 À1 À1 1 a2 6 1 1 1 1 À1 À1 b 4 1 i À1 Ài 0 0 ab 4 1 Ài À1 i 0 0 We can deduce that G is not isomorphic to A4 or D12 from the fact that the character table of G is different from those of A4 and D12. It is instructive to note that we produced the last two irreducible characters of G by simply using the orthogonality relations. â4 ˆ À1. This is typical of more advanced calculations. á4 ˆ À1. and illustrates the fact that it is usually much easier to construct an irreducible character of a group than to obtain an irreducible representation. Section 18.1: the group S4 .2: the group A4 . á5 ˆ 0.) Summary of Chapter 18 In this chapter we gave the character tables of various groups.6. â3 ˆ 1. 1.186 Hence Representations and characters of groups á3 ˆ À1. Section 18. 2.

It is a good idea to do the cases n odd and n even separately.8. as in Exercise 17. 5. Write down explicitly the character table of D12. Use the character table to ®nd seven distinct normal subgroups of D12. (Hint: use the result of Exercise 17. (Hint: use Proposition 17. and express ð as a sum of irreducible characters. as in Example 1. Find the values of ð on the elements of D8. with n odd. ba ˆ aÀ1 bÀ1 . Find the character table of G. and show that all its entries are integers. Let ð be the corresponding permutation character of D8. Find the character table of G. Let G ˆ V8 n ˆ ha.7. an ˆ b2 . Find the character table of G. b: a2n ˆ b4 ˆ 1. aÀ1 ba ˆ bÀ1 l.Some elementary character tables Exercises for Chapter 18 187 1.5. bÀ1 a ˆ aÀ1 bi.) 3. . Let G ˆ U6 n ˆ ka.6.) 4. b: a2 n ˆ 1. as in Exercise 17. bÀ1 ab ˆ aÀ1 i. Let G ˆ T 4n ˆ ha. as in Exercise 17. 2. b: a2n ˆ b3 ˆ 1. Regard D8 as a subgroup of S4 permuting the four corners of a square.1(3).6.

given those of G and H. respectively. then. . ÷ 4 . . An important special case of the product ÷ø occurs when ÷ ˆ ø. . which has character ÷ø. and indeed. It is therefore straightforward to calculate the product. and to put them together to form a new CG-module. increase. and more generally ÷ 3 . and it can be extended to include the product of any pair of characters ÷ and ø. we apply tensor products in a different way. wn . The value of the product ÷ø on an element g of G is simply ÷( g)ø( g). For each i. we have a chance of getting a large proportion of the character table of G from just one non-linear character ÷ of G. and so on. products of characters provide a very good source of new characters from given ones. we introduce a symbol v i  wj. Tensor product spaces Let V and W be vector spaces over C with bases v1 . At the end of the chapter. . 1 < j < n. . ÷ 2 . If ÷ is not linear. . The tensor product space V  W is de®ned to be the mn-dimensional vector space over C with a basis given by 188 . j with 1 < i < m.19 Tensor products The idea of multiplying a character of a group G by a linear character of G was introduced at the end of Chapter 17. called the tensor product of V and W. The plan is to take CG-modules V and W with characters ÷ and ø respectively. but a little ingenuity is required in order to justify the conclusion that the product ÷ø is a character of G. and by taking successive powers of ÷ we obtain arbitrarily many new characters. then the degrees of ÷. to ®nd all the irreducible characters of a direct product G 3 H. Potentially. . X X X . We shall illustrate this by constructing the character tables of S5 and S6 . v m and w1 . so we consider the character ÷ 2 .

. (2v1 À v2 )  (w1 ‡ w2 ) ˆ 2v1  w1 ‡ 2v1  w2 À v2  w1 À v2  w2 X Do not be misled by the notation into believing that every element of V  W has the form v  w. . j ëë i ì j (v i  wj ). . j ˆ ˆ i. j i. we de®ne v  w P V  W by ˆ vwˆ ë i ì j (v i  wj )X i. j 189 wˆ €n jˆ1 ì j wj For example. . it is impossible to express v1  w1 ‡ v2  w2 in the form v  w. 19. yb P W. ì j P C).Tensor products fv i  wj : 1 < i < m. For instance. because this is not the case. ë(v  w) ˆ ë ˆ i. j (ëv)  w ˆ 2 ˆ i 3 ëë i v i  2ˆ j 3 ì jw j ˆ i. i j i. w P W and ë P C.1 Proposition (1) If v P V. xa P V and y1. . then v  (ëw) ˆ (ëv)  w ˆ ë(v  w)X (2) If x1 . . . . then 2 a 3 H b I ˆ ˆ ˆ xi  d yj e ˆ xi  yj X iˆ1 jˆ1 i. Then jˆ1 2 3 2 3 ˆ ˆ ˆ v  (ëw) ˆ ëi vi  ëì j w j ˆ ëë i ì j (v i  wj ). 1 < j < ngX Thus V  W consists of all expressions of the form ˆ ë ij (v i  wj ) (ë ij P C)X €m For v P V and w P W with v ˆ iˆ1 ë i v i and (ë i . j ë i ì j (v i  wj ) ˆ ëë i ì j (v i  wj )X . j €m € Proof (1) Let v ˆ iˆ1 ë i v i and w ˆ n ì j w j.

. We know that the elements v i  wj (1 < i < m. 1 < l < n) V  W has dimension mn. . . fn is a basis of W.1. Tensor product modules We have introduced the tensor product of two vector spaces. 1 < j < n) are a basis of the mn elements ek  fl (1 < k < m. then the elements in fei  f j : 1 < i < m. .2 Proposition If e1 . Since elements ek  fl are v i  wj (1 < i < m. . . . . 19. . the next proposition shows that other bases of V and W work equally well. ì jl P C)X Then by Proposition 19. w n. . l Now the elements V  W . wj ˆ n ˆ lˆ1 ì jl f l (ë ik . 1 < j < ng give a basis of V  W. Let G be a ®nite group and let V and W be CG-modules with bases v1 . Proof Write vi ˆ m ˆ kˆ1 ë ik ek . and we leave it as an exercise. X X X .190 Representations and characters of groups Therefore v  (ëw) ˆ (ëv)  w ˆ ë(v  w). so we are now in a position to de®ne the tensor product of two CG-modules. 1 < j < n) give a basis of V  W. v m and w1 . em is a basis of V and f1 . The proof of part (2) is equally straightforward. respectively. The multiplication of v i  wj by an element of . we have ˆ v i  wj ˆ ë ik ì jl (ek  f l )X k. j Our construction of V  W depended upon choosing a basis of V and a basis of W at the beginning. it follows that the also a basis of V  W. and hence span V  W. . .

which is then extended linearly to a multiplication on the whole of V  W. we have (v  w) g ˆ v g  wgX Proof Let v ˆ € ë i v i and w ˆ n ì j wj.4 Proposition For all v P V. For all i. consider what happens when r is a scalar multiple of g.3 De®nition Let g P G. 19. j i. j ë i ì j (v i g  wj g) 3 ëivi g  2ˆ j j ˆ i 3 ì j wj g by Proposition 19X1 ˆ v g  wgX You should be warned that (v  w)r Tˆ vr  wr for most elements r in CG. .3. Then jˆ1 2ˆ 3 (v  w) g ˆ ë i ì j (v i  wj ) g by Proposition 19X1 iˆ1 i. 19. makes the vector space V  W into a CGmodule. j for arbitrary complex numbers ë ij . j €m ˆ ˆ ˆ 2 i. For example.5 Proposition The rule for multiplying an element of V  W by an element of G.Tensor products 191 G is de®ned in the following simple way. w P W and all g P G. given in De®nition 19. let 2ˆ 3 ˆ ë ij (v i  wj ) g ˆ ë ij (v i g  wj g) i. 19. j. de®ne (v i  wj ) g ˆ v i g  w j g and. more generally.

respectively. if ö is the character of V  W then . em of V and a basis f1 . By Proposition 9. Hence. ø( g) ˆ ì jX Now for 1 < i < m and 1 < j < n. . and V  W is a CG-module. 19. and g. Then the character of the CG-module V  W is the product character ÷ø. j by Proposition 19X4 3 ˆ ë ij (v i  wj ) g ˆ ë ij ((v i  wj ) g)X i. ì j . fn of W such that ei g ˆ ë i e i (1 < i < m) and f j g ˆ ì j f j m ˆ iˆ1 n ˆ jˆ1 (1 < j < n) for some complex numbers ë i . 1 < j < n. (ei  f j ) g ˆ ei g  f j g ˆ ë i ì j (ei  f j ). j Therefore all the conditions of Proposition 4. .2.192 Representations and characters of groups Proof Let 1 < i < m.6 are ful®lled.11 we can choose a basis e1 . and by Proposition 19. . . where ÷ø( g) ˆ ÷( g)ø( g) for all g P GX Proof Let g P G. . (v i  wj )( gh) ˆ v i ( gh)  wj ( gh) ˆ (v i g)h  (wj g)h ˆ (v i g  wj g)h ˆ ((v i  wj ) g)h. j We now calculate the character of V  W. these vectors ei  fj form a basis of V  W. . Then ÷( g) ˆ ë i . h P G. and 2ˆ i. .6 Proposition Let V and W be CG-modules with characters ÷ and ø. . Then (v i  wj ) g ˆ v i g  wj g P V  W . (v i  wj )1 ˆ v i  wj .

and ÷4 ÷4 ˆ ÷1 ‡ ÷3 ‡ ÷4 ‡ ÷5 X Powers of characters Corollary 19. Ker ˆ {1}). as can be seen from Theorem 19. j i j 193 as required. the product of ÷ with itself. the powers of ÷ carry a lot of information about the whole character table of G. More generally.7 shows that if ÷ is a character of G then so is ÷ 2 . i.Tensor products 2 32 ˆ ˆ ˆ 3 ö( g) ˆ ëi ì j ˆ ëi ì j ˆ ÷( g)ø( g). When ÷ is a faithful character (that is. and calculate ÷3 ÷4 and ÷4 ÷4 .1.7 Corollary The product of two characters of G is again a character of G. We reproduce it here. where ÷ 2 ˆ ÷÷. j 19. gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷3 ÷4 ÷4 ÷4 1 24 1 1 2 3 3 6 9 (1 2) 4 1 À1 0 1 À1 0 1 (1 2 3) 3 1 1 À1 0 0 0 0 (1 2)(3 4) 8 1 1 2 À1 À1 À2 1 (1 2 3 4) 4 1 À1 0 À1 1 0 1 We see that ÷3 ÷4 ˆ ÷4 ‡ ÷5 . 19. we de®ne ÷ n by ÷ n ( g) ˆ (÷( g)) n 0 for all g P GX Thus ÷ ˆ 1 G .10 below.8 Example The character table of S4 was given in Section 18. An inductive proof using Corollary 19. for every nonnegative integer n. .7 shows that ÷ n is a character of G.

xr are indeterminates. j . Suppose that x1 . We ®rst sketch a proof of this result. x2 must come from all the factors in the rÀ1 rÀ2 second row. . we must take x1 from the rÀ2 ®rst row of the matrix. we shall need the following result concerning the so-called `Vandermonde matrix'.9) If á1 . and consider H rÀ1 I 1 x1 x2 X X X x1 1 f 1 x x2 X X X x rÀ1 g 2 f g 2 2 Ä ˆ detf gX dX X X XXX X e 1 xr x2 r XXX x rÀ1 r If i Tˆ j and xi ˆ xj then two rows of the given matrix are equal. x2 from the second row. . and so on. . On the other hand. Hence rÀ1 rÀ2 the coef®cient of x1 x2 . to obtain x1 x2 X X X x rÀ1 rÀ1 in the expansion of the determinant Ä. It follows that ‰ ÄˆÆ (xi À xj )X i. j complex numbers. . . and so on. (19.10. It follows that Ä is divisible by ‰ (xi À x j ) ˆ (x1 À x2 )(x1 À x3 ) X X X (x1 À xr ) i. . . . . x rÀ1 in Ä is Æ1.194 Representations and characters of groups In the course of the proof of Theorem 19. so Ä ˆ 0. then the matrix rÀ1 I á2 X X X á1 1 rÀ1 á2 X X X á2 g g 2 g X XXX X e á2 r XXX á rÀ1 r ár 3 (x2 À x3 ) X X X (x2 À xr ) F F F 3 (x rÀ1 À xr )X rÀ1 rÀ2 Now the coef®cient of x1 x2 X X Xx rÀ1 in this product is 1: for in the rÀ1 way we have displayed the product. the term x1 must come from all rÀ2 the factors in the ®rst row. á r are distinct H 1 á1 f1 á 2 f Aˆf dX X 1 is invertible. .

øi Tˆ 0 for some j with 0 < j < r À 1. As ÷ is faithful. . X X X . Then ÷( g) takes four different values. and let b be the row vector which is given by b ˆ (â1 . then ÷( g) takes just two different values (see Proposition 13.9). j 19. and thus h÷ j . á r .20). We have seen that ÷ 2 ˆ ÷1 ‡ ÷3 ‡ ÷4 ‡ ÷5 . ÷ 1 .10 Theorem Let ÷ be a faithful character of G. so Theorem 19. as we wished to prove.10 says that every irreducible character of G is a constituent of 1 G or ÷. â r )X Now A is invertible by (19. (2) Let G ˆ S4 . and refer to Example 19. de®ne Gi ˆ f g P G: ÷( g) ˆ á i gX Take á1 ˆ ÷(1). . Let ÷ ˆ ÷4 . øi ˆ r 1 ˆ 1 ˆ (÷( g)) j ø( g) ˆ (á i ) j â i X jGj gPG jGj iˆ1 Let A be the r 3 r matrix with ij-entry (á i ) jÀ1 .8. Now let ø be an irreducible character of G. . Then every irreducible character of G is a constituent of one of the powers ÷ 0 . Proof Let the r values taken by ÷ be á1 . Then for all j > 0. . and b Tˆ 0 since â1 Tˆ 0.5. For 1 < i < r.9). ÷ rÀ1 . . gPG i and note that â1 ˆ ø(1) Tˆ 0. by Theorem 10. øi.Tensor products 195 To obtain (19. and suppose that ÷( g) takes precisely r different values as g varies over all the elements of G. . and for 1 < i < r. and deduce that the matrix A is invertible since its determinant is non-zero. . so that G1 ˆ Ker ÷.11 Examples (1) If G Tˆ {1} and ÷ is the regular character of G. G1 ˆ {1}. hence bA Tˆ 0. We must show that h÷ j . we know this already. øi Tˆ 0 for some j with 0 < j < r À 1. 19. . h÷ j . let ˆ âi ˆ ø( g). we substitute á i for xi (1 < i < r). But the ( j ‡ 1)th entry in the row vector bA is equal to jGjh÷ j .

v n be a basis of V. V  V ˆ S(V  V ) È A(V  V )X .10. This special case is particularly useful in ®nding irreducible characters.12 Proposition The subspaces S(V  V) and A(V  V) are CG-submodules of V  V. ÷ 1 . ÷ 2 . it is easy to see that S(V  V) and A(V  V) are subspaces of V  V (indeed. it is of some importance to be able to decompose powers of a character ÷ into sums of irreducible characters. . By Proposition 19. j i. they are eigenspaces of T). Let v1 . . the module V  V has character ÷ 2 .6. and the subspace A(V  V ) is known as the antisymmetric part of V  V. . We are going to provide a method for decomposing ÷ 2 . . 2ˆ 3 ˆ ë ij (v i  v j ) T ˆ ë ij (v j  v i )X i. Also. j and extending linearly ± that is. Decomposing ÷ 2 In view of Theorem 19. . j Check that for all v. Since T is linear.196 Representations and characters of groups h÷ 3 . we have (v  w)T ˆ w  vX Hence T is independent of the choice of basis. ÷5 of G. in this case. the square of ÷. as we shall see. A(V  V ) ˆ fx P V  V : xT ˆ Àxg. illustrating Theorem 19. 19. . ÷ 3 (indeed. . The subspace S(V  V) is called the symmetric part of V  V. ÷ 3 ) have between them as constituents all the irreducible characters ÷1 . just ÷ 2 .10. and de®ne a linear transformation T: V  V 3 V  V by (v i  v j )T ˆ v j  v i for all i. . Let V be a CG-module with character ÷. w P V. Now de®ne subsets of V  V as follows: S(V  V ) ˆ fx P V  V : xT ˆ xg. ÷2 i ˆ 1X and we ®nd that Thus ÷ 0 .

Tensor products Proof For all ë ij P C and g P G. . we have (xg)T ˆ (xT ) g ˆ xg. . The dimension of A(V  V ) is n(n À 1)a2. j i. A(V  V ). . j < n) form a basis of A(V  V). We now present bases of the symmetric and antisymmetric parts of V  V which consist of elements like these. Hence. . If x P S(V  V) ’ A(V  V) then x ˆ xT ˆ Àx. y P A(V  V) and g P G. Thus S(V  V) and A(V  V) are CG-submodules of V  V. for x P S(V  V). so xg P S(V  V) and yg P A(V  V). 2ˆ 3 ˆ ë ij (v i  v j ) Tg ˆ ë ij (v j g  v i g) i. j Therefore T is a CG-homomorphism from V  V to itself. w P V. j 197 ˆ ˆ ˆ i. while the antisymmetric part of V  V contains all vectors of the form v  w À w  v.13 Proposition Let v1 . The dimension of S(V  V ) is n(n ‡ 1)a2. . (1) The vectors v i  v j ‡ v j  v i (1 < i < j < n) form a basis of S(V  V ). (2) The vectors v i  v j À v j  v i (1 < i . j ë ij (v i g  v j g)T 3 ë ij (v i  v j ) gT X 2ˆ i. 19. and ( yg)T ˆ ( yT ) g ˆ À yg. for all x P V we have x ˆ 1(x ‡ xT ) ‡ 1(x À xT )X 2 2 Since T 2 is the identity. Therefore. Further. 1 2(x ‡ xT ) P S(V  V ) and 1 2(x À xT ) P j V  V ˆ S(V  V ) È A(V  V )X Note that the symmetric part of V  V contains all vectors which have the form v  w ‡ w  v with v. so x ˆ 0. v n be a basis of V.

19. j Hence ÷ A ( g) ˆ 1(÷ 2 ( g) À ÷( g 2 ))X 2 Also. i € i. j < n) are linearly independent elements of A(V  V ). and 2 ÷ A ( g) ˆ 1(÷ 2 ( g) À ÷( g 2 ))X 2 Proof By Proposition 9.11 we can choose a basis e1 .13(2). ˆ ÷ A ( g) ˆ ëi ë j X Now e i g ˆ € so ÷( g) ˆ i ë i and ÷( g2 ) ˆ i ë2 . which implies that ÷ S ( g) ˆ ÷ 2 ( g) À ÷ A ( g) ˆ 1(÷ 2 ( g) ‡ ÷( g 2 ))X 2 j .12. Therefore i ˆ 2 ˆ ÷ 2 ( g) ˆ (÷( g))2 ˆ ëi ‡ 2 ë i ë j ˆ ÷( g 2 ) ‡ 2÷ A ( g)X 2 ë2 ei. and ÷ A to be the character of the CG-module A(V  V ). j dim A(V  V ) > n(n À 1)a2X De®ne ÷ S to be the character of the CG-module S(V  V ). . and the vectors v i  v j À v j  v i (1 < i . we have ÷ S ( g) ˆ 1(÷ 2 ( g) ‡ ÷( g 2 )). . ÷2 ˆ ÷S ‡ ÷ AX The next result gives the values of the characters ÷ S and ÷ A . j i i. Hence dim S(V  V ) > n(n ‡ 1)a2. en of V such that e i g ˆ ë i e i (1 < i < n) for some complex numbers ë i . By Proposition 19. . and the result follows.14 Proposition For g P G. . Then (ei  ej À ej  ei ) g ˆ ë i ë j (ei  ej À ej  ei ). By Proposition 19. and hence from Proposition 19. dim S(V  V ) ‡ dim A(V  V ) ˆ dim V  V ˆ n2 X Hence the above inequalities are equalities. ÷ 2 ˆ ÷ S ‡ ÷ A .12.198 Representations and characters of groups Proof Clearly the vectors v i  v j ‡ v j  v i (1 < i < j < n) are linearly independent elements of S(V  V ).

14. The character table of G is given in Example 19. ÷2 .15 Example Let G ˆ S4 .13. The values of ÷.8. The techniques which we have developed so far give a useful method for ®nding new irreducible characters of a group. and use inner products to analyse ÷ S and ÷ A for new irreducible characters. obtained by lifting the irreducible characters of GaG9. conjugacy class sizes and centralizer orders |CG ( gi )| as follows: gi Class size |CG ( gi )| 1 1 120 (1 2) 10 12 (1 2 3) (1 2)(3 4) (1 2 3 4) (1 2 3)(4 5) (1 2 3 4 5) 20 15 30 20 24 6 8 4 6 5 Thus G has exactly seven irreducible characters.16 Example The character table of S5 Let G ˆ S5 . appear below. then form ø S and ø A and repeat. We have . given one or two irreducible characters to start with. We illustrate this strategy with two examples. The strategy is simple: (1) Given a character ÷. Let ÷ ˆ ÷4 . and the values of ÷ S and ÷ A . gi |CG ( gi )| ÷ ÷S ÷A 1 24 3 6 3 (1 2) 4 1 2 À1 (1 2 3) 3 0 0 0 (1 2)(3 4) 8 À1 2 À1 (1 2 3 4) 4 À1 0 1 We ®nd that ÷ S ˆ ÷1 ‡ ÷3 ‡ ÷4 and ÷ A ˆ ÷5 . G has conjugacy class representatives gi . (2) If ø is a new character found in (1).16(4). 19. G9 ˆ A5 and G has exactly two linear characters ÷1 . By Example 12. (a) Linear characters By Example 17. given by Proposition 19.Tensor products 199 19. the symmetric group of degree 5. form ÷ S and ÷ A .

200 Representations and characters of groups & ÷2 ( g) ˆ ÷1 ˆ 1 G . Proposition 17. . Write ÷ ˆ ÷3 . Next.14 shows that ÷4 ˆ ÷3 ÷2 is also an irreducible character.20. by Theorem 14.24 gives us a character ÷3 (b) Permutation character of G with values Observe that h÷3 . if g is an even permutation. At this point we have the following portion of the character table of G: gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 1 120 1 1 4 4 (1 2) 12 1 À1 2 À2 (1 2 3) (1 2)(3 4) (1 2 3 4) (1 2 3)(4 5) (1 2 3 4 5) 6 8 4 6 5 1 1 1 1 1 1 0 0 1 À1 0 0 1 À1 À1 1 1 1 À1 À1 (c) Tensor products We now use tensor products to construct the last three irreducible characters of G. ÷3 i ˆ ÷3 ( g) ˆ jfix ( g)j À 1 ( g P G)X 42 22 12 (À1)2 (À1)2 ‡ ‡ ‡ ‡ ˆ 1X 120 12 6 6 5 Hence ÷3 is irreducible. if g is an odd permutationX Proposition 13.14 the values of the characters ÷ S and ÷ A are gi |CG ( gi )| ÷S ÷A 1 120 10 6 (1 2) 12 4 0 (1 2 3) (1 2)(3 4) (1 2 3 4) (1 2 3)(4 5) (1 2 3 4 5) 6 8 4 6 5 1 0 2 À2 0 0 1 0 0 1 Thus. and 1. À1. By Proposition 19.

÷ S ˆ ÷1 ‡ ÷3 ‡ ø. ÷3 i ˆ ‡ ‡ À ˆ 1. which we call ÷5 . For ease of printing. Character table of S5 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 1 120 1 1 4 4 6 5 5 (1 2) 12 1 À1 2 À2 0 1 À1 (1 2 3) (1 2)(3 4) (1 2 3 4) (1 2 3)(4 5) (1 2 3 4 5) 6 8 4 6 5 1 1 1 1 0 À1 À1 1 1 0 0 À2 1 1 1 À1 0 0 0 À1 1 1 À1 À1 1 0 1 À1 1 1 À1 À1 1 0 0 19. of order 720. We have now found all seven irreducible characters of S5 . and 120 12 6 6 100 16 1 4 1 h÷ S . The character table of S5 is as shown. ÷1 i ˆ Therefore. Next. ÷ A i ˆ 36 4 1 ‡ ‡ ˆ 1. Finally. 120 8 5 201 and so ÷ A is a new irreducible character. 120 12 6 8 6 40 8 1 1 h÷ S . so that ÷6 ˆ ÷ S À ÷1 À ÷3 . 10 4 1 2 1 ‡ ‡ ‡ ‡ ˆ 1. Using . we then ®nd the last three irreducible characters by using the orthogonality relations. where ø is an irreducible character of degree 5. we use techniques similar to those of the previous example to ®nd 8 of the 11 irreducible characters of the symmetric group S6 .Tensor products h÷ A . ÷7 ˆ ÷6 ÷2 is a different irreducible character of degree 5. 120 12 6 8 6 h÷ S . Let G ˆ S6 . Let ÷6 ˆ ø. it is convenient to label each conjugacy class by the cycle-shape of its elements. ÷ S i ˆ ‡ ‡ ‡ ‡ ˆ 3.17 Example The character table of S6 In this example.

The values of ÷.2. 1. (a) Linear characters As with all symmetric groups Sn (n > 2). it has 11 irreducible characters.3) (4. ÷ S and ÷ A are as follows: Class |CG ( gi )| ÷ ˆ ÷3 ÷S ÷A (1) (2) 720 48 5 15 10 3 7 2 (3) 18 2 3 1 (2.2) (6) 120 144 15 40 90 120 6 5 48 18 8 6 Since G has 11 conjugacy classes.202 Representations and characters of groups this notation. (b) Permutation character and tensor products is given by ÷3 ( g) ˆ jfix ( g)j À 1 The function ÷3 which ÷1 ˆ 1 G . if g is even. ÷ S i ˆ 3X h÷ A . ÷ A i ˆ 1. where & ÷2 ( g) ˆ (see Example 17. by Proposition 13. h÷ S .13).2) 6 0 1 À1 (5) 5 0 0 0 (2.3) (4.24. À1. Let ÷ ˆ ÷3 . the derived subgroup is An .2) 16 1 3 À2 (4) 8 1 1 0 (3.2) 45 16 (4) 90 8 (3. h÷ S .2. the conjugacy class sizes and centralizer orders are as follows (see Exercise 12. ÷1 i ˆ 1.3): Cycle-shape Class size |CG ( gi )| (1) (2) 1 15 720 48 (3) 40 18 (2. if g is odd ( g P G) is a character of G. and we get exactly two linear characters ÷1 and ÷2 . h÷ S .2) (5) (2. .2) (6) 48 18 8 6 À1 3 À2 À1 0 1 À1 1 0 À1 0 1 We calculate that h÷3 . ÷3 i ˆ 1.2) (3. ÷3 i ˆ 1.2) (3.

. . of degree 9. so that we can guarantee that the solutions to the equations which we deal with are integers. . Finally.10).2) (6) 48 18 8 6 1 À1 À1 1 À2 2 3 À3 1 1 À1 À1 1 1 0 0 1 1 À1 À1 0 0 1 1 1 À1 À1 1 1 À1 0 0 (c) Orthogonality relations We now use the column orthogonality relations to complete the character table of G. Thus the conjugacy classes of s and t correspond to the second and fourth columns of the character table.Tensor products 203 Therefore ÷3 is irreducible. ÷8 ˆ ÷7 ÷2 is also irreducible. so is ÷4 ˆ ÷3 ÷2 . we call the three irreducible characters of G which have yet to be found ÷9 .2. Class |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 ÷8 (1) (2) 720 48 1 1 5 5 10 10 9 9 1 À1 3 À3 2 À2 3 À3 (3) 18 1 1 2 2 1 1 0 0 (2. ÷ S ˆ ÷1 ‡ ÷3 ‡ ÷7 . ÷5 ˆ ÷ A is irreducible. . as is ÷6 ˆ ÷5 ÷2 .2) 16 1 1 1 1 À2 À2 1 1 (4) 8 1 À1 1 À1 0 0 À1 1 (3.2) (3.2) 6 1 À1 0 0 À1 1 0 0 (5) 5 1 1 0 0 0 0 À1 À1 (2. Further. From Corollary 13.3) (4.10 we know that ÷(s) and ÷(t) are integers for all characters ÷ of G. Let s denote the permutation (1 2) and t denote the permutation (1 2)(3 4). Also. Ingeniously. but for the moment we know for certain only that ÷( g) is an integer if g2 ˆ 1 (see Corollary 13. in the ordering which we have adopted. It is therefore convenient ®rst to concentrate on elements of order 2. ÷10 and ÷11 . The column orthogonality relations give 11 ˆ iˆ1 ÷ i (s)2 ˆ 48X . ÷8 are recorded in the following portion of the character table of G. where ÷7 is another irreducible character. The irreducible characters ÷1 .16) that all the entries in the character tables of all symmetric groups are integers. respectively. It will be shown later (Corollary 22.

÷8 . 10. That is. and is not equal to any of ÷1 . . Moreover. c. ad ‡ be ‡ cf ˆ 10X The only solution to these equations in integers with a . d 2 ‡ e 2 ‡ f 2 ˆ 2. Therefore.2) d e f The column orthogonality relations give 11 ˆ iˆ1 11 ˆ iˆ1 ÷ i (1)÷ i (s) ˆ 0. ÷10 (s) ˆ À1X The plan is now to ®nd ÷ i (1) and ÷ i (t) for i ˆ 9. f ˆ 0. . 11. a ˆ b ˆ 5X . we see that ÷9 ÷2 is not ÷9 or ÷11 . e. ÷11 (s)2 ˆ 0X Now ÷9 ÷2 is an irreducible character. without loss of generality. that ÷9 (s)2 ˆ ÷10 (s)2 ˆ 1. we aim to evaluate the integers a. . . ÷9 ÷2 ˆ ÷10 X Once more.204 Hence Representations and characters of groups ÷9 (s)2 ‡ ÷10 (s)2 ‡ ÷11 (s)2 ˆ 2X We can assume. d. since ÷9 ÷2 (s) ˆ À÷9 (s). ÷ i (1)÷ i (t) ˆ 0. b. 11 ˆ iˆ1 11 ˆ iˆ1 ÷ i (s)÷ i (t) ˆ 0. 0 and b . ÷ i (t)÷ i (t) ˆ 16. d À e ˆ 0. 0 is d ˆ e ˆ 1. f in the following portion of the character table: Element Class ÷9 ÷10 ÷11 1 (1) a b c s (2) 1 À1 0 t (2. we lose no generality in assuming that ÷9 (s) ˆ 1. whence a À b ˆ 0.

since the column orthogonality relations will give three independent equations in these unknowns (as the above 3 3 3 matrix is invertible). v m .2) 16 1 1 1 1 À2 À2 1 1 1 1 0 (4) 8 1 À1 1 À1 0 0 À1 1 À1 1 0 (3.2) 1 1 0 We can now determine the three unknown entries in each further column.3) (4. Let V be a CG-module. we ®nd that the complete character table of S6 is as shown. .2. . . we ®nd that c ˆ 16 by using the relation 11 ˆ iˆ1 205 ÷ i (1)2 ˆ 720X The above portion of the character table is therefore Element Class ÷9 ÷10 ÷11 1 (1) 5 5 16 s (2) 1 À1 0 t (2. Character table of S6 Class |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 ÷8 ÷9 ÷10 ÷11 (1) (2) 720 48 1 1 5 5 10 10 9 9 5 5 16 1 À1 3 À3 2 À2 3 À3 1 À1 0 (3) 18 1 1 2 2 1 1 0 0 À1 À1 À2 (2. .Tensor products Finally. Having done these calculations.2) 6 1 À1 0 0 À1 1 0 0 1 À1 0 (5) 5 1 1 0 0 0 0 À1 À1 0 0 1 (2. with basis v1 .2) 48 1 À1 À1 1 À2 2 3 À3 À3 3 0 (3. and let W be a . given the character tables of G and H.2) (6) 18 8 6 1 1 À1 À1 1 1 0 0 2 2 À2 1 1 À1 À1 0 0 1 1 À1 À1 0 1 À1 À1 1 1 À1 0 0 0 0 0 Direct products We conclude the chapter by showing that tensor products can be used to determine the character table of a direct product G 3 H.

h) ˆ v g  wh. Then a proof similar to that of Proposition 19. . ÷ k i G hø j . y) ˆ (x À1 gx. h P H. l. . w P W. for ë ij P C. . j As in Proposition 19.5 shows that V  W is a C(G 3 H)-module. . wn. 1 < j < b)X Proof For all i. h P H)X 19. de®ne (v i  wj )( g. k. h) ˆ ÷( g)ø(h) ( g P G.18 Theorem Let ÷1 .206 Representations and characters of groups C H-module. ø l i H ˆ ä ik ä jl X (Here the subscripts G 3 H. with basis w1. y P H. the character of V  W is ÷ 3 ø. we ®nd that (v  w)( g. . Next. h) ˆ ë ij (v i g  wj h)X i. For all i. h)(x. . h÷ i 3 ø j . G and H indicate inner products of characters of G 3 H. for all v P V. 2ˆ 3 ˆ ë ij (v i  wj ) ( g. . respectively. ÷ k 3 ÷ l i G3 H ˆ ˆ 1 ÷ i ( g)ø j (h)÷ k ( g)ø l (h) jG 3 Hj gPG 2 ˆ 3 32 1 ˆ 1 ˆ ÷ i ( g)÷ k ( g) ø j (h)ø l (h) jGj gPG j Hj hP H hP H ˆ h÷ i . Then G 3 H has precisely ab distinct irreducible characters. that is. . and these are ÷ i 3 ø j (1 < i < a. By the proof of Proposition 19. j. G and H. j with 1 < i < m and 1 < j < n. .6. . where (÷ 3 ø)( g. y)À1 ( g. ø b be the distinct irreducible characters of H. note that for all g. and all g P G. we have (x. j i. x P G and h.) Thus the ab characters ÷ i 3 ø j are distinct and irreducible. . Let ÷ be the character of V and ø be the character of W. h) ˆ v i g  wj h and extend this de®nition linearly to the whole of V  W.4. y À1 hy)X . . ÷ a be the distinct irreducible characters of G and let ø1 .

18. h) and ( g9.3. the character table of S3 3 C2 is as shown. Character table of S3 gi |CG ( gi )| ÷1 ÷2 ÷3 1 6 1 1 2 (1 2) 2 1 À1 0 (1 2 3) 3 1 1 À1 Character table of C2 hi |CH (hi )| ø1 ø2 1 2 1 1 À1 2 1 À1 The conjugacy classes of S3 3 C2 are represented by (1. . h9) of G 3 H are conjugate if and only if the elements g and g9 are conjugate in G and the elements h and h9 are conjugate in H. . ga are representatives of the conjugacy classes of G and h1 . . Character table of S3 3 C2 ( gi . In particular. if g1 . 1) (1. hj ) (1 < i < a. À1) ((1 2 3). ((1 2). G 3 H has precisely ab conjugacy classes. 1) ((1 2 3). À1). and by Theorem 19. hj )| ÷1 ÷2 ÷3 ÷1 ÷2 ÷3 3 3 3 3 3 3 ø1 ø1 ø1 ø2 ø2 ø2 (1. 1 < j < b) are representatives of the conjugacy classes of G 3 H. 1) ((1 2).3(1). then the elements ( g i . hb are representatives of the conjugacy classes of H. ((1 2 3). (1. 1). hj ) |CG3 H ( g i . À1) 12 4 6 12 4 6 1 1 2 1 1 2 1 À1 0 1 À1 0 1 1 À1 1 1 À1 1 1 2 À1 À1 À2 1 À1 0 À1 1 0 1 1 À1 À1 À1 1 . À1).Tensor products 207 Hence elements ( g. . . 1). By Theorem 15. . ((1 2). so the irreducible characters ÷ i 3 ø j which we have found must be all the irreducible characters of G 3 H. We reproduce it here. . . alongside the character table of C2 . j 19. 1). Consequently.19 Example The character table of S3 3 C2 The character table of S3 ( D6 ) is given in Example 16. À1) ((1 2). G 3 H has exactly ab irreducible characters. ((1 2 3). À1).

3. 1 G i ˆ 0. 2 ÷ A ( g) ˆ 1(÷ 2 ( g) À ÷( g 2 )) 2 for all g P G. h÷ø. h) ˆ ÷( g)ø(h) for all g P G.5 shows that D12  S3 3 C2 ).10. The values of ÷ 3 ø are given by (÷ 3 ø)( g. where we give the character table of D12 (Exercise 1. if ÷ Tˆ øX 3. Let ÷. ø and ö be characters of the group G.13 of the next chapter we shall show that there exist irreducible characters ÷ and ö of A5 which take the following values: . Show that there is some irreducible character ø of G such that k÷ n . Exercises for Chapter 19 1. 2. h P H. The irreducible characters of G 3 H are those characters ÷ 3 ø. øl ˆ 0 for all integers n with n > 0. then so are ÷ S and ÷ A . öi ˆ h÷. Prove that & 1.2. where ÷ S ( g) ˆ 1(÷ 2 ( g) ‡ ÷( g 2 )). The product of any two characters of G is a character of G. In Example 20. Let ÷ be a character of G which is not faithful. (This shows that the hypothesis that ÷ is faithful cannot be dropped from Theorem 19. where ÷ is an irreducible character of G and ø is an irreducible character of H. If ÷ is a character of G. Show that h÷ø. ÷öiX 2. øöi ˆ hø. if ÷ ˆ ø. Summary of Chapter 19 1. Suppose that ÷ and ø are irreducible characters of G.) 4.208 Representations and characters of groups Compare the solution to Exercise 18.

Find ÷ S and ÷ A . g2 . 5. . and show that both are irreducible. g1 . .13. g5 . g2 . further. Write down the character table of D6 3 D6. ÷ A . g2 are con1 2 3 4 5 6 7 jugate to g1 .Tensor products 1 ÷ ö 5 3 (1 2 3) À1 0 (1 2)(3 4) 1 À1 (1 2 3 4 5) (1 ‡ 0 p 5)a2 (1 3 4 5 2) (1 À 0 p 5)a2 209 Calculate the values of ÷ S . respectively. g4 . g2 . . . g2 . 6. g7 . . g4 . A certain group G or order 24 has precisely seven conjugacy classes with representatives g1 . Express these characters as linear combinations of the irreducible characters ø1 . g2 . g2 . . . g5 . . . G has a character ÷ with values as follows: gi |CG ( gi )| ÷ g1 24 2 g2 24 À2 g3 4 0 g4 6 Àù2 g5 6 Àù g6 6 ù g7 6 ù2 where ù ˆ e2ðia3 . Moreover. ö S and ö A . ®nd the character table of G. By forming products of the irreducible characters found so far. g2 . ø5 of A5 which are given in Example 20.

5(1). h P G.1 Example Let G ˆ D8 ˆ ka. Restriction Let H be a subgroup of the ®nite group G. b: a4 ˆ b2 ˆ 1. let V be the CG-module with basis v1 . v1 b ˆ v1 . We write this character of H as ÷ 5 H. and refer to it as the restriction of ÷ to H. As in Example 4. and illustrate its use. The character of V 5 H is obtained from the character ÷ of V by evaluating ÷ on the elements of H only. We apply this result in the situation where H is of index 2 in G. Then C H is a subset of CG.20 Restriction to a subgroup In this chapter and the next. then f 5 H denotes the restriction of f to H (so that ( f 5 H)(h) ˆ f (h) for all h P H). The case where H is a normal subgroup of G is of particular interest. which occurs. If V is a CG-module. bÀ1 ab ˆ aÀ1 l. v2 a ˆ Àv1 . for example. and Clifford's Theorem 20. If V is a CG-module then we write the corresponding C H-module as V 5 H. when G ˆ S n and H ˆ A n .8 gives important information in this case. then V is also a C H-module. Here. 20. and call it the restriction of V to H. v2 b ˆ Àv2 X 210 . h P H if they hold for all g. This simple way of converting a CG-module into a C H-module is known as restricting from G to H. since properties (1)±(5) of De®nition 4. we introduce the elementary idea of restricting a CG-module to a subgroup H of G.2 certainly hold for all g. More generally. v2 for which v1 a ˆ v2 . we are going to look at ways of relating the representations of a group to the representations of its subgroups. if f: G 3 C is any function.

then dim V ˆ dim (V 5 H). The character ÷ of V is given by g ÷( g) 1 2 a 0 a2 À2 a3 0 b 0 ab 0 a2 b 0 a3 b 0 v1 b ˆ v1 . a2 . if V 5 H is an irreducible C H-module then V is an irreducible CG-module. then V 5 H is the C Hmodule with basis v1 .Restriction to a subgroup 211 If H is the subgroup {1.2 Example Let G ˆ S5 and let H be the subgroup A4 of G consisting of all even permutations of {1. b. On the other hand.1 illustrates this fact. v2 b ˆ Àv2 X and the character ÷ 5 H of V 5 H is given by h ÷(h) 1 2 a2 À2 b 0 a2 b 0 If V is a CG-module and H is a subgroup of G. for if U is a CG-submodule of V. By 18. 4} ®xing 5. the character table of H is gi jC H ( g i )j ø1 ø2 ø3 ø4 1 12 1 1 1 3 (1 2)(3 4) 4 1 1 1 À1 (1 2 3) 3 1 ù ù2 0 (1 3 2) 3 1 ù2 ù 0 . 20. v2 for which v1 a2 ˆ Àv1 . 2. However.2. 3. it might be the case that V is an irreducible CG-module while V 5 H is not an irreducible C H-module. v2 a2 ˆ Àv2 . Example 20. then U 5 H is a C H-submodule of V 5 H. a2 b} of G.

÷7 . if W1 and W2 are functions from G to C. l H is the inner product on the vector space of functions from H to C. ÷5 5 H and ÷6 5 H. we calculate the character ÷ i 5 H as a sum of irreducible characters ø j . ÷3 5 H. ÷5 5 H ˆ 2ø4 . l G is the inner product on the vector space of functions from G to C which we have de®ned earlier.3 De®nitions The inner product k . ÷3 5 H ˆ ÷4 5 H and ÷6 5 H ˆ ÷7 5 HX Therefore we need only consider ÷1 5 H. W2 i G ˆ W1 ( g)W2 ( g). and k . From Example 19.212 Representations and characters of groups (where ù ˆ e2ðia3 ). de®ned similarly. . with irreducible characters labelled ÷1 . These have values 1 ÷1 5 ÷3 5 ÷5 5 ÷6 5 H H H H 1 4 6 5 (1 2)(3 4) 1 0 À2 1 (1 2 3) (1 3 2) 1 1 0 À1 1 1 0 À1 It is easy to spot that ÷ 1 5 H ˆ ø1 . .16 we see that ÷1 5 H ˆ ÷2 5 H. ÷ 6 5 H ˆ ø2 ‡ ø 3 ‡ ø 4 X Constituents of a restricted character To help us discuss the way in which a restricted character ÷ 5 H can be expressed in terms of the irreducible characters of H. . The character table of G is given in Example 19. we introduce the following notation. jGj gPG . For each i with 1 < i < 7. ÷ 3 5 H ˆ ø1 ‡ ø 4 .16. 20. . Thus. then 1 ˆ hW1 .

Recall from Theorem 13. øi H X j Hj iˆ1 j Therefore h÷ i 5 H.4 Proposition Let H be a subgroup of G and let ø be a non-zero character of H. . ø i i H X We say that ø i is a constituent of ÷ 5 H if the coef®cient d i in the above expression is non-zero. ÷ 5 H ˆ d 1 ø1 ‡ X X X ‡ d r ø r for some non-negative integers d 1 . iˆ1 Now 0 Tˆ k ˆ jGj ø(1) ˆ h÷reg 5 H. then by Theorem 14.Restriction to a subgroup and if ö1 and ö2 are functions from H to C. X X X . . . øi H Tˆ 0X Proof Let ÷1 . ÷reg ( g) ˆ and ÷reg ˆ ÷ i (1)÷ i X 0 if g Tˆ 1. ÷ k be the irreducible characters of G. The next proposition shows that every irreducible character of H is a constituent of the restriction of some irreducible character of G. øi H ˆ ÷ i (1)h÷ i 5 H.4. .20 that the regular character ÷reg of G satis®es & k ˆ jGj if g ˆ 1. 20. Then there exists an irreducible character ÷ of G such that h÷ 5 H.17. . it may be very dif®cult in practice to write down the restrictions ÷ 5 H in terms of irreducible characters of H. d r which are given by d i ˆ h÷ 5 H. ö2 i H ˆ ö1 (h)ö2 (h)X j Hj hP H 213 If ÷ is a character of G and ø1 . . Unfortunately. ø r are the irreducible characters of the subgroup H of G. we could hope to ®nd the character table of the subgroup H by restricting the irreducible characters ÷ of G to H.19 and Proposition 13. then 1 ˆ hö1 . . The best . øi H Tˆ 0 for some i. . Suppose that we know the character table of G. In the light of Proposition 20.

and let ø1 . jGj iˆ1 i € where K ˆ (1ajGj) ga H ÷( g)÷( g)X Now K > 0. 1 ˆ h÷. j We can say more about the constituents of ÷ 5 H in the case where H is a normal subgroup of G. since ÷ is irreducible.214 Representations and characters of groups chance of doing this occurs when the index jG: Hj(ˆ jGjaj Hj) is small. where the non-negative integers d 1 .5 Proposition Let H be a subgroup of G.6) if and only if ÷( g) ˆ 0 for all elements g of G which lie outside H. The conclusions of the proposition a follow at once. we have equality in (20.17. X X X . ø r be the irreducible characters of H. we will see that all the constituents of ÷ 5 H have the same degree. For example. ÷i G ˆ ˆ ˆ 1 ˆ ÷( g)÷( g) jGj gPG 1 ˆ ÷(h)÷(h) ‡ K jGj hP H r j Hj ˆ 2 d ‡ K. let ÷ be an irreducible character of G. 20. ÷ 5 Hi H ˆ i 1 ˆ ÷(h)÷(h)X j Hj hP H Also. and K ˆ 0 if and only P if ÷( g) ˆ 0 for all g with g P H. . Proof By Theorem 14. F F F . The way to exploit the fact that H v G us revealed in the following proposition. Then ÷ 5 H ˆ d 1 ø1 ‡ X X X ‡ d r ø r . d r satisfy (20X6) r ˆ iˆ1 d 2 < jG: HjX i Moreover. since the restrictions ÷ 5 H then do not have many constituents. we have r ˆ iˆ1 d 2 ˆ h÷ 5 H. as the following result shows.

Further. since U is irreducible. (2) As a C H-module. (3) If g1 . è is a C H-isomorphism. and the proof of the proposition is complete.12. and (wgh)è ˆ (wh9 g)è ˆ (wh9ö) g ˆ (wö)h9 g ˆ (wö) gh ˆ (wgè)hX Therefore. Suppose that h P H. so dim U ˆ dim Ug. u 3 ug (u P U) is an invertible linear transformation from U to Ug.7 Proposition Suppose that H v G. as claimed.Restriction to a subgroup 215 20. then WgÀ1 is a C H-submodule of U. For every g P G let Ug ˆ fug: u P U g. proving that Ug is a C H-submodule of V 5 H. Then gh ˆ h9 g for some h9 P H. WgÀ1 ˆ {0} or U. De®ne è : Ug1 g 3 Ug2 g by è : wg 3 (wö) g (w P Ug 1 )X Then è is clearly an isomorphism of vector spaces. Proof (1) Clearly. V is a direct sum of some of the C Hmodules Ug. g P G and Ug1 and Ug2 are isomorphic C H-modules. Then (1) The set Ug is an irreducible C H-submodule of V 5 H and dim Ug ˆ dim U . and since H v G we have ghgÀ1 P H for all h P H. we have ˆ Vˆ UgX gPG Then by Proposition 7. Ug is an irreducible C H-module. Therefore. since V is irreducible. Therefore. V is a direct sum of some of the C Hmodules Ug. then Ug1 g and Ug2 g are isomorphic C H-modules. Moreover. (2) The sum of all the subspaces Ug with g P G is a CG-submodule of V. so (ug)h ˆ u( ghg À1 ) g P Ug (u P U). (3) Now let ö be a C H-isomorphism from Ug1 to Ug2 . j . if W is a C Hsubmodule of Ug. Ug is a subspace of V. whence W ˆ {0} or Ug. Let V be an irreducible CG-module and U be an irreducible C H-submodule of V 5 H. g2 .

Proof Let V be a CG-module with character ÷. each having character ø1 . Then V contains a C H-module X1 whose character is eø1 . if g P G then X1 g is a direct sum of isomorphic C H-modules. . Then (1) all the constituents of ÷ 5 H have the same degree. by Proposition 20.14 in Chapter 22 to see an advanced use of the theorem.216 Representations and characters of groups We now come the fundamental theorem on the restriction of a character to a normal subgroup. . parts (1) and (2).7(2). Therefore. but you might like to look at Corollary 22. and which is therefore a direct sum of e isomorphic C H-modules. say X 1 ˆ U1 È X X X È U e X By Proposition 20. and Xi T X j if i Tˆ j. ø1 i. 20. and (2) if ø1 . . Hence V has the form V ˆ X1 È X X X È X m where each Xi is a direct sum of e isomorphic C H-modules. ÷ 5 H ˆ e(ø1 ‡ X X X ‡ ø m )X j Our main applications of Clifford's Theorem will concern the case where jG: Hj ˆ 2. then ÷ 5 H ˆ e(ø1 ‡ X X X ‡ ø m ) for some positive integer e. V is a sum of C Hmodules of the form X1 g.7. Normal subgroups of index 2 We are shortly going to give more precise information about the constituents of ÷ 5 H when H is a normal subgroup of G of index 2 (that is. Examples where this happens are G ˆ Sn . Then it follows from Proposition 20. On the other hand. Let e ˆ h÷ 5 H. .8 Clifford's Theorem Suppose that H v G and that ÷ is an irreducible character of G. jG: Hj ˆ 2). that all the constituents of ÷ 5 H have the same degree. .7(3). ø m are the constituents of ÷ 5 H.

2 where iˆ1 d i < 2. j with i Tˆ j. then by Proposition 20. if H is a subgroup of index 2 in G. When H is a normal subgroup of index 2 in G. We describe this relationship in (20. by Clifford's Theorem 20. Also. and let ÷ be an irreducible character of G.16). Proof If ø1 . ÷ and ÷ë are irreducible characters of the same degree (see Proposition 17. ø r are the irreducible characters of H. it is often desirable to have more details about the two cases in Proposition 20. we deduce that either ÷ 5 H ˆ ø i for some i. or (2) ÷ 5 H is the sum of two distinct irreducible characters of H of the same degree. and we shall supply these next.13) below.9 Proposition Suppose that H is a normal subgroup of index 2 in G. In fact. . then H must be normal in G (see Exercise 1. . X X X .10 Proposition Suppose that H is a normal subgroup of index 2 in G. since ë(h) ˆ 1 for all h P H. and that ÷ is . H ˆ hai. . 20. b: a n ˆ b2 ˆ 1.8 j €r ÷ 5 H ˆ d 1 ø1 ‡ X X X ‡ d r ø r . ø i and ø j have the same degree. Since d 1 . ë( g) ˆ À1 if g P HX a Note that for all irreducible characters ÷ of G. d r are non-negative integers.14). 20. and we then illustrate the results by ®nding the character table of A5 from that of S5 (which we have already obtained in Example 19. or ÷ 5 H ˆ ø i ‡ ø j for some i.5. For practical purposes.10). .9. the character tables of G and H are closely related. ÷ 5 H ˆ ÷ë 5 H. we may lift the non-trivial linear character of GaH to obtain a linear character ë of G which satis®es @ 1 if g P H. Since GaH  C2 . In the latter case. bÀ1 ab ˆ aÀ1 i. or G ˆ D2 n ˆ ha.Restriction to a subgroup 217 H ˆ A n . Then either (1) ÷ 5 H is irreducible.

and ÷ is an irreducible character of G. ÷ 5 H is irreducible if and only if the inequality in (20.5. ÷ë( g) ˆ À÷( g) if g P H. 20. observe that @ ÷( g) if g P H.9. and this happens if and only if ÷( g) Tˆ 0 for some g P G with g P H. Then the following three conditions are equivalent: (1) ÷ 5 H is irreducible.6) is strict. If ö is an irreducible character of G which satis®es ö 5 H ˆ ÷ 5 H. if g P HX a . To see that (2) is equivalent to (3). Proof We use Proposition 20. In the next proposition we consider the ®rst possibility.218 Representations and characters of groups an irreducible character of G. a (3) the characters ÷ and ÷ë of G are not equal. a so ÷( g) Tˆ 0 for some g with g P H if and only if ÷ë Tˆ ÷. Thus (1) is equivalent to a (2). (2) ÷( g) Tˆ 0 for some g P G with g P H. since jG: Hj ˆ 2. then ÷ 5 H is the sum of one or two irreducible characters of H. if H is a normal subgroup of G of index 2. and that ÷ is an irreducible character of G for which ÷ 5 H is irreducible.11 Proposition Suppose that H is a normal subgroup of index 2 in G. Proof We have (÷ ‡ ÷ë)( g) ˆ Hence @ 2÷( g) 0 if g P H. a j According to Proposition 20. then either ö ˆ ÷ or ö ˆ ÷ë.

÷( g) ˆ 0 for all g with g P H. so kö. ÷i G ˆ ö( g)÷( g) ˆ ö( g)÷( g) jGj gPG jGj gP H ˆ 1hö 5 H. ö 5 Hi H X Now k÷ 5 H.10. Such characters of G occur in pairs (÷ and ÷ë) which have the same restriction to H (Propositions 20. G. a Therefore. (20.10. since ÷ 5 H is irreducible and ö 5 H ˆ ÷ 5 H. and so either ö ˆ ÷ or j ö ˆ ÷ë. and hence ö ˆ ÷. then hö 5 H. 1 ˆ 1 ˆ hö. ö 5 Hl H ˆ 1. ÷ 5 Hi H X 2 If ö 5 H has ø1 or ø2 as a constituent. .13) Let H be a normal subgroup of index 2 in the group G. say ÷ 5 H ˆ ø1 ‡ ø2. (1) Each irreducible character ÷ of G which is nonzero somewhere outside H restricts to be an irreducible character of H. j We summarize our results on subgroups of index 2. 20. ÷l G Tˆ 0. by explaining how to list the irreducible characters of H on the assumption that we know the character table of G. and that ÷ is the sum of two ö is an irreducias a constituent. 20. Finally. Therefore k÷ ‡ ÷ë. Proof In view of Proposition 20. öl G ˆ 1. öi G ˆ ˆ 1 ˆ 2÷( g)ö( g) jGj gP H 1 ˆ ÷( g)ö( g) j Hj gP H 219 ˆ h÷ 5 H.Restriction to a subgroup h÷ ‡ ÷ë. ÷ 5 Hi H Tˆ 0. If ble character of G such that ö 5 H has ø1 or ø2 then ö ˆ ÷.12 Proposition Suppose that H is a normal subgroup of index 2 in an irreducible character of G for which ÷ 5 H is irreducible characters of H.11). we look at the case where ÷ 5 H is reducible.

.9. They can also be veri®ed by calculating inner products. and note that H is a normal subgroup of index 2 in the group S5 . a so by (20.13)(2). Observe that ÷1 . ÷3 and ÷6 are non-zero somewhere outside H.12). Call them ø1 . .220 Representations and characters of groups (2) If ÷ is an irreducible character of G which is zero everywhere outside H. . ÷3 5 H and ÷6 5 H are irreducible characters of H. The two characters of H which we get from ÷ in this way come from no other irreducible character of G (Propositions 20. ÷5 5 H ˆ ø4 ‡ ø5 where ø4 and ø5 are distinct irreducible characters of H of degree 3. ÷7 of S5 can be found in Example 19.13)(1). Note that ÷2 5 H ˆ ÷1 5 H. . We have established that the character table of H is gi |CG ( g i )| ø1 ø2 ø3 ø4 ø5 1 60 1 4 5 3 3 (1 2 3) 3 1 1 À1 á2 â2 (1 2)(3 4) 4 1 0 1 á3 â3 (1 2 3 4 5) 5 1 À1 0 á4 â4 (1 3 4 5 2) 5 1 À1 0 á5 â5 . (3) Every irreducible character of H appears among those obtained by restricting irreducible characters of G. . The conjugacy classes of H and their sizes are given in Example 12. . ÷1 5 H. 20. In case (2) of (20. so by (20. then ÷ restricts to be the sum of two distinct irreducible characters of H of the same degree. ÷4 5 H ˆ ÷3 5 H and ÷7 5 H ˆ ÷6 5 H. in practice case (1) occurs more frequently than case (2). and the irreducible characters ÷1 .13). . and hence ø1 . 20. ø5 are the distinct irreducible characters of H by (20. Fortunately.18(2).13)(3). as in parts (1) and (2) (Proposition 20.4).13) of facts about characters of subgroups of index 2. 20. ø2 and ø3 .16. The results we have obtained so far have been deduced from our summary (20.14 Example The character table of A5 Write H ˆ A5 . ÷5 ( g) ˆ 0 for all g P H. Also. respectively. extra work is needed to calculate the values taken by the two constituents of ÷ 5 H.10. .

we see that each element of A5 is conjugate to its inverse. we may take p p á4 ˆ 1(1 ‡ 5). á3 ˆ â3 ˆ À1. Hence by Proposition 13. By the orthogonality relation for column i with itself (2 < i < 5). ⠈ 1(1 À 2 p 5). and we ®nd that á4 and â4 are the solutions of the quadratic x 2 À x À 1 ˆ 0. we obtain 3 ˆ 3 ‡ á2 ‡ â2 . Since ø4 Tˆ ø5 .Restriction to a subgroup 221 We use the column orthogonality relations to calculate the unknowns á i and â i .9(4). á5 and â5 are 1(1 Æ 5). we have 2 p p á5 ˆ 1(1 À 5). Since we have not yet distinguished between ø4 and ø5 . . â4 ˆ 1(1 À 5)X 2 2 p Similarly. 3 3 5 ˆ 2 ‡ á2 ‡ â2 ˆ 2 ‡ á2 ‡ â2 X 4 4 5 5 Hence á2 ˆ â2 ˆ 0. â5 ˆ 1(1 ‡ 5)X 2 2 Thus the character table of A5 is as shown. all the numbers in the character table are real. The values of á i ‡ â i for 2 < i < 5 are given by noting that ø4 ‡ ø5 ˆ ÷5 5 H (or by using the column orthogonality relations for column 1 and column i). 2 2 4 ˆ 2 ‡ á2 ‡ â2 . We get á2 ‡ â2 ˆ 0. á3 ‡ â3 ˆ À2. Character table of A5 gi |CG ( g i )| ø1 ø2 ø3 ø4 ø5 1 60 1 4 5 3 3 (1 2 3) 3 1 1 À1 0 0 (1 2)(3 4) 4 1 0 1 À1 À1 (1 2 3 4 5) 5 1 À1 0 á â (1 3 4 5 2) 5 1 À1 0 â á where á ˆ 1(1 ‡ 2 p 5).13. á4 ‡ â4 ˆ á5 ‡ â5 ˆ 1X Using Proposition 12.

X X X . If ÷ is a character of G. ÷ and ÷ 5 H have the same degree. and let ÷ be an irreducible character of G. express ÷ 5 H as a sum of irreducible characters of H. Use the restrictions of the irreducible characters of S6 . then all the constituents of ÷ 5 H have the same degree. Let G ˆ S4 and let H be the subgroup k(1 2 3 4).1). then r ˆ iˆ1 d 2 < jG : HjX i 3. to ®nd the character table of A6 .4. Exercises for Chapter 20 1. The number of irreducible constituents of ÷ 5 H is bounded above by jG : Hj.17. . then ÷ 5 H is a character of H. (The seven conjugacy classes of A6 can be found by consulting the solutions to Exercises 12. if ø1 . ÷ 5 Hi H ˆ 1. 2. Summary of Chapter 20 Assume throughout that H is a subgroup of G. 1.222 Representations and characters of groups Proposition 17.3 and 12. ø r are the irreducible characters of H. In particular. 2. (a) Show that H  D8. 2 or 3X Give examples to show that each possibility can occur. Let G be a group with an abelian subgroup H of index n. The values of ÷ 5 H are given by (÷ 5 H)(h) ˆ ÷(h) for all h P H. Suppose that G is a group with a subgroup H of index 3. (1 3)l of G. 4. given in Example 19. Prove that h÷ 5 H. Prove that ÷(1) < n for every irreducible character ÷ of G. If H v G and ÷ is an irreducible character of G. and ÷ 5 H ˆ d 1 ø1 ‡ X X X ‡ d r ø r. (b) For each irreducible character ÷ of G (given in Section 18.6 allows us to deduce from the character table that A5 is a simple group.) 3. Indeed.

6. 15. 21. 14. 35. 6. 14. . Find the complete list of degrees of the irreducible characters of A7 . 14. 14. 20. 21. A7 has nine conjugacy classes. 1. 35X Also.Restriction to a subgroup 223 5. It is known that the complete list of degrees of the irreducible characters of S7 is 1. 15.

As H is smaller than G. then there exists r P CG such that uW ˆ ru for all u P U X 224 . since for all s P C H. it is usually the case that C H-modules are easier to understand and construct than CG-modules. If W is a C H-homomorphism from U to CG. We shall see many applications of this method in later chapters. C H-homomorphisms and CG-homomorphisms Let U be a C H-submodule of the regular C H-module C H. We now prove the striking fact that every C Hhomomorphism from U to CG has this simple form. and let U be a C H-submodule of C H. Before describing the process of induction. which constructs a CG-module from a given C H-module. then W: u 3 ru (u P U ) de®nes a C H-homomorphism from U to CG. If r P CG. Much more subtle than this is the process of induction. We saw in the last chapter that restriction gives a simple way of converting a CG-module into a C H-module. 21.1 Proposition Assume that H < G. so induction can often give us an important handle on the representations of a group if we know some representations of its subgroups.21 Induced modules and characters Throughout this chapter we assume that H is a subgroup of the ®nite group G. (us)W ˆ rus ˆ (uW)s. and induction is the main concern of this chapter. we require some results which connect C H-homomorphisms with CG-homomorphisms.

2 Corollary Let U be a CG-submodule of CG. For u P U. v P V. 21.1. (2) there exists r P CG such that for all u P U. Then the following two statements are equivalent: (1) U ’ V ˆ {0}. v P V. 21. r(u ‡ v) ˆ uX Then ru ˆ u if u P U. v P V ) is a function. Consequently U ’ V ˆ {0}. De®ne ö: C H 3 CG by ö: u ‡ w 3 uW (u P U. Then the sum U ‡ V is a direct sum. Then every CG-homomorphism from U to CG has the form u 3 ru (u P U ) for some r P CG. and rv ˆ 0 if v P V. If x P U ’ V then rx ˆ x and rx ˆ 0.2. and so x ˆ 0. j . there is a C H-submodule W of C H such that C H ˆ U È W. there exists r P CG such that for all u P U. and so W is of the required form. the ®rst of which is just the case H ˆ G of the proposition.Induced modules and characters 225 Proof By Maschke's Theorem 8. Conversely. assume that for some r P CG we have ru ˆ u and rv ˆ 0 for all u P U.1. j We give two corollaries of Proposition 21. ru ˆ u and rv ˆ 0X Proof Assume that U ’ V ˆ {0}. Therefore by Corollary 21.3 Corollary Let U and V be CG-submodules of CG. v P V. uW ˆ uö ˆ (1u)ö ˆ (1ö)u ˆ ru. moreover. w P W )X Then ö is easily seen to be a C H-homomorphism.11). so u‡v3 u (u P U . Let r ˆ 1ö. it is a CG-homomorphism from U È V to CG (see Proposition 7.

U3 ˆ sp (1 ‡ ù2 a ‡ ùa2 . b ‡ ù2 ab ‡ ùa2 b). b ‡ ù2 ab ‡ ùa2 b). where U1 ˆ sp (1 ‡ a ‡ a2 ‡ b ‡ ab ‡ a2 b). g P GgX Clearly. That is.5 Example Let G ˆ D6 ˆ ka. W 2 4 G ˆ sp (1 ‡ ùa ‡ ù2 a2 . Let ù ˆ e2ðia3 . W 1 4 G ˆ sp (1 ‡ ù2 a ‡ ùa2 . 21.4 De®nition Assume that H is a subgroup of G.8(2) that CG ˆ U1 È U 2 È U3 È U 4 .8(1)). X(CG) is then a CG-submodule of CG. W 2 4 G ˆ U4 X . U2 ˆ sp (1 ‡ a ‡ a2 À b À ab À a2 b). Remember that H is a subgroup of G. Then U 4 G is called the CG-module induced from U. and let U 4 G denote the CG-module U(CG). and de®ne W 0 ˆ sp (1 ‡ a ‡ a2 ). so C H is a subset of CG. W 0 4 G ˆ sp (1 ‡ a ‡ a2 . b ‡ ab ‡ a2 b). W 1 4 G ˆ U3 . W 2 ˆ sp (1 ‡ ùa ‡ ù2 a2 )X These are C H-submodules of C H (see Example 10. a cyclic subgroup of G of order 3. b ‡ ùab ‡ ù2 a2 b)X Thus W 0 4 G ˆ U1 È U2 . Clearly. bÀ1 ab ˆ aÀ1 l. and let H ˆ kal. W 1 ˆ sp (1 ‡ ù2 a ‡ ùa2 ). 21.226 Representations and characters of groups Induction from H to G For any subset X of CG. we write X(CG) for the subspace of CG which is spanned by all the elements xg with x P X. b ‡ ùab ‡ ù2 a2 b)X Recall from Example 10. b: a3 ˆ b2 ˆ 1. Let U be a C H-submodule of C H. a direct sum of irreducible CG-modules U i . g P G. U4 ˆ sp (1 ‡ ùa ‡ ù2 a2 . X (CG) ˆ sp fxg: x P X .

so ra is a linear combination of elements rug. we have sra ˆ a. while W 1 4 G and W 2 4 G are irreducible. 21.6 Proposition Assume that H < G. and hence ra P V 4 G. and also there exists s P CG such that vWÀ1 ˆ sv for all v P V. proving that U 4 G is CG-isomorphic to V 4 G. Then U 4 G is CG-isomorphic to V 4 G. g P G). Moreover. 21. Consequently sru ˆ u and rsv ˆ v for all u P U . there exists r P CG such that uW ˆ ru for all u P U. Proof Let W: U 3 V be a C H-isomorphism. as (aö) g ˆ rag ˆ (ag)ö (a P U 4 G. for all a P U 4 G. ö is a CG-homomorphism. v P V X If a P U 4 G then a is a linear combination of elements ug (u P U.Induced modules and characters 227 In particular. Then (U 4 G) ’ (V 4 G) ˆ f0g. Therefore ö is a CG-isomorphism. j The next proposition and its corollary enable us to de®ne the induced module U 4 G for an arbitrary C H-module U. We now show that isomorphic C H-modules give isomorphic induced CG-modules.7 Proposition Assume that U and V are C H-submodules of C H with U ’ V ˆ {0}. v P V. W 0 4 G is reducible. rsb ˆ b Hence the function b 3 sb (b P V 4 G) is the inverse of ö. Since sru ˆ u and rsv ˆ v for all u P U.1. g P G). Therefore ö: a 3 ra (a P U 4 G) is a function from U 4 G to V 4 G. By Proposition 21. b P V 4 GX . Suppose that U and V are C H-submodules of C H and that U is C H-isomorphic to V.

21. g P G). a direct sum of C H-submodules U i . using (2.3. v P V. this implies that ru9 ˆ u9 and similarly. as required. there exists r P CG such that ru ˆ u and rv ˆ 0 for all u P U. It is trivial for m ˆ 1. 21.7. U 4 G ˆ (U1 4 G) È (V 4 G)X By induction. Then U 4 G ˆ (U1 4 G) È X X X È (U m 4 G)X Proof We prove this by induction on m.228 Representations and characters of groups Proof By Corollary 21. we obtain U 4 G ˆ (U 1 4 G) È X X X È (Um 4 G).8 Corollary Let U be a C H-submodule of C H. and hence. U  U1 È X X X È Um .5). Then for all u P U.10).3. j We can now de®ne the induced module U 4 G for an arbitrary C Hmodule U (where U is not necessarily a C H-submodule of C H). Then (by Theorems 8. v P V and all g P G.7 and 10. and suppose that U ˆ U1 È X X X È Um . The de®nition of U 4 G implies that U 4 G ˆ (U1 4 G) ‡ (V 4 G)X Therefore by Proposition 21. rv9 ˆ 0 for all v9 P V 4 GX j for all u9 P U 4 G. Therefore (U 4 G) ’ (V 4 G) ˆ f0g by Corollary 21.9 De®nition Let U be a C H-module. Now U ˆ U1 È V. where V ˆ U2 È X X X È U m . V 4 G ˆ (U 2 4 G) È X X X È (U m 4 G). rug ˆ ug and rvg ˆ 0X Since U 4 G is spanned by elements of the form ug (u P U.

U 4 K  (U1 4 K) È X X X È (Um 4 K)X Since K < G.11 Theorem Suppose that H and K are subgroups of G such that H < K < G.Induced modules and characters 229 for certain irreducible C H-submodules U i of C H. it follows that (U(CK))(CG) ˆ U(CG). g P G)X (U 4 K) 4 G ˆ U 4 GX U  U1 È X X X È Um for certain irreducible C H-submodules U i of C H. (U(CK))(CG) is spanned by elements of the form (u P U . We emphasize that the de®nition of the induced module U 4 G in the case where U is a C H-submodule of C H is a natural one: U 4 G ˆ U (CG)X We shall always prove results for general induced modules U 4 G by ®rst dealing with the special case where U is a C H-submodule of C H.9) that (21X10) (U1 È X X X È Um ) 4 G  (U1 4 G) È X X X È (U m 4 G)X Our ®rst major result on general induced modules is known as `the transitivity of induction'. If U is a C H-module. De®ne U 4 G to be the following (external) direct sum: U 4 G ˆ (U 1 4 G) È X X X È (Um 4 G)X Proposition 21. k P K)X Therefore. k P K. and then applying the fact (which is immediate from De®nition 21.6 and Corollary 21. Now let U be an arbitrary C H-module. That is. then (U 4 K) 4 G  U 4 GX Proof Assume ®rst that U is a C H-submodule of C H.4. By (21. Then . Then U(CK) is spanned by elements of the form uk ukg (21X12) (u P U .8 ensure that this de®nition is consistent with De®nition 21. 21.10).

The next example illustrates an important relationship between induced characters and restrictions of characters.230 Therefore Representations and characters of groups (U 4 K) 4 G  (U1 4 K) 4 G È X X X È (Um 4 K) 4 G by (21X10) ˆ (U1 4 G) È X X X È (U m 4 G) U4G by Definition 21. and ø1 . ÷ 6 5 H ˆ ø 2 ‡ ø 3 ‡ ø4 .2. . then the character of the induced CG-module U 4 G is denoted by ø 4 G.2) then ÷ 1 5 H ˆ ø1 .9. ÷7 are the irreducible characters of G (given in Example 19. ÷5 5 H ˆ 2ø4 .17. the coef®cients which appear here are the values of h÷ i 5 H.14 Example Let G ˆ S5 and let H be the subgroup A4 of G. ÷ 2 5 H ˆ ø1 . . We showed in that example that if ÷1 . . . as in Example 20. . ÷ 4 5 H ˆ ø1 ‡ ø4 . . j. . We record these coef®cients in a . ÷ 3 5 H ˆ ø1 ‡ ø4 . 21. ÷ 7 5 H ˆ ø 2 ‡ ø 3 ‡ ø4 X By Theorem 14. ø j i H for appropriate i.16). . ø4 are the irreducible characters of H (given in 18. and is called the character induced from ø.13 De®nition If ø is the character of a C H-module U. by (21X12) j Induced characters 21.

the seven integers in column 1 give ø1 4 G ˆ 1 . øi H for all characters ÷ of G and ø of H. ÷6 ‡ 0 . ÷2 ‡ 1 . ø j i H : ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 H ø1 1 f 1 f f 1 f f 1 f f 0 f d 0 0 ø2 0 0 0 0 0 1 1 ø3 0 0 0 0 0 1 1 ø4 I 0 0 g g 1 g g 1 g g 2 g g 1 e 1 231 The rows of this matrix tell us how the irreducible characters of G restrict to H. which is known as the Frobenius Reciprocity Theorem. ø4 X Remarkably. row 3 gives ÷3 5 H ˆ 1 .15 Proposition Assume that H < G. is also equal to h÷ i . V 5 H) have equal dimensions. ÷3 ‡ 1 . V ) and HomC H (U . ÷4 ‡ 0 . and let V be a CG-submodule of CG. we need the following preliminary result. it is true that h÷. ÷1 ‡ 1 . 21. For example. ÷5 ‡ 0 . . ÷7 X Similarly. it turns out that the columns of the matrix tell us how the irreducible characters of H induce to G. Then the vector spaces HomCG (U 4 G. ø j i H . ø3 ‡ 1 . The Frobenius Reciprocity Theorem Before proving the theorem. and ø4 4 G ˆ ÷3 ‡ ÷4 ‡ 2÷5 ‡ ÷6 ‡ ÷7 X Thus the ij-entry in our matrix. which we already know to be equal to h÷ i 5 H. ø2 ‡ 0 . ø2 4 G ˆ ø3 4 G ˆ ÷6 ‡ ÷7 . ø j 4 Gi G .Induced modules and characters matrix whose ij-entry is h÷ i 5 H. Let U be a C H-submodule of C H. In fact. and we devote the next section to a proof of this result. To be precise. ø 4 Gi G ˆ h÷ 5 H. ø1 ‡ 0 .

Then hø 4 G. By Theorem 14. then r1 s ˆ r2 s for all s P U 4 G. V 5 H). Then there is a C H-submodule U of C H which has character ø. Moreover. Finally. V ) to HomC H (U . there is an element r P CG such that sW ˆ rs for all s P U 4 GX De®ne W: U 3 CG to be the restriction of W to U. as required.1. Let ö P HomC H (U .232 Representations and characters of groups Proof Suppose that W P HomCG (U 4 G. Hence the function W 3 W is injective. that is. V 5 H))X .16 The Frobenius Reciprocity Theorem Assume that H < G. there exists r P CG such that uö ˆ ru for all u P U. since s is a linear combination of elements ug with u P U. Thus the function W3W is a linear transformation from HomCG (U 4 G. V 5 H). ö ˆ W. and there is a CG-submodule V of CG which has character ÷. V ). and so it is an invertible linear transformation from HomCG (U 4 G. ÷i G ˆ hø. V ). ÷ 5 Hi H ˆ dim (HomC H (U .2. and hø. ÷ 5 Hi H X Proof First assume that the characters ÷ and ø are irreducible.24. These two vector spaces therefore have the same dimension. ÷i G ˆ dim (HomCG (U 4 G. V 5 H). g P G. De®ne W from U 4 G to CG by sW ˆ rs (s P U 4 G)X Then W P HomCG (U 4 G. V 5 H). note that if r1. j 21. uW ˆ ru for all u P U X Then W P HomC H (U . Let ÷ be a character of G and let ø be a character of H. r2 P CG and r1 u ˆ r2 u for all u P U. Therefore the function W 3 W is surjective. Then by Corollary 21. Then by Proposition 21. V )). V ) to HomC H (U . we have hø 4 G. We shall show that this linear transformation is invertible.

we therefore deduce that (21X17) hø 4 G. the characters of G span the vector space of class functions on G. . then hø 4 G.18 Corollary If f is a class function on G. f 5 Hi H X Proof This follows at once from the Frobenius Reciprocity Theorem. For the general case. . ÷ 5 Hi H 233 in the special case we are considering. ÷i G ˆ ˆ ˆ ˆ k ˆ iˆ1 d i ÷ i and ø ˆ m ˆ jˆ1 ej ø j X C B m ˆ jˆ1 ej ø j 4 G. Then for some integers di . . ÷ i i G ej d i hø j . and for convenience of notation. .Induced modules and characters From Proposition 21. . de®ne the j . 21. ø m be the irreducible characters of H. and ø is a character of H. ÷ k be the irreducible characters of G and let ø1 .15. since by Corollary 15. ÷ 5 Hi H X This completes the proof of the Frobenius Reciprocity Theorem. . k ˆ iˆ1 d i ÷i G m k ˆˆ jˆ1 iˆ1 m k ˆˆ ej d i hø j 4 G. ÷ i 5 Hi H k ˆ iˆ1 by (21X17) B jˆ1 iˆ1 m ˆ jˆ1 C di÷i 5 H H ej ø j . f i G ˆ hø. . namely when ÷ and ø are irreducible. ej we have ÷ˆ Therefore hø 4 G. j The values of induced characters We now show how to evaluate induced characters. . Let ø be a character of the subgroup H of G. ˆ hø. let ÷1 . ÷i G ˆ hø.4.

Let ÷ be an irreducible character of G. Proof Let f: G 3 C be the function given by 1 ˆ • f ( g) ˆ ø( y À1 gy) ( g P G)X j Hj yPG We aim to prove that f ˆ ø 4 G. it is suf®cient to show that h f . and so by Corollary 15. ÷i G for all irreducible characters ÷ of G. If w P G then 1 ˆ • f (w À1 gw) ˆ ø( y À1 w À1 gwy) ˆ f ( g) j Hj yPG since wy runs through G as y runs through G. ÷i G ˆ hø 4 G. Then 1 ˆ h f . ÷i G ˆ ˆ 1 1 ˆˆ • ø(x)÷( yxy À1 )X jGj j Hj xPG yPG 1 ˆ ø(x)÷(x) j Hj xP H 1 1 ˆˆ • ø( y À1 gy)÷( g)X jGj j Hj gPG yPG . Then h f . Therefore f is a class function. ÷i G ˆ f ( g)÷( g) jGj gPG ˆ Put x ˆ y À1 gy.234 Representations and characters of groups @ • function ø: G 3 C by • ø( g) ˆ ø( g) 0 if g P H. if g P HX a 21.4.19 Proposition The values of the induced character ø 4 G are given by 1 ˆ • (ø 4 G)( g) ˆ ø( y À1 gy) j Hj yPG for all g P G.

de®ne the class function f G on G by x & 1 if y P x G f G ( y) ˆ ( y P G)X x 0 if y P x G a Thus f is the characteristic function of the conjugacy class x G . 21. j For practical purposes. then h÷. and ÷( yxy À1 ) ˆ ÷(x) for all y P G. ÷i G ˆ hø 4 G. f G i G ˆ x ÷(x) X jCG (x)j .20 Corollary If ø is a character of the subgroup H of G. Therefore a h f .Induced modules and characters 235 • since ø(x) ˆ 0 if x P H. then the degree of ø 4 G is given by (ø 4 G)(1) ˆ jGj ø(1)X j Hj Proof This follows immediately by evaluating (ø 4 G)(1) using Proposition 21.3).23 below). For x P G.21 Proposition If ÷ is a character of G and x P G. j 21. ÷i G X This was the equation required to show that f ˆ ø 4 G. a formula for the values of induced characters different from that given in Proposition 21. h f . ÷i G ˆ hø. Alternatively. ÷ 5 Hi H and so by the Frobenius Reciprocity Theorem. and we shall derive this next (it is given in Proposition 21.19.19 is more useful. so the proof is complete. the stated degree of ø 4 G can be found using just the de®nition of induced modules (see Exercise 21.

(2) If some element of x G lies in H. .22)). xm P H such that H H f G 5 H ˆ f x1 ‡ X X X ‡ f x m X x Statement (2) just says that H ’ x G breaks up into m conjugacy classes of H.22) Suppose that x P G. xm P H and f G 5 H ˆ f x1 ‡ F F F ‡ f x m (as in (21. f G i G ˆ hø. .18.23 Proposition Let ø be a character of the subgroup H of G. (1) If no element of x G lies in H. x m . . . we have: (21. jCH (x1 )j jCH (xm )j H H where x1 . f G i G ˆ x ˆ ˆ Note that a result similar to Proposition 21. with representatives x1 . we have (ø 4 G)(x) ˆ hø 4 G. . then (ø 4 G)(x) ˆ 0. . f G 5 Hi H X x x jCG (x)j . 2 or more conjugacy classes of H.21 was used in the proof of Theorem 16. then   ø(x1 ) ø(xm ) (ø 4 G)(x) ˆ jCG (x)j ‡X X X‡ . but if g P G then g G may contain 0. If H < G and h P H then h H  h G . x (2) If some element of x G lies in H. then f G 5 H ˆ 0. 1. and suppose that x P G. To put this another way. .4. then there are elements x1 . x Proof By Proposition 21. X X X . 21. (1) If no element of x G lies in H.21 and Corollary 21.236 Proof We have Representations and characters of groups 1 ˆ ÷( g) f G ( g) x jGj gPG 1 ˆ jx G j ÷(x) ÷( g) ˆ jGj gPx G jGj ÷(x) jCG (x)j by Theorem 12X8X j h÷. .

with representatives (1 3)(2 4) and (1 2)(3 4). fb ˆ (1 3). We have G H f1 5 H ˆ f1 . bl.Induced modules and characters 237 If no element of x G lies in H. fa2 ˆ (1 3)(2 4)g. the statement G H H f (1 2)(3 4) 5 H ˆ f (1 3)(2 4) ‡ f (1 2)(3 4) records the fact that the G-conjugacy class (1 2)(3 4) G contains exactly two H-conjugacy classes. fa ˆ (1 2 3 4). since a4 ˆ b2 ˆ 1 and bÀ1 ab ˆ aÀ1 .24 Example Let G ˆ S4 and let H ˆ ka. 21. . f x1 i H ‡ X X X ‡ hø. G f (1 2 3) 5 H ˆ 0. then f G 5 H ˆ 0. then x (ø 4 G)(x) H H ˆ hø. and hence x (ø 4 G)(x) ˆ 0.12). f x m i H ˆ ø(x1 ) ø(xm ) ‡X X X‡ jCH (x1 )j jCH (xm )j by Proposition 21X21X j The result follows. the conjugacy classes of H are f1g.22)(2). And if some element of x G lies in H. f x1 ‡ X X X ‡ f x m i H jCG (x)j H H ˆ hø. where a ˆ (1 2 3 4). and G H f (1 2 3 4) 5 H ˆ f (1 2 3 4) X For example. By (12. G H f (1 3) 5 H ˆ f (1 3) . and H H f G 5 H ˆ f x1 ‡ X X X ‡ f x m as in (21. G H H f (1 2)(3 4) 5 H ˆ f (1 3)(2 4) ‡ f (1 2)(3 4) . a2 b ˆ (2 4)g. a3 ˆ (1 4 3 2)g. b ˆ (1 3)X Then H  D8.

b in S7 by a ˆ (1 2 3 4 5 6 7). ÷5 of H  D8.25 Example (cf. we have (ø 4 G)(1) ˆ 24 ø(1) . 4   ø((1 3)(2 4)) ø((1 2)(3 4)) (ø 4 G)((1 2)(3 4)) ˆ 8 ‡ . we use induced characters to ®nd the character table of a group of order 21. 21.23. 8 ø((1 3)) (ø 4 G)((1 3)) ˆ 4 . Referring to Example 16. . Then according to Proposition 21.238 Representations and characters of groups The orders of the centralizers of the elements of H are as follows: h |CG (h)| |C H (h)| 1 24 8 (1 3)(2 4) 8 8 (1 2 3 4) 4 4 (1 3) 4 4 (1 2)(3 4) 8 4 Suppose that ø is a character of H. .3(3) for the irreducible characters ÷1 . 8 4 (ø 4 G)((1 2 3 4)) ˆ 4 ø((1 2 3 4)) X 4 (ø 4 G)((1 2 3)) ˆ 0. b ˆ (2 3 5)(4 7 6) . . we therefore have 1 ÷1 4 G ÷2 4 G ÷3 4 G ÷4 4 G ÷5 4 G 3 3 3 3 6 (1 2) 1 À1 1 À1 0 (1 2 3) 0 0 0 0 0 (1 2)(3 4) 3 À1 À1 3 À2 (1 2 3 4) 1 1 À1 À1 0 In the next example.2) De®ne permutations a. Exercise 17. .

b and b2 to be representatives of the conjugacy classes. Let ç ˆ e2ðia7 . note that H H H f G 5 H ˆ f a ‡ f a2 ‡ f a4 a . we obtain three linear characters ÷1 . Their values are shown below: g |CG ( g)| ÷1 ÷2 ÷3 1 21 1 1 1 a 7 1 1 1 a3 7 1 1 1 b 3 1 ù ù2 b2 3 1 ù2 ù where ù ˆ e2ðia3 . We aim to ®nd the character table of G. 0 < j < 2. a2 . 21. G has order 21. a4 g. a. 7 divides jC G (a)j. fa3 . Hence jC G (a)j ˆ 7. bl of S7 . Check that a7 ˆ b3 ˆ 1. there is a character ø k of H given by ø k (a j ) ˆ ç jk (0 < j < 6)X To use the formula in Proposition 21. First we ®nd the conjugacy classes. Since hai < C G (a). a jC G (a)j . a3 . For 1 < k < 6. fa. we see that the conjugacy classes of G are f1g. Since kal v G and Gahai  C3. a6 g.23 for calculating ø k 4 G. Using this. ÷2 . We shall obtain the last two irreducible characters of G by inducing linear characters of H. fai b: 0 < i < 6g. and similarly jC G (b)j ˆ 3. Let H ˆ kal. bÀ1 ab ˆ a2 X 239 It follows from these relations that the elements of G are all of the form ai bj with 0 < i < 6. a5 . ÷3 of G as the lifts of the linear characters of Gahai. fai b2 : 0 < i < 6gX We take 1.Induced modules and characters and let G be the subgroup ka. Also. and since b P C G (a). Notice that G has exactly ®ve irreducible characters.

240 Representations and characters of groups since no two of the elements a. ø6 are linearly independent. Therefore ÷4 Tˆ ÷5 . (ø1 4 G)(b) ˆ (ø1 4 G)(b2 ) ˆ 0X Similarly (ø3 4 G)(1) ˆ 3. bÀ1 ab ˆ a2 i g |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 1 21 1 1 1 3 3 a 7 1 1 1 ç ‡ ç2 ‡ ç4 ç3 ‡ ç5 ‡ ç6 a3 7 1 1 1 ç3 ‡ ç5 ‡ ç6 ç ‡ ç2 ‡ ç4 b 3 1 ù ù2 0 0 b2 3 1 ù2 ù 0 0 . then the values of ÷4 and ÷5 are 1 ÷4 ÷5 3 3 a ç ‡ ç2 ‡ ç4 ç3 ‡ ç5 ‡ ç6 a3 ç3 ‡ ç5 ‡ ç6 ç ‡ ç2 ‡ ç4 b 0 0 b2 0 0 Now ÷4 5 H ˆ ø1 ‡ ø2 ‡ ø4 and ÷5 5 H ˆ ø3 ‡ ø5 ‡ ø6 . a4 are conjugate in H. 21 7 7 3 3 and similarly h÷5 . a2 . Character table of ha. ÷4 i G ˆ 9 2 2 0 0 ‡ ‡ ‡ ‡ ˆ 1. (ø3 4 G)(a) ˆ ç3 ‡ ç5 ‡ ç6 . . (ø3 4 G)(a3 ) ˆ ç ‡ ç2 ‡ ç4 . since ø1 . and the character table of G is as shown. Hence by Proposition 21. and (ø3 4 G)(b) ˆ (ø3 4 G)(b2 ) ˆ 0X Thus if we write ÷4 ˆ ø1 4 G and ÷5 ˆ ø3 4 G. (ø1 4 G)(a) ˆ ç ‡ ç2 ‡ ç4 and similarly (ø1 4 G)(a3 ) ˆ ç3 ‡ ç5 ‡ ç6 . .23. ÷5 i G ˆ 1. We now calculate that h÷4 . . . (ø1 4 G)(1) ˆ 3X And since no G-conjugate of b or b2 lies in H. b: a7 ˆ b3 ˆ 1. Thus ÷4 and ÷5 are our last two irreducible characters.

(b) Find a basis of the induced CG-module U 4 G. . where ø is a character of H and ÷ is a character of G. then   ø(x1 ) ø(xm ) (ø 4 G)( g) ˆ jCG ( g)j ‡X X X‡ jCH (x1 )j jCH (xm )j H H where f G 5 H ˆ f x1 ‡ X X X ‡ f x m . 241 1. ÷i G ˆ hø. Let G ˆ D8 ˆ ka. then U 4 G is simply U(CG). Is U 4 G irreducible? 2. If U is a C H-module of C H. Let G ˆ S4 and let H be the subgroup k(1 2 3)l  C3 . (a) If ÷1 . (c) Write down the character of the C H-module U and the character of the CG-module U 4 G. ÷ 5 Hi H . If no element of g G lies in H. g 4. . 3. bl.Induced modules and characters Summary of Chapter 21 Assume that H is a subgroup of G. The Frobenius Reciprocity Theorem states that hø 4 G. b: a4 ˆ b2 ˆ 1. Exercises for Chapter 21 1. ÷5 are the irreducible characters of G. De®ne U to be the 1-dimensional subspace of C H spanned by 1 À a2 ‡ b À a2 bX (a) Check that U is a C H-submodule of C H. an induced CG-module U 4 G can be de®ned. For each C H-module U. as given in . . If ø is a character of H then the induced character ø 4 G is given by 1 ˆ • (ø 4 G)( g) ˆ ø( y À1 gy)X j Hj yPG In particular. bÀ1 ab ˆ aÀ1 l and let H be the subgroup ka2 . . the degree of ø:G is |G: H|ø(1). 2. then (ø 4 G)( g) ˆ 0X If some element of g G lies in H.

) . bl. ø2 .25. Show that the integers d 1 . let ø be a character of H.5. Let ø be an irreducible character of H. where a ˆ (1 2 3 4 5 6 7). and let ÷ be a character of G. ø3 of C3 . 6. b ˆ (2 3 5)(4 7 6). . .25). Calculate the values of the induced characters ö 4 G and ø 4 G. satisfy k ˆ iˆ1 d 2 < jG : HjX i (Compare Proposition 20. ÷ k be the irreducible characters of G. Let ö and ø be the irreducible characters of H which are given by gi |C H ( g i )| ö ø 1 21 1 3 a 7 a3 7 b 3 1 0 b2 3 1 0 1 1 ç ‡ ç2 ‡ ç4 ç3 ‡ ç5 ‡ ç6 where ç ˆ e2ðia7 (see Example 21. X X X . Show direct from the de®nition that if H < G and ø is a character of H.) 5. . then jGj (ø 4 G)(1) ˆ ø(1)X j Hj 4.242 Representations and characters of groups Section 18. (b) Calculate the induced characters ø j 4 G (1 < j < 3) as sums of the irreducible characters ÷ i of G.1. Let G ˆ S7 and let H ˆ ka. You are given that jC G (a)j ˆ 7 and jC G (b)j ˆ 18. Let H be a subgroup of G. which are given by ø 4 G ˆ d 1 ÷1 ‡ X X X d k ÷k . . d k . work out the restrictions ÷ i 5 H (1 < i < 5) as sums of the irreducible characters ø1 . Prove that (ø(÷ 5 H)) 4 G ˆ (ø 4 G)÷X (Hint: consider the inner product of each side with an arbitrary irreducible character of G. 3. and use the Frobenius Reciprocity Theorem. Suppose that H is a subgroup of G. as in Example 21. and let ÷1 .

and let ø be an irreducible character of H.Induced modules and characters 243 7. . Discover and prove results for ø 4 G which are similar to those presented in Chapter 20 for the restriction of irreducible characters of G to H. Suppose that H is a normal subgroup of index 2 in G.

then ÷( g)  ÷(1) mod p for any character ÷ of G for which ÷( g) is an integer. we require that det (A À ëI) ˆ 0 for some square matrix A with integer entries. Most of the results concern arithmetic properties of character values. And we prove some useful congruence properties. Algebraic integers 22. perhaps the most opaque is that which states that the degree of an irreducible character of a ®nite group G must divide the order of G. We discuss properties of a group element g P G which ensure that ÷( g) is an integer for all characters ÷ of G. Thus. all of whose entries are integers. we have uA ˆ ëu for some non-zero row vector u. if p is a prime number and g P G is an element of order pr for some r.1 De®nition A complex number ë is an algebraic integer if and only if ë is an eigenvalue of some matrix. We remark that ë is an algebraic integer if and only if ë is a root of a polynomial of the form 244 . for ë to be an algebraic integer. This is one of several results which we shall prove in this chapter. using algebraic integers. for example.22 Algebraic integers Among the properties of characters which may be regarded as fundamental. for the same matrix A. Equivalently.

then u(ÀA) ˆ (Àë)u and uA ˆ ëu. . note that if A is an integer matrix and u is a row vector with uA ˆ ëu. then so are Àë and the complex conjugate ë of ë.7). Proof There exist square matrices A and B. and let u be the row vector (1.3 Theorem If ë and ì are algebraic integers. vB ˆ ìvX Suppose that A is an m 3 m matrix and B is an n 3 n matrix. (4) Let A be the n 3 n matrix H 0 0 f1 0 f f0 1 Aˆf f f d0 0 0 0 given by 0 XXX 0 XXX 0 XXX XXX 0 XXX 0 XXX 0 0 0 1 0 I 0 1 0 0g g 0 0g gX g g 0 0e 1 0 Suppose that ù is an nth root of unity. since it is an eigenvalue of the 0 1 matrix X 2 0 (3) If ë is an algebraic integer. ù. all of whose entries are integers. . since n is an eigenvalue of the 1 3 1 matrix (n). To see this. then ëì and ë ‡ ì are also algebraic integers. p (2) 2 is an algebraic integer. X X X . 22. . such that uA ˆ ëu.2 Examples (1) Every integer n is an algebraic integer. ù2 . In fact. 1) ˆ ùuX This shows that every nth root of unity is an algebraic integer. . 22. algebraic integers are usually de®ned in this way. where u is the row vector obtained from u by replacing each entry by its complex conjugate.Algebraic integers x n ‡ a nÀ1 x nÀ1 ‡ X X X ‡ a1 x ‡ a0 245 where a0 . ù2 . ù nÀ1 ). Then uA ˆ (ù. a nÀ1 are integers (see Exercise 22. and non-zero row vectors u and v. ù nÀ1 . X X X . .

It is easy to check as in the proof of Proposition 19. then ë is an integer. j Theorem 22.246 Representations and characters of groups Let e1 .9. we have (x  y)(A  B) ˆ xA  yBX Hence (u  v)(A  B) ˆ uA  vB ˆ ëu  ìv ˆ ëì(u  v)X Therefore ëì is an eigenvalue of A  B. ( ë ij (ei  f j ))(A  B) ˆ ë ij (ei A  f j B)). ÷( g) is a sum of roots of unity. . Then (u  v)(A  I n ‡ I m  B) ˆ uA  vI n ‡ uI m  vB ˆ ëu  v ‡ u  ìv ˆ (ë ‡ ì)(u  v). 1 < j < n). X X X . De®ne an endomorphism A  B of V by (ei  f j )(A  B) ˆ ei A  f j B (1 < i < m. Hence ÷( g) is an algebraic integer. € € extending linearly (that is.3 shows that the set of all algebraic integers forms a subring of C. by Example 22.3. The next result provides a link between algebraic integers and characters.4 Corollary If ÷ is a character of G and g P G. f n be a basis of C n . and we deduce as above that ë ‡ ì is an algebraic integer. Let Im and In denote the identity m 3 m and n 3 n matrices. Then the vectors ei  fj (1 < i < m. Each root of unity is an algebraic integer. . so any sum of roots of unity is an algebraic integer by Theorem 22. 1 < j < n) has integer entries. em be a basis of C m and f 1 . . then ÷( g) is an algebraic integer.2(4).4 that for all vectors x P C m . Since the matrix of A  B relative to the basis ei  fj (1 < i < m.5 Proposition If ë is both a rational number and an algebraic integer. . respectively. Proof By Proposition 13. 22. 1 < j < n) form a basis of the tensor product space V ˆ C m  C n. y P C n. it follows that ëì is an algebraic integer. . j 22.

as a special case of Proposition 22. then ÷( g) is an integer. p the well known result that 2 is irrational. Recall from De®nition 12. with character ÷. the entries of sA À rI which are not on the diagonal are divisible by s. j The next result is an immediate consequence of Corollary 22.5. (Example 22.21 that if C is a conjugacy class of G.6 Corollary Let ÷ be a character of G and let g P G. Thus  n 1 det (A À ëI) ˆ det (sA À rI) Tˆ 0. where r and s are coprime integers and s Tˆ Æ1. In passing. then ˆ Cˆ x P CGX xPC 22. 22. Therefore det (sA À rI) ˆ (Àr) n ‡ mp for some integer m.7 Lemma Suppose that g P G and that C is the conjugacy class of G which contains g. and hence also by p. we establish two preliminary lemmas. Let U be an irreducible CG-module. If ÷( g) is a rational number. which is enough to establish the proposition.2(2) shows p that 2 is an algebraic integer. Let p be a prime number which divides s. As p does not divide r (since r and s are coprime). we deduce that det (sA À rI) Tˆ 0.Algebraic integers 247 Proof Suppose that ë is a rational number which is not an integer.5. s and hence ë is not an algebraic integer. Write ë ˆ ras. We shall show that ë is not an algebraic integer. For every n 3 n matrix A of integers.) The degree of every irreducible character divides |G| To prepare for the proof that |G| is divisible by the degree of each irreducible character of G. note that we have.4 and Proposition 22. Then .

we know by Proposition 9.8 Lemma € Let r ˆ gPG á g g P CG. Then for 1 < i < n.14 that there exists ë P C such that uC ˆ ëu for all u P U. we obtain ˆ ÷(x) ˆ ë÷(1).248 Representations and characters of groups uC ˆ ëu for all u P U . the result follows. gn be the elements of G. j 22. that is. this yields jCj÷( g) ˆ ë÷(1)X Thus ë ˆ jCj÷( g)a÷(1). Then ë is an algebraic integer.8. . we have gi r ˆ n ˆ jˆ1 aij g j . As |C| ˆ |G:CG ( g)| by Theorem 12. where ëˆ jGj ÷( g) X jCG ( g)j ÷(1) Proof Since C lies in the centre of CG (see Proposition 12. where each á g is an integer. .22). ˆ u x ˆ ëu for all u P U X xPC Consequently if B is a basis of U. Suppose that u is a non-zero element of CG such that ur ˆ ëu. . Proof Let g1 . where ë P C. then ˆ [x]B ˆ ëIX xPC Taking the traces of both sides of this equation. . xPC and since ÷ is constant on the conjugacy class C.

aij ˆ á g where g ˆ gÀ1 gj . k ˆ iˆ1 jGj ÷( g i )÷( g i ) jCG ( g i )j ÷(1) .10 and 22.Algebraic integers 249 for certain integers aij .2(4). and let C be the sum of the elements in the conjugacy class of G which contains g.) The i statement that ur ˆ ëu (with u Tˆ 0) says that ë is an eigenvalue of the integer matrix A ˆ (aij ). Then for all i. Then ux ˆ ùu and so Lemma 22. Proof Let g1 . Therefore ë is an algebraic integer. Proof Let U be an irreducible CG-submodule of CG with character ÷.3. j 22. where ù is an nth root of unity.7. 22. gk be representatives of the conjugacy classes of G.11 Theorem If ÷ is an irreducible character of G. (In fact. then ëˆ is an algebraic integer.8 con®rms that ù is an algebraic integer. .4. then ÷(1) divides |G|. . by Lemma 22. Then uC ˆ ëu for all u P U. . by Lemma 22.8. by Corollaries 22.9 Example Let G ˆ Cn ˆ kx: x n ˆ 1l. both jGj ÷( g i ) jCG ( g i )j ÷(1) and ÷( g i ) jGj ÷( g) jCG ( g)j ÷(1) are algebraic integers. Therefore ë is an algebraic integer. Hence by Theorem 22. and de®ne u ˆ 1 ‡ ùx À1 ‡ ù2 x À2 ‡ X X X ‡ ù nÀ1 x P CG. j 22. .10 Corollary If ÷ is an irreducible character of G and g P G. Notice that this example is just a reworking of Example 22.

As jGja÷(1) is a rational number. Since Ker r v G and G is simple. Therefore by Theorem 17. Proof Suppose that G is a simple group which has an irreducible character ÷ of degree 2. Theorem 22. Theorem 16. But g 3 det ( gr) is a linear character of G (see Exercise 13. (Note that ÷(1) . In particular.250 Representations and characters of groups is an algebraic integer. and so r is injective. This algebraic integer is equal to jGja÷(1).13 Corollary No simple group has an irreducible character of degree 2. as the sum of the squares of the degrees of the irreducible characters is equal to |G|. Proposition 22. By Theorem 22.12 Examples (1) If p is a prime number and G is a group of order pn for some n. Let r: G 3 GL(2.11. and so G9 ˆ G as G is simple. and hence satis®es p < n. by the row orthogonality relations.7(a)). observe that G is non-abelian. the number of linear characters of G divides |G|. where p is prime.11.5 now implies that jGja÷(1) is an integer. then ÷(1) is a power of p for all irreducible characters ÷ of G. by Proposition 9. (Recall that a group G is simple if it has no normal subgroups apart from {1} and G.18. either the degrees of the irreducible characters of G are all 1.5. 1. if jGj ˆ p2 then ÷(1) ˆ 1 for all irreducible characters ÷. (2) Let G be a group of order 2 p.) 22. we have Ker r ˆ f1g. X X X . and this implies that det ( gr) ˆ 1 for all g P GX . G has no non-trivial linear characters. using Proposition 9. Hence.11. ÷(1) divides |G|. 2 (with ( p À 1)a2 degrees 2)X (3) If G ˆ Sn then every prime p which divides the degree of an irreducible character of G also divides n!. the degree of each irreducible character of G is 1 or 2 (it cannot be p for the reason noted in (1) above). or they are 1. C) be a representation of G with character ÷. By Theorem 17. That is. 2. j 22.) Hence.4(1). Hence G9 Tˆ 1.11 also has the following interesting consequences concerning irreducible characters of simple groups. First. we recover the well known result that groups of order p2 are abelian. p.

11. and we deduce that G has an abelian normal pcomplement N. we conclude that   À1 0 À1 T (xr)T ˆ X 0 À1 Thus xr ˆ T(ÀI)T À1 ˆ ÀIX Consequently (xr)( gr) ˆ ( gr)(xr) for all g P G. so we assume that G is non-abelian. Then G has an abelian normal p-complement. we assume that every irreducible character of G has degree a power of a prime p. and by Proposition 9.Algebraic integers 251 Now G has even order. G is not simple unless G has prime order. As r is injective. by Theorem 22.11 give us the equation ˆ jGj ˆ jGaG9j ‡ ÷(1)2 . This time. there is a 2 3 2 matrix T such that T À1 (xr)T is a diagonal matrix with diagonal entries Æ1. while jG: N j is a power of p. that is. Since det (xr) ˆ 1. N is an abelian normal subgroup of G. ÷(1) . As r is injective. Then ÷(1) is divisible by p. Theorems 11. 1. xr has order 2.6). this means that xg ˆ gx for all g P G. so p divides |G| by . j Our next result again shows that information about character degrees can sometimes be used to learn about the structure of a group. Therefore G contains an element x of order 2 (see Exercise 1.12 and 17. Since G is non-abelian. Consider the 2 3 2 matrix xr. and hence hxi v GX This is a contradiction. as G is simple.11. and jN j is coprime to p. In particular.8). 22. the sum being over the irreducible characters ÷ of G for which ÷(1) .14 Corollary Suppose that p is a prime and the degree of every irreducible character of G is a power of p. by our hypothesis. 1 for some irreducible character ÷ of G. Proof The result is correct if G is abelian (see Theorem 9.

8 shows that ø(1) divides ÷(1). We may now apply induction on |G| to deduce that H has an abelian normal p-complement N. and we deduce from the equation above that p divides the order of the abelian group GaG9. Also. it follows that GaG9 has a subgroup of index p. we have Z(G) ˆ G. so either N ˆ {1} or N ˆ G. Since every ®nite abelian group is isomorphic to a direct product of cyclic groups. j A condition which ensures that ÷( g) is an integer In Theorem 22. Let ø be an irreducible character of H. Hence N consists of those elements of G whose order is coprime to p.4. so it remains to prove that N v G. Bearing in mind the dif®culties we encountered in constructing the character tables of Sn for small values of n (we reached n ˆ 6 in Example 19. then we denote their highest common factor by (a. øi Tˆ 0 for some irreducible character ø of H. Before proving Theorem 22.17). since otherwise p divides the order of gH which in turn divides the order of g. by Proposition 20. by Exercise 1.11. But an abelian simple group has prime order. This result implies. every entry in the character table of Sn is an integer (see Corollary 22.16. On the other hand. and from this fact it follows easily that N v G. Therefore. we require a preliminary lemma concerning roots of unity.16 below we give a group-theoretic condition on an element g of G which implies that ÷( g) is an integer for every character ÷ of G.252 Representations and characters of groups Theorem 22.15 Lemma If ù is an nth root of unity. then . Next. 22.1. Suppose that g P G and the order of g is coprime to p. If a and b are positive integers. if N ˆ G then G is again abelian. we write d|n to denote the fact that d divides n.16 is evidently a useful result. a similar argument shows that g P N. assume that G is simple. b). because G is simple. Clifford's Theorem 20. We have that |N| is coprime to p and jG : N j is a power of p. for integers d and n. Theorem 22. for example. G has prime order. If N ˆ f1g then G is a p-group so Z(G) Tˆ {1} (see Exercise 12. Finally. Then g P H. so G is abelian.17). that for all n. Then h÷ 5 H. Hence G has a normal subgroup H of index p.7). so ø(1) is a power of p.

(i. Suppose that g is conjugate to gi for all i with 1 < i < n and (i.n)ˆ1 253 is an integer. Then ÷( g) is an integer for all characters ÷ of G.11.nad)ˆ1 It follows that ˆ 1<i< n. ˆ ù dj P ZX 1< j< nad. Proof We prove the result by induction on n.Algebraic integers ˆ ùi 1<i< n. 1< j< nad.n)ˆd ùi ˆ ˆ ˆ ù dj X dj n 1< j< nad.nad)ˆ1 If djn then ù d is an (nad)th root of unity. (i.n)ˆ1 ùi ˆ n ˆ iˆ1 ùi À ˆ ˆ ù dj P Z.nad)ˆ1 j as required. ( j. 1 ( j. So suppose that ù is an nth root of unity and ù Tˆ 1. ( j.16 Theorem Let g be an element of order n in G. and if in addition d . 22. d . 1. if ù ˆ 1 then the result is immediate. Then ù is a root of the polynomial €n (x n À 1)a(x À 1) ˆ x nÀ1 ‡ X X X ‡ x ‡ 1X Therefore iˆ1 ù i ˆ 0. €n i Now we partition the sum iˆ1 ù according to the highest common factor d of i and n: 0ˆ n ˆ iˆ1 ùi ˆ ˆ ˆ dj n 1<i< n (i. Proof Let V be a CG-module with character ÷ of degree m. there is a basis B of V such that H I ù1 0 f g FF [ g]B ˆ d e F ùm 0 . By Proposition 9. n) ˆ 1. Also. It is trivial for n ˆ 1. dj n. then by our induction hypothesis.

17 Corollary All the character values of symmetric groups are integers. Consequently ÷( g) is a rational number. namely that if ÷( g i ) P Z for all characters ÷ of G. . one particularly useful consequence of our results is that if p is a prime number. 22. then the permutations g and g i have the same cycle-shape. Proof If g P Sn and i is coprime to the order of g. then g is conjugate to g i whenever i is coprime to n.16.254 Representations and characters of groups where ù1 . ù m are nth roots of unity. ˆ ÷( g i ) P ZX 1<i< n. . For example. . we need to de®ne the p9-part of a group element.16. . and hence are conjugate by Theorem 12. For 1 < i < n. j We remark that using Galois theory it is possible to prove the converse of Theorem 22. .6. The result now follows from Theorem 22. g is an element of G of order pr for some r. . Before going into the character theory. we have ÷( g i ) ˆ ÷( g) for such i. (i. then ÷( g)  ÷(1) mod p. n) ˆ 1. and so ÷( g) is an integer by Corollary 22.15. where s is the number of integers i with 1 < i < n and (i. and so i ÷( g i ) ˆ ù1 ‡ X X X ‡ ù im X Therefore by Lemma 22. .n)ˆ1 As g is conjugate to gi if 1 < i < n and (i. . and ÷ is a character of G such that ÷( g) P Z. The de®nition will emerge from the following lemma. the matrix i [ g i ]B has diagonal entries ù1 . n) ˆ 1. j The p9-part of a group element The rest of the chapter is devoted to some important congruence properties of character values.15. and hence s÷( g) P Z. . ù im .

Then there exist x. Now suppose that x9. x p ˆ g aup ˆ 1. (2) the order of x is a power of p.18 Lemma Let p be a prime number and let g P G. Then xy ˆ yx ˆ g au‡bp ˆ g. so is coprime to p. g ˆ x9 y9 ˆ y9x9. that is. Moreover. as required. b such that au ‡ bpv ˆ 1X Put x ˆ gau and y ˆ gbp . xy ˆ g ˆ x9 y9.Algebraic integers 255 22. hence also with gau ˆ x. and (3) the order of y is coprime to p. v P Z and (u. y P G such that (1) g ˆ xy ˆ yx. y9 commutes with y and y( y9)À1 has order coprime to p. the order of x9 is a power of p and the order of y9 is coprime to p. Proof Let the order of g be upv . p) ˆ 1. j v v v v v . Since both x and x9 have order a power of p. Finally. where u. Similarly. y u ˆ g bup ˆ 1X Hence the order of x is a power of p and the order of y divides u. so x9 commutes with g. the elements x and y of G which satisfy conditions (1)±(3) are unique. so x À1 x9 ˆ y( y9)À1 X If z ˆ x À1 x9 ˆ y( y9)À1 . then we have shown that the order of z is both a power of p and coprime to p. y9 P G also satisfy (1)±(3). We have x9 g ˆ x9 y9x9 ˆ gx9. Therefore x and y satisfy conditions (1)±(3). and so x ˆ x9 and y ˆ y9. Therefore z ˆ 1. it follows that x À1 x9 has order a power of p. Then there exist integers a. We must show that x ˆ x9 and y ˆ y9.

21 Proposition There are only ®nitely many ideals I of Z[æ] which contain pZ[æ]. . . æ nÀ1 . we need a few basic facts about a certain subring of C in which all our character values will lie. if p ˆ 2 and g has order 6.256 Representations and characters of groups 22. where u. . . with ai P Z. æ. Proof Consider the factor ring Z[æ]a pZ[æ]. so in fact Z[æ] ˆ f f (æ): f (x) P Z[x] of degree < n À 1gX Now let p be a prime number and let pZ[æ] ˆ f pr: r P Z[æ]g. The ideals of Z[æ] which contain pZ[æ] are in . For example. the expression g ˆ xy in Lemma 22. Z[æ] ˆ f f (æ): f (x) P Z[x]gX Clearly. v Then the p9. p) ˆ 1. æ2 . v P Z and (u.18 has x ˆ g3 .20) Let the order of g be upv .18. a principal ideal of Z[æ]. b with au ‡ bpv ˆ 1. By de®nition. Every such coset contains an element of the form a0 ‡ a1 æ ‡ X X X ‡ a nÀ1 æ nÀ1 . We extract the following statement from the proof of Lemma 22. y ˆ gÀ2 . À2 A little ring theory To prepare for our main result on congruence properties of character values. we conclude that Z[æ]a pZ[æ] is ®nite. Let n be a positive integer and let æ ˆ e2ðia n . De®ne Z[æ] to be the subring of C generated by Z and æ. 0 < ai < p À 1 for all iX As there are only ®nitely many such elements. 22. this has as its elements all the cosets pZ[æ] ‡ r with r P Z[æ].18 the p9-part of g.19 De®nition We call the element y which appears in Lemma 22.part of g is gbp . every element of Z[æ] is an integer combination of the powers 1. then the p9-part of g is g . (22. and choose integers a. that is.

b with j am ‡ bp ˆ 1. Since n r ˆ rr nÀ1 . b P P such that 1 ˆ ra ‡ bX Then s ˆ rsa ‡ sbX As rs P P and b P P. P is a proper ideal which is contained in no larger proper ideal of Z[æ].) We now prove two easy results about the maximal ideal P. a Since r P P. assume that r n P P. 22. then either r P P or s P P. Since p P P. which establishes that P ’ Z  pZ. there exist a P Z[æ]. we therefore have rZ[æ] ‡ P ˆ Z[æ]X Consequently. For the last statement of the proposition.23 Proposition We have P ’ Z ˆ pZ. the ideal rZ[æ] ‡ P of Z[æ] strictly contains P. it follows that s P P.21 that there is a maximal ideal of P of Z[æ] which contains pZ[æ]. In particular. that is. if r n P P for some positive integer n. As P is a maximal. since P Tˆ Z[æ]. as required.Algebraic integers 257 bijective correspondence with the ideals of Z[æ]a pZ[æ] (the correspondence being I 3 Ia pZ[æ]). If p B m then there are integers a. we conclude that r P P. but this implies that 1 P P. this implies that either r P P or r nÀ1 P P. we also have pZ  P ’ Z. which is false. Proof Assume that rs P P and r P P. Proof Let m P P ’ Z.22 Proposition If r. then r P P. and the proof is complete. Repeating this argument. j 22. We must show that s P P. Thus pjm. s P Z[æ] and rs P P. (A proper ideal of Z[æ] is an ideal which is not equal to Z[æ]. Therefore there are only ®nitely many such ideals. j We deduce from Proposition 22. j .

Then y ˆ g bp (see (22. If ÷ is any character of G. if p Tˆ 2. where á P Z[æ]. then ÷( g) À ÷( y) P PX Proof Suppose that g has order m ˆ upv . so it follows that (ù À ù bp ) p P pZ[æ]X Thus (ù À ù bp ) p lies in the maximal ideal P. pv .11). b with au ‡ bpv ˆ 1. p) ˆ 1. Then v ù ˆ ù au‡bp . we have @ 0.part of g. let p be a prime number and let P be a maximal ideal of Z[æ] containing pZ[æ]. ù bp ˆ ù bp X Consider the number (ù À ù bp ) p . Hence (ù À ù bp ) p ˆ ù p ‡ (À1) p ù bp ‡ pá. The orders of g and of y divide n ˆ |G|.22 now forces v v v v v 2v v v v v 2v v v 2v v v . As in the previous section. since ù p ˆ ù bp . where u. Moreover.  v vÀ r v p bpv pv pv v pvÀ1 bpv (ù À ù ) ˆ ù À p ù ù ‡ X X X Æ ù p ù rbp r ‡ X X X ‡ (À1) p ù bp X For 0 . Choose integers a. and hence lie in Z[æ]. By the Binomial Theorem. and so v v 2v 2v ù p ˆ ù aup . the binomial coef®cient ( rp ) is divisible by p. Now let ù be an mth root of unity (so ù P Z[æ] as mjn). 22. The ring Z[æ] is of interest because all the character values of G lie in Z[æ] (see Proposition 9. v P Z and v (u.20)).24 Theorem Let g P G and let y be the p9. r .258 Representations and characters of groups Congruences At last we are ready to prove our results on congruences of character values. so ÷( g) and ÷( y) are both sums of nth roots of unity. Application of Proposition 22. pv pv bp2v ù ‡ (À1) ù ˆ pv 2ù . Let G be a group of order n and let æ ˆ e2ðia n . if p ˆ 2.

then ÷( g)  ÷(1) mod pX Proof As the order of g is a power of p. lies in P. .11.27 extensively in our character calculations in Chapters 25±7. Suppose that g P G and that y is the p9.27 in which g has order 2.23 give ÷( g) À ÷( y) P P ’ Z ˆ pZX Therefore ÷( g)  ÷( y) mod p. j 22. For the moment. If ÷ is a character of G such that ÷( g) P Z. .Algebraic integers (22X25) (ù À ù bp ) P PX v 259 By Proposition 9. If ÷ is a character of G such that ÷( g) and ÷( y) are both integers.24±22. so the result is immediate from Corollary 22. j 22. . by (22.26. . then ÷( g)  ÷( y) mod pX Proof As ÷( g) and ÷( y) are both integers. Theorem 22. Suppose that g P G and the order of g is a power of p.10 is the special case of Corollary 22. d v v which.25). there are mth roots of unity ù1 .part of g.27 Corollary Let p be a prime number.26 Corollary Let p be a prime number. . ù d such that bp ÷( g) ˆ ù1 ‡ X X X ‡ ù d and ÷( y) ˆ ù1 ‡ X X X ‡ ù bp X d v v Then bp ÷( g) À ÷( y) ˆ (ù1 À ù1 ) ‡ X X X ‡ (ù d À ù bp ). we just illustrate the results with reference to some character tables which we already know. We shall use the congruence results 22. j Notice that Corollary 13.24 and Proposition 22. the p9-part of g is 1.

24 for this a value. Thus the corresponding entries in columns 1 and 2 Character table of A5 1 ÷1 ÷2 ÷3 ÷4 ÷5 1 4 5 3 3 (1 2 3) 1 1 À1 0 0 (1 2)(3 4) 1 0 1 À1 À1 (1 2 3 4 5) 1 À1 0 á â (1 3 4 5 2) 1 À1 0 â á of the character table are congruent modulo 3. 3X However. If we take p ˆ 5 and g ˆ (1 2 3 4 5). 1(1 À 5) ˆ â 5X 2 Put æ ˆ e2ðia60 .28 Example Recall from Example 20.24. Since â P Z[æ] (see p Proposition 9.22. p p Then ( 5)2 P P. we have â 5 P P. .11). as can be seen by inspecting the table. and let P be a maximal ideal of Z[æ] containing 5Z[æ].26 implies that ÷( g)  ÷(1) mod 3 whenever ÷( g) P Z. We illustrate Theorem 22. and p ÷4 ( g) À ÷4 (1) ˆ á À 3 ˆ 1(1 ‡ 5 À 6) 2 p p p ˆ 5 . ÷4 ( g) À ÷4 (1) P PX This illustrates Theorem 22. That is. then the p9-part of g is 1.14 that the character table of A5 is as shown. Also ÷ i ((1 2 3 4 5))  ÷ i (1) mod 5 for i ˆ 1. ⠈ (1 À 5)a2. so 5 P P by Proposition 22. 2. ÷4 ((1 2 3 4 5)) ˆ á P Z. Summary of Chapter 22 1. If g ˆ (1 2 3) then Corollary 22. 2. Character values are algebraic integers. Similarly the entries in columns 1 and 3 are congruent modulo 2. p p where á ˆ (1 ‡ 5)a2.260 Representations and characters of groups 22. The degree of every irreducible character of G divides |G|.

11 to show that every irreducible character of G has degree 1. for all characters ÷. then g ˆ 1. Exercises for Chapter 22 1. If g P G and y is the p9-part of g. (a) Find the degrees of all the irreducible characters of G. 17.12. Let G be a group and let ö be a character of G such that ö( g) ˆ ö(h) for all non-identity elements g and h of G. jGj where á is an algebraic integer. If g is conjugate to g i for all integers i which are coprime to the order of g. Let p and q be prime numbers with p . 3. . Suppose that G is a group of odd order. and let G be a nonabelian group of order pq. Prove that h÷. (c) Show that q divides p À 1 and that G has q ‡ (( p À 1)aq) conjugacy classes. 2. (b) Show that |G9| ˆ p. Use Theorems 11. Let G be a group of order 15.Algebraic integers 261 3. (c) Deduce that ÷ ˆ 1 G . (b) Now let ÷ be an irreducible character of G with ÷ ˆ ÷. then ÷( g) is an integer. (c) Show that if ÷ is a non-trivial irreducible character of G. (b) Show that a ‡ b and a ‡ b|G| are integers. Deduce that G is abelian. then b÷(1) is an integer. Prove that the number of conjugacy classes in a group of order 16 is 7. (a) Prove that if g P G and g ˆ gÀ1 .11 and 22. 1 G i ˆ 1 (÷(1) ‡ 2á). Let p be a prime number. 10 or 16. 4. 4. b P C. then ÷( g)  ÷( y) mod p for all characters ÷ of G such that ÷( g) and ÷( y) are integers. q. 5. (d) Deduce that both a and b are integers. This exercise shows that the only irreducible character ÷ of G such that ÷ ˆ ÷ is the trivial character. (a) Show that ö ˆ a1 G ‡ b÷reg for some a.

1 or À1. (d) You are given that all entries in the character table of G are integers. Moreover. and contains an element g of order 5 such that jCG ( g)j ˆ 5.262 Representations and characters of groups 6. . (c) Find ÷(1) and ÷( g) for all irreducible characters ÷ of G. g. A certain group G of order 120 has exactly seven conjugacy classes. g2 . where each ar (0 < r < n À 1) is an integer.27 to deduce that G has two irreducible characters of degree 5. (a) Show that for every irreducible character ÷ of G. ÷( g) is 0. and that the conjugacy classes of G have representatives g1 . Prove that a complex number ë is an algebraic integer if and only if ë is a root of a polynomial of the form x n ‡ a nÀ1 x nÀ1 ‡ X X X ‡ a1 x ‡ a0 . (b) Use Corollary 22. . . It is often possible to calculate the character table from limited arithmetic information about the group G.26 and the column orthogonality relations. This exercise illustrates this point with the group G ˆ S5 . . g3 and g4 are conjugate in G. 7. . ®nd the character table of G. g7 with orders and centralizer orders as follows: gi Order of gi |CG ( gi )| g1 1 120 g2 2 12 g3 2 8 g4 3 6 g5 4 4 g6 6 6 g7 5 5 Using Corollary 22.

the subject of real representations not only is elegant and interesting. which consist largely of the calculation of character tables and applications of character theory. If all the matrices gr ( g P G) have real entries. Often. There is a subtle interplay between representations over C and representations over R. Real characters An element g of the ®nite group G is said to be real if g is conjugate to gÀ1 . However. but also gives delicate information about characters which often comes into play in more dif®cult calculations. while there is no representation ó equivalent to r with all the matrices gó having real entries. The material in this chapter is perhaps at a slightly more advanced level than that in the rest of the book. Let r be a representation of G. This is used in the last section to prove a famous result of Brauer and Fowler concerning centralizers of involutions in ®nite simple groups. characters of CG-modules are real-valued. and if g is real. and is not used in the ensuing chapters. which we shall explore in this chapter. Various criteria for whether or not a character corresponds to a representation over R lead us to the remarkable Frobenius±Schur Count of Involutions. we have always taken our representations to be over the ®eld C of complex numbers. some results in representation theory work equally well for the ®eld R of real numbers. However. the converse is not true: it can happen that the character of r is realvalued. and the ®rst main result of the chapter describes the number of real-valued irreducible characters of G.23 Real representations Since Chapter 9. then of course the character of r is real-valued. then the conjugacy class g G is also said to be 263 . Nevertheless.

5. For every conjugacy class g G of G. Thus for example. X is invertible. Since the trace of a permutation matrix is equal to the number of points ®xed by the corresponding permutation. and the trivial character of G is real.4). On the other hand. Part of the following corollary was obtained by a different method in Exercise 22.2.264 Representations and characters of groups real. Therefore Q ˆ X À1 X ˆ X À1 PX X Consequently P and Q have the same trace. Hence there is a permutation matrix P such that PX ˆ X (see Exercise 4. and let X denote the complex conjugate of the matrix X. Therefore X can be obtained from X be permuting the columns. so X can be obtained from X by permuting the rows. the entries in the column of X which corresponds to g G are the complex conjugates of the entries in the column of X which corresponds to ( g À1 ) G . and the number of real conjugacy classes of G is tr (Q)X j As these numbers are equal. and so there is a permutation matrix Q such that XQ ˆ X By Proposition 16. then it contains the inverse of each of its elements. . the conjugacy class {1} of G is real. a character ÷ of G is real if ÷( g) is real for all g P G. the complex conjugate ÷ is also an irreducible character (see Proposition 13. we have the number of real irreducible characters of G is tr (P). since ( g À1 ) G ˆ fx À1 : x P g G g.15). Proof Let X denote the character table of G. the result is proved. Notice that if a conjugacy class is real.2.1 Theorem The number of real irreducible characters of G is equal to the number of real conjugacy classes of G. by Proposition 13. For every irreducible character ÷ of G. 23.

then no non-identity element of G is real (see the solution to Exercise 23. bÀ1 ab ˆ aÀ1 l. since     À1 0 0 1 .3(3)). .1. and there is a basis v1 . . G has an element g of order 2. Proof If G has odd order. Hence G has at least two real conjugacy classes. then by Exercise 1. b: a4 ˆ b2 ˆ 1.8. . .1).1). b: a4 ˆ 1. v i g is a linear combination of v1 .3 Examples (1) Let G ˆ D8 ˆ ka. . v n of V.Real representations 265 23. and let ÷ be the irreducible character of G of degree 2 (see Example 16. . C) with character ÷. The values of ÷ are as follows: 1 ÷ 2 a2 À2 a 0 b 0 ab 0 . and so G has at least two real irreducible characters by Theorem 23. Then ÷ can be realized over R. br ˆ ar ˆ 0 1 À1 0 provides a representation r of G with character ÷ such that all the matrices gr ( g P G) have real entries. If G has even order. . v n with real coef®cients. such that all the entries in each matrix gr ( g P G) are real.1. the only real character of G is the trivial character. and let ÷ be the irreducible character of G of degree 2 (see Exercise 17. 23. We say that ÷ can be realized over R if there is a representation r: G 3 GL (n.2 Corollary The group G has a non-trivial real irreducible character if and only if the order of G is even. b2 ˆ a2 . bÀ1 ab ˆ aÀ1 l. such that for all g P G and 1 < i < n. j Characters which can be realized over R Let ÷ be a character of the group G. {1} and g G. This is the same as saying that there is some CG-module V with character ÷. Therefore by Theorem 23. . (2) Let G ˆ Q8 ˆ ka.

. In fact. where F is R or C. .3(2) tells us that the converse is false.266 Representations and characters of groups Thus ÷ is real. v n .3(1)). Therefore we obtain a representation r: G 3 GL (n. . but it is at the moment unclear how to prove this. Simply take a basis v1 . if we de®ne v1 x ˆ v2 . . C). .) À1 À1 Every RG-module can easily be converted into a CG-module. and hence also in C. the matrix gr has its entries in R. and consider the vector space over C with basis v1 . with a multiplication by elements of G satisfying the conditions of De®nition 4. where C3 ˆ kx: x 3 ˆ 1l. .18(3) below. The construction is even easier to understand in terms of representations: if r: G 3 GL (n. (We shall eventually establish this in Example 23. RG-modules Recall that in Chapter 4 we de®ned an FG-module. R) is a representation then for each g P G. v1 b ˆ Àv1 . In this section we shall study the relationship between RG-modules and CGmodules. . v n of the RG-module.2. Thus an RG-module is a vector space over R. This new vector space is clearly a CG-module (with v i g de®ned as before). Example 23. (1) V becomes an RD8 -module if we de®ne v1 a ˆ v2 . .) Although every character which can be realized over R is perforce a real character.4 Examples Let V be a 2-dimensional vector space over R. v2 a ˆ Àv1 .2. v2 b ˆ v2 (compare Example 23. 23. v2 . (2) V becomes an RC3 -module. Notice that a character ÷ of G can be . ÷ cannot be realized over R. v2 x ˆ Àv1 À v2 X   0 1 (This gives the representation x 3 of Exercise 3. with basis v1 .

5) makes VR into an RGmodule. In this way we de®ne v g for all v P VR and all g P G. v n . .Real representations 267 realized over R if and only if there exists an RG-module with character ÷. . regarding v j and iv j as elements of VR . . X X X . and (À yjk v k ‡ xjk (iv k )) (1 < j < n). 1 < j < nX It follows easily that.6. iv n X Write z jk ˆ x jk ‡ iyjk with x jk . we have (v j g)h ˆ v j ( gh) and ((iv j ) g)h ˆ (iv j )( gh)X Hence using Proposition 4. . We de®ne a multiplication of VR by g by putting (23X5) vj g ˆ (iv j ) g ˆ n ˆ kˆ1 n ˆ kˆ1 (xjk v k ‡ yjk (iv k )). and let g P G. Regarding v j as an element of the CG-module V. We summarize the basic properties of VR in the next proposition. If ÷ is the character of V. is 2 n ˆ kˆ1 xkk ˆ ÷( g) ‡ ÷( g)X Hence the character of VR is ÷ ‡ ÷. . we see that (23. Let V be a CG-module with basis v1 . y jk P R. iv1 . we have (v j g)h ˆ v j ( gh) for all g. and extending linearly to de®ne v g for all v P VR . then ÷( g) ˆ n ˆ kˆ1 zkk X The character of VR . h P G. Rather more subtle than this is the construction of an RG-module from a given CG-module. X X X . v n . There exist complex numbers z jk such that vj g ˆ n ˆ kˆ1 zjk v k (1 < j < n)X Now let VR be the vector space over R with basis v1 . evaluated at g.

namely U. where H I H I 0 0 1 0 0 1 0 0 f 0 0 0 À1 g f g g. v3 . Proof We have already proved part (1). v1 b ˆ v2 . Then VR has basis v1 . we obtain the representation r. (2) If V is an irreducible CG-module and VR is a reducible RGmodule. (1) The RG-module VR has character ÷ ‡ ÷.268 Representations and characters of groups 23. With respect to this basis. x is represented by the matrix   p À1a2 3a2 p X À 3a2 À1a2 (2) Let G ˆ D8 ˆ ka.6 Proposition Let V be a CG-module with character ÷. v2 . and so ÷ can be realized over R. iv1 . Then by part (1). v4 . and let V be the 1-dimensional CGmodule with basis v1 such that p v1 x ˆ 1(À1 ‡ i 3)v1 2 p (note that 1(À1 ‡ i 3) ˆ e2ðia3 ). bÀ1 ab ˆ aÀ1 l.7 Examples (1) Let G ˆ C3 ˆ kx: x 3 ˆ 1l. where v3 ˆ iv1 . and with 2 respect to this basis. b: a4 ˆ b2 ˆ 1. and let V be the 2dimensional CG-module with basis v1 . VR ˆ U È W where U is an RG-module with character ÷ and W is an RG-module with character ÷. with character ÷. v2 b ˆ v1 X Then VR has basis v1 . suppose that V is an irreducible CG-module and VR is a reducible RG-module. For part (2). . j 23. br ˆ f 1 0 0 0 gX ar ˆ f d À1 0 0 e d0 0 0 1e 0 0 1 0 0 0 0 1 0 The subspace of VR which is spanned by v1 ‡ v4 and v2 ‡ v3 is an RG-submodule. Therefore the character of V can be realized over R. Thus there is an RG-module. then ÷ can be realized over R. v2 a ˆ Àiv2 . in particular. dim VR ˆ 2 dim V. v4 ˆ iv2 . v2 such that v1 a ˆ iv1 .

y) are both linear ± hence the term bilinear. v) for all u. v) and y 3 â(u. v. u) for all u. v P V X And the bilinear form â is skew-symmetric if â(u. u2 . u1 . v) of F. A similar result for CG-modules was given in Exercise 8.) The bilinear form â is symmetric if â(u. 23. then there exists a G-invariant symmetric bilinear form â on V such that â(v. v1 . v. where F is R or C.Real representations 269 by Proposition 23. ë1 v1 ‡ ë2 v2 ) ˆ ë1 â(u. (Thus for ®xed u. In fact. v2 P V and ë1 . v) . Let V be a vector space over F. for all u. then a bilinear form â on V is said to be Ginvariant if â(ug. 0 for all non-zero v P V X .8 Theorem If V is an RG-module. v) ˆ Àâ(v.3(1). v). v1 ) ‡ ë2 â(u. â(u. v P V X If V is an FG-module. v2 ). v) ˆ ë1 â(u1 .6. v) of vectors in V an element â(u. we already know this from Example 23. A bilinear form â on V is a function which associates with each ordered pair (u. Bilinear forms The question of whether or not a given character can be realized over R turns out to be related to the existence of a certain bilinear form on the corresponding CG-module. and which has the following properties: â(ë1 u1 ‡ ë2 u2 . u) for all u. v) ‡ ë2 â(u2 . vg) ˆ â(u. ë2 P F.6(2). v P V and g P GX Our next result shows that every RG-module has a G-invariant symmetric bilinear form with a strong positivity property. v) ˆ â(v. the functions x 3 â(x.

v) . v P V with â(u. v) ˆ n ˆ jˆ1 vˆ ë j ì jX Then ã is a symmetric bilinear form on V. If U is an RG-submodule of V then so is . v)X xPG Therefore â is G-invariant and the theorem is proved. and â(v. v) . W ˆ fw P V : â(u. wg) ˆ â(ug À1 . wgg À1 ) ˆ â(ug À1 . Moreover. v gx) ˆ â(u. 0X j ˆ xPG ã(ux. â is a symmetric bilinear form on V. For all u P U.270 Representations and characters of groups €n jˆ1 ë j v j . ã(v. w) ˆ 0X Thus wg P W. . v n be a basis of V. de®ne ã(u. for non-zero v P V. 0. and that there exist u.9 Proposition Let V be an RG-module and let â be a G-invariant bilinear form on V. . then gx runs through G as x runs through G. . so â(u. we have ugÀ1 P U. If g P G. u) . so W is an RG-submodule of V. and hence ˆ â(ug. j 23. v P V )X Again. vx) (u. 0 and â(v. Now let w P W and g P G. ì j P R. Proof Let v1 . v) ˆ n ˆ jˆ1 ì2 . For u ˆ €n jˆ1 ì j v j P V with ë j . j 23. . 0 for all non-zero v P V. v) ˆ Now let â(u. . w) ˆ 0 for all u P U gX Proof It is easy to see that W is a subspace of V. Then V is a reducible RG-module. vg) ˆ ã(ugx.10 Proposition Suppose that â is a G-invariant symmetric bilinear form on the RGmodule V.

so is ã. then W is non-zero. and is an RG-submodule of V by Proposition 23. €n But for all v ˆ iˆ1 ë i v i P V (ë i P R). v) À â(u. 1 ã(v2 . v2 ) . and a non-zero Ginvariant symmetric bilinear form on V . for all i. j if i Tˆ j. v) (u. . v2 ) . 23. w) .7). v2 ) ˆ 1 À â(v2 . (2) there exists a CG-module V with character ÷. v i ) ˆ 1 â(v1 .8 supplies us with a G-invariant symmetric bilinear form â1 on V such that â1 (w. v1 ) ˆ x. 0 for all non-zero w P V X By a general result on bilinear forms (see Exercise 23. We can now relate bilinear forms to the question of whether or not a given character of G can be realized over R. v j ) ˆ 0 and â1 (v i . there is a basis v1 .Real representations 271 Proof Theorem 23. 0. . v P V )X x Since â and â1 are G-invariant symmetric bilinear forms on V. x so W Tˆ V. . v j ) ˆ â(v i .9. and de®ne ã by 1 ã(u. â(v2 . 0X Let â(v1 .11 Theorem Let ÷ be an irreducible character of G. . The following two conditions are equivalent: (1) ÷ can be realized over R. w) ˆ 0 for all v P V g. . v) ˆ â1 (u. v1 ) ˆ ë1 ã(v1 . Therefore V is a reducible RG-module. if we de®ne W ˆ fw P V : ã(v. 0. we have ã(v. v n of V such that â1 (v i . Moreover. v1 ) . v1 ) ˆ 0X Therefore.

X X X . ì j P R)X jˆ1 jˆ1 jˆ1 Then W is a bijection.6(2) that ÷ can be realized over R. suppose that ÷ can be realized over R. X X X .8. ìk uk ˆ ë j ì k ã(u j . all ë P R and all g P G. iv1 ) ˆ À1X Therefore VR is a reducible RG-module. Let â(w. Then â(v1 . u n be a basis of U.12). Conversely. . we have (23X12) (w1 ‡ w2 )W ˆ w1 W ‡ w2 W. Notice that ~ ~ â(v1 . v P VR . Then v1 . . . iv1 . Let V be a CG-module with character ÷. De®ne a function W from VR to V by n n n ˆ ˆ ˆ W: ë jv j ‡ ì j (iv j ) 3 (ë j ‡ iì j )v j (ë j . ” De®ne ã on V by 2 n 3 n n n ˆ ˆˆ ˆ ” ã ë j u j. v).10. (ëv)W ˆ ë(vW). . . v1 ) ˆ 1X Extend v1 to a basis v1 . v n . u) ‡ â(v. . there is a non-zero Ginvariant symmetric bilinear form ã on U. there exists w P V with â(w. w) ˆ z and v1 ˆ z À1a2 w. . u n . and let V be the vector space over C with basis u1 . There exist u. by Proposition 23. v n of V. . that â is a Ginvariant symmetric bilinear form on VR . and for all w1 . . . (v g)W ˆ (vW) gX ~ Now de®ne a function â on ordered pairs of elements of VR by ~ â(u. w2 . u k ) jˆ1 kˆ1 jˆ1 kˆ1 . Let u1 . v) ˆ â(v. u ‡ v) ˆ â(u. Since â(u ‡ v. v1 ) ˆ 1 and â(iv1 . v P V with â(u. and suppose that â is a non-zero G-invariant symmetric bilinear form on V. using the properties (23. iv n is a basis of the RG-module VR . v P VR )X ~ You can readily check. . vW) (u. V is a CG-module (with u i g de®ned as for U). v) ˆ the real part of â(uW. v) ‡ 2â(u. This establishes that (2) implies (1) in the theorem. . It now follows from Proposition 23. u) Tˆ 0. By Theorem 23. and let U be an RG-module with character ÷. As explained earlier.272 Representations and characters of groups Proof We ®rst show that (2) implies (1). w) Tˆ 0.

Then ã is a non-zero G-invariant symmetric bilinear form on the CG-module V. Thus (1) implies (2). 1 or À1.Real representations 273 ” (where ë j . Recall from Chapter 19 that ÷ 2 is the character of the CG-module V  V. then we de®ne the indicator é÷ of ÷ by V if ÷ is not a constituent of ÷ S or ÷ A . 1 G i ˆ ÷( g)÷( g) ˆ h÷. b ` é÷ ˆ 1. if 1 G is a constituent of ÷ S . We shall see later that this number tells us whether or not ÷ can be realized over R. and V has character ÷. relating it to the internal structure of the group G. b b X À1. and ÷2 ˆ ÷S ‡ ÷ A . Observe that 1 ˆ h÷ 2 . if ÷ is realX Let V be a CG-module with character ÷. if ÷ is not real. ÷iX jGj gPG Therefore. ì k P C). j The indicator function We now associate with each irreducible character ÷ of G a certain number. b 0. . Note that é÷ Tˆ 0 if and only if ÷ is real. where ÷ S is the character of the symmetric part of V  V. then precisely one of ÷ S and ÷ A has 1 G as a constituent. for irreducible characters ÷. we have @ 0. 1 G i ˆ 1. which is always 0. and ÷ A is the character of the antisymmetric part of V  V. called the indicator of ÷. 23. if 1G is a constituent of ÷ A X We call é the indicator function on the set of irreducible characters of G. 2 h÷ . Hence if h÷ 2 .13 De®nition If ÷ is an irreducible character of G. 1 G i ˆ 1. The next result gives a signi®cant property of the indicator function. and the proof of the theorem is complete.

23. Proof De®ne a function W: G 3 C by W(x) ˆ jf y P G: y 2 ˆ xgj (x P G)X Note that W is a class function on G. since for g P G we have y 2 ˆ x D ( g À1 yg)2 ˆ g À1 xgX Therefore by Corollary 15. 1 G i 1 ˆ ˆ ÷( g 2 ) by Proposition 19X14 jGj gPG ˆ ˆ 1 ˆ ˆ ÷( g 2 ) jGj xPG gPG: g2 ˆx 1 ˆ W(x)÷(x) jGj xPG ˆ hW. The de®nition of é÷ gives é÷ ˆ h÷ S À ÷A . The character table of G is 1 ÷1 ÷2 ÷3 1 1 2 (1 2) 1 À1 0 (1 2 3) 1 1 À1 j . Representations and characters of groups ˆ (é÷)÷(x) ˆ jf y P G: y 2 ˆ xgj. ÷iX € Therefore. and the result follows. W ˆ (é÷)÷. W is a linear combination of the irreducible characters of G.274 23.15 Example Let G ˆ S3 . ÷ where the sum is over all irreducible characters ÷ of G.4.14 Theorem For all x P G.

squares to be (1 2 3). four elements of G square to be 1. By Proposition 8. (1 2). and deduce the Frobenius±Schur Count of Involutions. Back to reality We now relate the indicator function to the previous material on bilinear forms. namely 1. and de®ne a multiplication of C by elements of G by ëg ˆ ë (ë P C. Using this. in accordance with Theorem 23. g P G)X In this way. (3) There exists a non-zero G-invariant skew-symmetric bilinear form on V if and only if é÷ ˆ À1.16 Theorem Let V be an irreducible CG-module with character ÷. and one element. there is a non-zero CG-homomorphism from V  V onto this trivial CG-submodule.8. (1) Suppose that é÷ Tˆ 0.14 we calculate that é÷ ˆ 1 for each irreducible € character ÷ of G.14. . Then 1 G is a constituent of ÷ 2 .Real representations 275 Using Proposition 19. and hence the CG-module V  V has a trivial CG-submodule. 23. (1 3) and (2 3). (1) There exists a non-zero G-invariant bilinear form on V if and only if é÷ Tˆ 0. and hence there is a non-zero CG-homomorphism W from V  V onto the trivial CG-module C. C becomes a trivial CG-module. we show that the indicator of an irreducible character determines whether or not it can be realized over R. (1 3 2). no elements square to be (1 2). (2) There exists a non-zero G-invariant symmetric bilinear form on V if and only if é÷ ˆ 1. Proof In this proof we regard C as a 1-dimensional vector space over C. which takes the following values: 1 ÷ 1 ‡ ÷2 ‡ ÷3 4 (1 2) 0 (1 2 3) 1 Sure enough. so (é÷)÷ ˆ ÷1 ‡ ÷2 ‡ ÷3 .

v)X Thus â is G-invariant. V  V has a trivial CGsubmodule. j < n) and extending linearly. and W is a non-zero CG-homomorphism from S(V  V) onto the trivial CG- . vg) ˆ (ug  v g)W ˆ ((u  v) g)W ˆ ((u  v)W) g ˆ (u  v)W ˆ â(u. Conversely. X X X . we have â(ug. Conversely. Let v1 . and for u. v j ) as â is G-invariant ˆ (v i  v j )WX Hence W is a non-zero CG-homomorphism from V  V onto the trivial CG-module C. .8 that there is a non-zero CGhomomorphism W from S(V  V) onto the trivial CG-module C. we have ((v i  v j ) g)W ˆ (v i g  v j g)W ˆ â(v i g. Since â is symmetric. . 1 < j < n) and extending linearly to the whole of V  V.1. suppose that there is a non-zero G-invariant bilinear form â on V. De®ne â(u. so that v i  v j (1 < i < n. W is well-de®ned. Then 1 G is a constituent of ÷ S . by Proposition 10. Let v1 . suppose that there exists a non-zero G-invariant symmetric bilinear form â on V. it follows by Proposition 8. v j g) ˆ â(v i . v n be a basis of V. . v P V )X Then â is a non-zero bilinear form on V. v n be a basis of V. the symmetric part of V  V. 1 < j < n) form a basis of V  V. As in (1). For g P G. (2) Suppose that é÷ ˆ 1. v P V and g P G.276 Representations and characters of groups Now de®ne â by â(u. and therefore é÷ Tˆ 0. v j ) (1 < i < n. v) ˆ (u  v ‡ v  u)W (u. and de®ne W: S(V  V) 3 C by putting (v i  v j ‡ v j  v i )W ˆ â(v i . Thus. v) ˆ (u  v)W (u. v P V )X Then â is a non-zero G-invariant symmetric bilinear form on V. De®ne W: V  V 3 C by putting (v i  v j )W ˆ â(v i . which is the character of the CG-module S(V  V). . If follows that ÷ 2 has the trivial character 1 G as a constituent. v j ) (1 < i.

Real representations

277

module C. Hence ÷ S has the trivial character 1 G as a constituent, and so é÷ ˆ 1. (3) The proof of (3) is very similar to that of (2), and is omitted. j We can now relate real representations of G to involutions in G, where by an involution we mean an element of order 2. 23.17 Corollary (The Frobenius±Schur Count of Involutions) For each irreducible character ÷ of G, we have V if ÷ is not real, b 0, b ` é÷ ˆ 1, if ÷ can be realized over R, b b X À1, if ÷ is real, but ÷ cannot be realized over RX Moreover, for all x P G, ˆ (é÷)÷(x) ˆ jf y P G: y 2 ˆ xgj,
÷

where the sum is over all irreducible characters ÷ of G. In particular, ˆ (é÷)÷(1) ˆ 1 ‡ t,
÷

where t is equal to the number of involutions in G. Proof When we de®ned the indicator function, we showed that é÷ Tˆ 0 if and only if ÷ is real. And Theorems 23.11 and 23.16(2) show that ÷ can be realized over R if and only if é÷ ˆ 1. This proves that é÷ is determined as in the statement of the corollary. € The expression for ÷ (é÷)÷(x) was obtained in Theorem 23.14. € Putting x ˆ 1, we see that ÷ (é÷)÷(1) is equal to the number of elements y of G satisfying y 2 ˆ 1. These elements are just the involutions in G, together with the identity, so the number of them is precisely 1 ‡ t. j We conclude with some examples illustrating the use of Corollary 23.17. 23.18 Examples (1) Let ÷ be a linear character. Then é÷ ˆ 1 if ÷ is real, and é÷ ˆ 0 if ÷ is non-real.

278

Representations and characters of groups

For an abelian group, the Frobenius±Schur Count of Involutions shows that the number of real irreducible (linear) characters is equal to the number of real conjugacy classes, since in this case g is conjugate to gÀ1 if and only if g2 ˆ 1. This special case of Theorem 23.1 can be proved directly without much dif®culty (see Exercise 23.2). (2) We know that all the irreducible characters of D8 ˆ ka, b: a4 ˆ b2 ˆ 1, bÀ1 ab ˆ aÀ1 l can be realized over R (see Example 23.3(1), and note that all four linear characters are real). Thus é÷ ˆ 1 for all irreducible characters ÷ of D8, and so ˆ (é÷)÷(1) ˆ 1 ‡ 1 ‡ 1 ‡ 1 ‡ 2 ˆ 6X
÷

The ®ve involutions in D8 which are predicted by the Frobenius± Schur Count of Involutions are a2 , b, ab, a2 b and a3 b. (3) In the group Q8 ˆ ka, b: a4 ˆ 1, a2 ˆ b2 , bÀ1 ab ˆ aÀ1 l, there is just one involution, namely a2 . Now é÷ ˆ 1 for each of the four linear characters, and ˆ (é÷)÷(1) ˆ 2
÷

by the Frobenius±Schur Count of Involutions. Therefore, if ø is the irreducible character of degree 2, then éø ˆ À1. In particular ø cannot be realized over R. (4) The symmetric group S4 has ten elements whose square is 1, namely the identity, the six elements which are conjugate to (1 2), and the three elements which are conjugate to (1 2)(3 4). Since the degrees 1, 1, 2, 3, 3 of the irreducible characters of S4 sum to be 10 (see Section 18.1), we see that all the characters of S4 can be realized over R. The Brauer±Fowler Theorem We now apply Corollary 23.17 to give a proof of a famous theorem of Brauer and Fowler. 23.19 Brauer±Fowler Theorem Let n be a positive integer. Then there exist only ®nitely many nonisomorphic ®nite simple groups containing an involution with centralizer of order n.

Real representations

279

Despite its fairly elementary proof, this result is of great historical importance in ®nite group theory. It led Brauer to propose the programme of determining, for each ®nite group C, all the simple groups G possessing an involution u with CG (u)  C. This programme was the start of the modern attempt to classify all ®nite simple groups, which was ®nally completed in the early 1980s. For further information about this, see the book by D. Gorenstein listed in the Bibliography. In Exercise 10 at the end of the chapter you are asked to carry out Brauer's programme in the case where C  C2 . This should not trouble you too much. In Chapter 30, Theorem 30.8, you will ®nd a much more sophisticated case, in which C  D8. For proof of Theorem 23.19 we require two preliminary lemmas. 23.20 Lemma € € If a1 , . . . , a n are real numbers, then a2 > … a i †2 an. i Proof This follows from the Cauchy±Schwarz inequality kvk kwk > jvXwj, taking v ˆ (a1 , . . . , a n ) and w ˆ (1, . . . , 1). j 23.21 Lemma Let G be a group of even order m, and let t be the number of involutions in G (so t . 0 by Exercise 8 of Chapter 1). Write a ˆ (m À 1)at. Then G contains a non-identity element x such that jG : C G (x)j < a2 . Proof By Corollary 23.17, we have ˆ t< ÷(1)
÷

where the sum is over all non-trivial irreducible characters ÷ of G. Writing k for the number of irreducible characters of G, we deduce using Lemma 23.20 and Theorem 11.12 that ˆ ˆ t2 < ( ÷(1))2 < (k À 1) ÷(1)2 ˆ (k À 1)(m À 1),
÷

and hence m À 1 < (k À 1)(m À 1)2 at 2 ˆ (k À 1)a2 . Now k À 1 is the number of non-identity conjugacy classes of G. If every nonidentity conjugacy class has size more then a2 , then (k À 1)a2 .

280

Representations and characters of groups

jGj À 1 ˆ m À 1, a contradiction. Therefore some non-identity class x G has size at most a2 . Then jG : C G (x)j < a2 , giving the result. j Proof of Theorem 23.19 Suppose G is simple and contains an involution u such that jC G (u)j ˆ n. Let jGj ˆ m, and let t be the number of involutions in G. By Proposition 12.6, every element of the conjugacy class u G is an involution, and hence t > ju G j ˆ jG : C G (u)j ˆ manX Therefore (m À 1)at , n, and so by Lemma 23.21, there is a nonidentity element x P G such that jG : C G (x)j , n2 . Let H ˆ C G (x). If H ˆ G then x lies in Z(G), the centre of G, which is a normal subgroup of G. Since G is simple it follows that G ˆ Z(G), so G is abelian and therefore G  C2 . Now suppose that H Tˆ G. Write r ˆ jG : Hj, so r , n2 . By Exercise 9 at the end of the chapter, there is a non-trivial homomorphism è from G to the symmetric group S r . As G is simple, the normal subgroup Ker è ˆ f1g. Thus G is isomorphic to a subgroup of S r , hence of S n2 . In particular, given n, there are only ®nitely many possibilities for G. j

Summary of Chapter 23 1. The number of real irreducible characters of G is equal to the number of real conjugacy classes of G. Let é be the indicator function, and let ÷ be an irreducible character of G. V if ÷ is non-real, b 0, b b b 1, ` if there exists an RG-module U with character ÷, 2X é÷ ˆ with character ÷ b b b b X À1, if ÷ is real, but there does not exist an RG-moduleX V if 1 G is not a constituent of ÷ S or ÷ A , b 0, ` 3X é÷ ˆ 1, if 1 G is a constituent of ÷ S , b X À1, if 1 G is a constituent of ÷ A X € 2 4. ÷ (é÷)÷(1) ˆ |{ g P G: g ˆ 1}|.

Real representations Exercises for Chapter 23

281

1. Prove that if G is a group of odd order then no non-identity element of G is real. 2. Let G be a ®nite abelian group. Use the description of the irreducible characters of G, given in Theorem 9.8, to prove directly that the number of real irreducible characters of G is equal to the number of elements g in G for which g2 ˆ 1. 3. Let G ˆ D2 n and consult Section 18.3 for the character table of G. How many elements g of G satisfy g2 ˆ 1? Deduce that é÷ ˆ 1 for all irreducible characters ÷ of G. 4. Let r be an irreducible representation of degree 2 of a group G, and let ÷ be the character of r. Prove that ÷ A ( g) ˆ det ( gr) for all g P G. Deduce that é÷ ˆ À1 if and only if det ( gr) ˆ 1 for all g P G. 5. Let G ˆ T 4 n ˆ ha, b: a2 n ˆ 1, a n ˆ b2 , bÀ1 ab ˆ aÀ1 i, as in Exercise 17.6. Let V be a 2-dimensional vector space over C with basis v1 , v2 and let å be a (2n)th root of unity in C with å Tˆ Æ1. Exercise 17.6 shows that V becomes an irreducible CG-module if we de®ne v1 a ˆ åv1 , v2 a ˆ å À1 v2 , v1 b ˆ v2 , v2 b ˆ å n v1 X

Let ÷ be the character of this CG-module V. (a) Note that å n ˆ Æ1. Use Exercise 4 to show that é÷ ˆ 1 if å n ˆ 1 and é÷ ˆ À1 if å n ˆ À1. (b) Let â be the bilinear form on V for which â(v1 , v1 ) ˆ â(v2 , v2 ) ˆ 0, â(v1 , v2 ) ˆ 1, â(v2 , v1 ) ˆ å n X Prove that the bilinear form â is G-invariant, and use Theorem 23.16 to provide a second proof that é÷ ˆ 1 if å n ˆ 1 and é÷ ˆ À1 if å n ˆ À1. (c) Prove that a n is the only element of order 2 in T4 n. (d) Use the character table of G, which appears in the solution to Exercise 18.3, to ®nd é÷ for each irreducible character ÷ of G. Check that

282

Representations and characters of groups ˆ (é÷)÷(1) ˆ 2,
÷

in agreement with the Frobenius±Schur Count of Involutions. 6. Prove that if ÷ is an irreducible character of a group G, and é÷ ˆ À1, then ÷(1) is even. (Hint: the solution uses a well known result about skew-symmetric bilinear forms.) 7. Suppose that V is a vector space over R and that â1 and â are symmetric bilinear forms on V. Assume that â1 (w, w) . 0 for all non-zero w in V. Prove that there is a basis e1 , X X X , e n of V such that â1 (ei , ei ) ˆ 1 for all i, and â1 (ei , ej ) ˆ â(ei , ej ) ˆ 0 for all i Tˆ jX 8. Schur's Lemma is crucial for the development of the theory of CG-modules. This exercise indicates the extent to which results like Schur's Lemma hold for RG-modules. Let V and W be irreducible RG-modules. (a) Prove that if W: V 3 W is an RG-homomorphism then either W is an RG-isomorphism or vW ˆ 0 for all v P V. (b) Prove that if W: V 3 V is an RG-isomorphism and V remains irreducible as a CG-module, then W ˆ ë1 V for some real number ë. . (c) Give an example of a group G, an irreducible RG-module V and an RG-homomorphism W: V 3 V which is not a multiple of 1 V . 9. Let G be a group with a subgroup H of index n. Let Ù be the set of n right cosets Hx of H in G. For g P G, de®ne a function r g : Ù 3 Ù by ( Hx)r g ˆ Hxg for all x P G. Prove that r g is a permutation of Ù, and that the function r : g 3 r g is a homomorphism from G to the symmetric group on Ù. „ Show that the kernel of r is xPG x À1 Hx. Deduce that if a group G has a subgroup H of index n, then there is a homomorphism G 3 S n with kernel contained in H. 10. Suppose that G is a ®nite group containing and involution t with CG (t)  C2 . Prove that |G : G9| ˆ 2. Deduce that if G is simple, then G  C2 .

24 Summary of properties of character tables

In this short chapter we present no new results, but instead we gather together from previous chapters various properties which are helpful when we try to ®nd the character table of a particular group. In the next four chapters we shall calculate several character tables in detail. Usually we begin by working out the conjugacy classes and centralizer orders of our given ®nite group G. The size of the character table is determined by the number k of conjugacy classes of G; the character table is then a k 3 k matrix, with columns indexed by the conjugacy classes of G (the ®rst column corresponding to the conjugacy class {1}), and with rows indexed by the irreducible characters of G. When doing calculations, we commonly come across a new character ÷, which may or may not be irreducible. We can then calculate h÷, ÷i, which is given by h÷, ÷i ˆ 1 ˆ ÷( g)÷( g)X jGj gPG

The character ÷ is irreducible if and only if h÷, ÷i ˆ 1 (see Theorem 14.20). If ÷ is reducible then we calculate h÷, ÷ i i for each of the irreducible characters ÷ i which we already know, and then ÷À ˆ h÷, ÷ i i÷ i
i

will also be a character. We can thus determine whether ÷ is a linear combination of the irreducible characters we already know; and if it is not, then we can obtain from ÷ a linear combination of irreducible characters, all of which are new. We have developed a number of methods for producing characters ÷ 283

284

Representations and characters of groups

on which to perform such calculations. For example, every subgroup of Sn has a permutation character (see (13.22)); the product of two characters is again a character (Proposition 19.6); given a character ø we can form the symmetric and antisymmetric parts of its square, ø S and ø A (see Proposition 19.14); and if H is a subgroup of G then we can restrict characters of G to H, and induce characters of H to G. These and other properties of characters are summarized in the following list.

Properties of characters Assume that ÷1 , . . . , ÷ k are the irreducible characters of G. (1) (Example 13.8(3)) There is a (trivial) character ÷ of G which is given by ÷( g) ˆ 1 for all g P GX (2) (Theorem 17.11) The group G has precisely jGaG9j linear characters. These are the characters ÷ given by ÷( g) ˆ ø( gG9) ( g P G) as ø varies over the irreducible (linear) characters of GaG9. (3) (Theorem 17.3) As a generalization of (2), if N v G and ø is an irreducible character of GaN , then we get an irreducible character ÷ of G which is given by ÷( g) ˆ ø( gN ) ( g P G) (÷ is the lift of ø). This method gives precisely those irreducible characters of G which have N contained in their kernel. (4) (Theorem 19.18) If G ˆ G1 3 G2 then all the irreducible characters ÷ of G are given by ÷( g1 , g 2 ) ˆ ö1 ( g 1 )ö2 ( g 2 ) ( g 1 P G1 , g 2 P G2 ), as ö i varies over the irreducible characters of Gi (i ˆ 1, 2). (5) (Proposition 13.24) If G is a subgroup of Sn , then the function í: G 3 C de®ned by í( g) ˆ jfix ( g)j À 1 is a character of G. ( g P G)

Summary of properties of character tables

285

(6) (Theorems 11.12 and 22.11) The entries ÷ i (1) (1 < i < k) in the ®rst column of the character table of G are positive integers, and satisfy
k ˆ iˆ1

÷ i (1)2 ˆ jGjX

Moreover, each integer ÷ i (1) divides |G|. (7) (Row orthogonality relations, Theorem 16.4(1)) For all i, j, we have h÷ i , ÷ j i ˆ ä ij . (8) (Column orthogonality relations, Theorem 16.4(2)) For all g, h P G, we have @ k ˆ jCG ( g)j, if g and h are conjugate, ÷ i ( g)÷ i (h) ˆ 0, otherwiseX iˆ1 (9) (Exercise 13.5) If ÷ is an irreducible character of G and z P Z(G), then there exists a root of unity å such that for all g P G, ÷(zg) ˆ å÷( g)X (10) (Proposition 13.9(2)) If g is an element of order n in G, and ÷ is a character of G, then ÷( g) is a sum of nth roots of unity. Moreover, |÷( g)| < ÷(1). (11) (Proposition 13.9(3, 4)) If g P G and ÷ is a character of G, then ÷( g À1 ) ˆ ÷( g)X In particular, if g is conjugate to gÀ1 then ÷( g) is real for all characters ÷ of G. (12) (Corollary 15.6) If g P G and g is not conjugate to gÀ1 , then ÷( g) is non-real for some character ÷ of G. (13) (Theorem 22.16) Let g P G. If g is conjugate to gi for all positive integers i which are coprime to the order of g, then ÷( g) is an integer for all characters ÷ of G. (14) (Corollary 22.26) Suppose that p is a prime number and that y is the p9-part of the element g of G. If ÷ is a character of G such that ÷( g) and ÷( y) are both integers, then ÷( g)  ÷( y) mod pX

then ÷ë is an irreducible character of G. For example. indeed. where (ø 5 H)(h) ˆ ø(h) (h P H)X We have seen that the character table of a group G gives grouptheoretic information about G. We can see from the character table whether or not G is simple (Proposition 17. Proposition 21.15) If ÷ is an irreducible character of G. these can be determined in the following ways. The centre Z(G) can be found by noting which elements g of G satisfy . then so is the product ÷ø. 2 ÷ A ( g) ˆ 1(÷ 2 ( g) À ÷( g 2 ))X 2 (20) (De®nition 21. then so is ÷. ÷ S ( g) ˆ 1(÷ 2 ( g) ‡ ÷( g 2 )).23. then ø 5 H is a character of H. then so are ÷ S and ÷ A .1) The number of real irreducible characters of G is equal to the number of real conjugacy classes of G. (21) (Chapter 20) If H is a subgroup of G and ø is a character of G. where ÷( g) ˆ ÷( g) ( g P G)X (16) (Theorem 23.6). then ÷( g)  ÷(1) mod pX (15) (Proposition 13. The derived subgroup G9 consists of those elements g in G which satisfy ÷( g) ˆ 1 for all linear characters ÷ of G.23) If H is a subgroup of G and ø is a character of H. the ®rst column determines |G| and jGaG9j (by (6) and (2)).14) If ÷ is a character of G.14) If ÷ is an irreducible character of G and ë is a linear character of G.286 Representations and characters of groups In particular. (17) (Proposition 17. where for all g P G. Two important normal subgroups are G9 and Z(G). where ÷ë( g) ˆ ÷( g)ë( g) ( g P G)X (18) (Proposition 19. where ÷ø( g) ˆ ÷( g)ø( g) ( g P G)X (19) (Proposition 19.5). then ø 4 G is a character of G. if the order of g is a power of p. with values given by Proposition 21. we can ®nd all the normal subgroups of G (Proposition 17.6) If ÷ and ø are characters of G.13.

1 we gave examples of non-isomorphic groups. D8 and Q8 .Summary of properties of character tables 287 € ÷( g)÷( g) ˆ |G|. the sum being over all irreducible characters ÷ of G. however. with the same character table. In Chapter 30 we shall see some more impressive results about subgroups of G. As a ®nal remark. it is of course true that isomorphic groups have the same character table. which can be deduced from the character table. . the converse is false: in Exercise 17.

that is. and in this chapter we shall describe a class of Frobenius groups and ®nd the character tables of the groups in this class. and n r T 1 mod p for 0 . and Zà ˆ Z p À f0g is an abelian group p under multiplication. In particular. but for a good 288 . r . A number of these groups are so-called Frobenius groups. Z p is an abelian group under addition.25 Characters of groups of order pq By the end of the next chapter. that Zà is cyclic: p 25. but not at all obvious. with addition and multiplication modulo p. that is. p will denote a prime number. we shall have determined the character tables of all groups of order less than 32. generated by 1. We shall not provide a proof of Theorem 25. p À 1g. there exists an integer n p such that n pÀ1  1 mod p. It is also true. is a ®eld. Clearly Z p is a cyclic group under addition. X X X . Primitive roots modulo p Recall that the set Z p ˆ f0.1. this will give the character tables of all groups whose order is the product of two prime numbers. Throughout the chapter. p À 1X An integer n of order p À 1 in Zà is called a primitive root modulo p p.1 Theorem The multiplicative group Zà is cyclic. 1.

Fraleigh listed in the Bibliography. 0 1 and so we have the relations (25X5) Ap ˆ Bq ˆ I. De®ne p     1 1 1 0 Aˆ . 0 1 0 u and let F ˆ hA. and let u be an element of order q in the multiplicative group Zà . we see that every element of F is of the form A i B j with 0 < i < p À 1. B. and u r T 1 mod p for 0 . 11 and 13. BÀ1 AB ˆ Au X Using these relations.2 Example The number 2 is a primitive root modulo 3.3 of the book by J. where q| p 2 1 25. Bˆ . As an immediate consequence of Theorem 25. such that u q  1 mod p. Now let q| p À 1. Bi.4 Example De®ne Gˆ & 1 0  ' y Ã. but not modulo 7. These pq elements are dis- . 5.Characters of groups of order pq 289 account. the subgroup of G generated by A and B.1). y P Z X : xPZp p x Under matrix multiplication.1 we have 25. 25.3 Proposition If q| p À 1 then there is an integer u such that u has order q modulo p ± that is. r . 0 < j < q À 1. qX Frobenius groups of order pq. G is a group of order p( p À 1) (see Exercise 25. and 3 is a primitive root modulo 7. Then   1 u À1 B AB ˆ ˆ Au . we refer you to Theorem 45.

Proof Assume that G is non-abelian. BÀ1 AB ˆ Au iX 25.q . Fraleigh listed in the Bibliography).q . 25. p It is not hard to show that. The groups F p.3). a ˆ bÀq abq ˆ a u q and so u q  1 mod p. (Alternatively. so we have the presentation F ˆ hA.5) determine all products in F.q does not depend on which integer u of order q we choose (see Exercise 25.q ˆ ha. Then either G is abelian. B: Ap ˆ Bq ˆ I.) Both H and Ga H are cyclic.6 De®nition If p is a prime and q| p À 1. so jFj ˆ pq. S. Now H v G.7 Proposition Suppose that G is a group of order pq. where u is an element of order q in Zà . b: a p ˆ b q ˆ 1. bÀ1 ab ˆ au i. the interested reader can ®nd more information in the book by D. Further. or q divides p À 1 and G  F p. Moreover the relations (25. where p and q are prime numbers with p . q. then we write F p. so bÀ1 ab ˆ au for some integer u. up to isomorphism. it follows that b has order q. p .q for the group of order pq with presentation F p. then G is generated by a and b. Since bq P H but b does not have order pq (as G is non-abelian). Suppose that H ˆ kal and Ga H ˆ h Hbi. F p.290 Representations and characters of groups tinct.q belong to a wider class of groups known as Frobenius groups. Passman listed in the Bibliography. We shall not give the general de®nition of these here. as we shall only be dealing with F p. B. The next result classi®es all groups whose order is the product of two distinct prime numbers. these facts follow readily from Sylow's Theorems (see Chapter 18 of the book by J. since they have prime order. It follows from Exercise 22. Thus the order of u in the group Zà divides q.3 that q divides p À 1 and G has a normal subgroup H of order p.

q in general. Hence (av i ) G has size q. and u has order q modulo p. and since kal < CG (av i ). Let S be the subgroup of Zà consisting of the powers of u. .Characters of groups of order pq 291 If the order of u were 1 then we would have bÀ1 ab ˆ a. and choose coset representatives v1 . isomorphic to C3 3 C5 ). Thus p jSj ˆ q.3. (bn ) G ˆ fam bn : 0 < m < p À 1g (1 < n < q À 1)X Proof The equation bÀ j av b j ˆ avu j shows that av is conjugate to avs for all s P S.8 Example By Proposition 25. and in Example 21. The character table of F p. We have now established that a p ˆ b q ˆ 1.7. X X X . Write r ˆ ( p À 1)aq.q are f1g. p 25. and has the form stated in the proposition.9 Proposition The conjugacy classes of G ˆ F p.q ˆ ha. b: ap ˆ bq ˆ 1.q : the dihedral group of order 2 p is the case where q ˆ 2. bÀ1 ab ˆ au . (av i ) G ˆ fav i s : s P Sg (1 < i < r). also the size of this conjugacy class is equal to |G: CG (av i )|. Therefore the conjugacy class of av i has size at least q.q . and G would be abelian. in fact. and the groups of order 21 are C3 3 C7 and F7. Thus let G ˆ Fp.3 . Therefore the order of u is q. every group of order 15 is abelian (indeed.q We have. j 25. this size is at most q. bÀ1 ab ˆ au i where p is prime. already found the character tables of certain of the groups F p. q| p À 1 (q not necessarily prime).25 we dealt with F7. order of u in Zà is qX p Hence G  F p. We now construct the character table of F p. v r for S in Zà .

and kbl has index p in G. j By Proposition 25. observe that the derived subgroup G9 ˆ kal. øv j s ihai ˆ hö j . hö j 5 hai. we have |CG (bn )| ˆ q. so GaG9 has order q and therefore by Theorem 17. let p ö j ˆ øv j 4 GX We now prove that each ö j is irreducible. so we seek q ‡ r irreducible characters. every conjugate of bn has the form am bn for some m.16. denote by øv the character of kal which p is given by øv (ax ) ˆ å vx (0 < x < p À 1)X We calculate the values of the induced character øv 4 G. G has q ‡ r conjugacy classes. G has precisely q linear characters. These are given by ÷ n (0 < n < q À 1).292 Representations and characters of groups Since CG (bn ) contains kbl. On the other hand. First. ö j i G X Hence ö j 5 hai ˆ hö j . By the Frobenius Reciprocity Theorem 21. and so the conjugacy class of bn has size p. and øv 4 G ˆ øvs 4 G if s P SX For each coset representative v j (1 < j < r) of S in Zà . For v P Zà . øv j s 4 Gi G ˆ hö j . using Proposition 21. as Gahai is abelian. Let å ˆ e2ðia p . it follows that for n T 0 mod q. .9. and ˆ (øv 4 G)(ax ) ˆ å vsx (0 < x < p À 1)X sPS Note that øv 4 G has degree q. We obtain (øv 4 G)(ax by ) ˆ 0 if 1 < y < q À 1. ö j i G ˆ sPS øv j s ‡ ÷. 0 < y < q À 1)X We shall show that G has r irreducible characters of degree q.11. Hence (bn ) G ˆ fam bn : 0 < m < p À 1g and the proof is complete.23. where ÷ n (a x b y ) ˆ e2ði nyaq (0 < x < p À 1. for all s P S.

25. ö j of G (0 < n < q À 1. . .Characters of groups of order pq 293 where ÷ is either 0 or a character of kal. we deduce that kö j . ˆ ö j (ax ) ˆ e2ðiv j sxa p . Of these. ö r are distinct. so we have the complete character table of G. bÀ1 ab ˆ au i . and also that ˆ ö j 5 hai ˆ hö j . sPS for 1 < j < r. . This proves that ö j is irreducible. v r S are the cosets in Zà of the p subgroup S generated by u. ö j i G øv j s X sPS By Theorem 14.10 Theorem Let p be a prime number. X X X . 0 < y < q À 1g has q ‡ r irreducible characters. 25. Taking degrees. b: a p ˆ b pÀ1 ˆ 1. ö j i G X Since ö j (1) ˆ q ˆ jSj. q| p À 1 and r ˆ ( p À 1)aq. ö j l G ˆ 1.10 in some examples. . 1 < j < r). We have now found q ‡ r distinct irreducible characters ÷ n . bÀ1 ab ˆ au i ˆ fax by : 0 < x < p À 1. We summarize in the following theorem. . the characters øv (v P Zà ) are linearly indepenp dent. . . We conclude by illustrating Theorem 25. Consequently the irreducible characters ö1 . . where v1 S. b: a p ˆ bq ˆ 1. it follows that ö j (1) > jSjhö j . and hence ö1 5 kal. Then the group F p.11 Example Let G ˆ F p. pÀ1 ˆ ha.23. q have degree 1 and are given by ÷ n (ax by ) ˆ e2ði nyaq (0 < n < q À 1) and r have degree q and are given by ö j (ax by ) ˆ 0 if 1 < y < q À 1. ö r 5 kal are distinct.q ˆ ha.

b ˆ (2 3 5 4)X Check that a5 ˆ b4 ˆ 1. In Example 21.4 is as shown opposite. Here F13.3. the character table of F13.4 .4 ˆ ha. then G  F5. bÀ1 ab ˆ a2 X Hence if G ˆ ka. ⠈ å 2 ‡ å 3 ‡ å 10 ‡ å 11 .10. You may like . 㠈 å4 ‡ å6 ‡ å7 ‡ å9 X By Theorem 25. bÀ1 ab ˆ a5 iX Write å ˆ e2ðia13 . if 1 < x < p À 1X 25. with values given by ö(ax by ) ˆ 0 ö(a ) ˆ À1 25. bl.12 Example Let a. and let á ˆ å ‡ å 5 ‡ å 8 ‡ å 12 . b P S5 be the permutations a ˆ (1 2 3 4 5).294 Representations and characters of groups where u is a primitive root modulo p. Then G has p À 1 linear characters.4 gi |CG ( gi )| ÷0 ÷1 ÷2 ÷3 ö 1 20 1 1 1 1 4 a 5 1 1 1 1 À1 b 4 1 i À1 Ài 0 b2 4 1 À1 1 À1 0 b3 4 1 Ài À1 i 0 x if 1 < y < p À 2. b: a13 ˆ b4 ˆ 1. Character table of F5. and one irreducible character ö of degree p À 1. q ˆ 4.25 we found the character table of F7.13 Example We consider the case p ˆ 13. and so by the previous example the character table of G is as shown.

2. 2. Let p and q be prime numbers with p . bÀ1 ab ˆ au iX The irreducible characters of Fp.5 of order 55. q. Prove that  & ' 1 y : x P Zà . Character table of F13. If G has order pq. Let u and v be two integers which are of order q modulo p.Characters of groups of order pq 295 to check that this agrees with the description of the character table provided by Theorem 25. b: ap ˆ bq ˆ 1.4 gi |CG ( gi )| ÷0 ÷1 ÷2 ÷3 ö1 ö2 ö3 1 52 1 1 1 1 4 4 4 a 13 1 1 1 1 á â ã a2 13 1 1 1 1 â ã á a4 13 1 1 1 1 ã á â b 4 1 i À1 Ài 0 0 0 b2 4 1 À1 1 À1 0 0 0 b3 4 1 Ài À1 i 0 0 0 Summary of Chapter 25 1. and de®ne .q.q are described in Theorem 25. with p prime and q| p À 1. p 0 x under matrix multiplication. Let u be an element of order q in Zà . Write down the character table of the non-abelian group F11.q ˆ ha.10. Let p be a prime number. Let p and q be positive integers. is a group of order p( p À 1). then either G is abelian or G  F p. 3. Suppose that p is prime and q divides p À 1. Exercises for Chapter 25 1. Then p Fp. y P Z p .10.

show that the two irreducible characters ö1 . Note that ka. cÀ1 ac ˆ aÀ1 . Suppose that p is a prime number. ö2 of G of degree q have values p 1 (ä p)) 2(À1 Æ on the element a. where ä ˆ 1 if p  1 mod 4. Let q ˆ ( p À 1)a2 and let G ˆ Fp.q ˆ ha. 22 . as in Exercise 5.q in 25. (( p À 1)a2)2 g). (d) Deduce that if å ˆ e2ðia p then ˆ p å s ˆ (À1 Æ (ä p)). bÀ1 ab ˆ au i. b: ap ˆ bq ˆ 1. bl is a normal subgroup of E which is isomorphic to C3 3 C3 . (c) Using the orthogonality relations.) 7. G2 ˆ ha. b: ap ˆ bq ˆ 1. obtain the character table of E.16. cÀ1 bc ˆ bÀ1 i. (b) Deduce that a is conjugate to aÀ1 if and only if p  1 mod 4. Q ˆ f12 . E provides a counterexample to the converse of Proposition 9. Show that the group E of Exercise 5 has the properties that Z(E) is cyclic. b.4. bÀ1 ab ˆ av iX Prove that G1  G2 . c: a3 ˆ b3 ˆ c2 ˆ 1.) 4. 6.296 Representations and characters of groups G1 ˆ ha. ab ˆ ba.6. bÀ1 ab ˆ au i. b: ap ˆ bq ˆ 1. (a) Show that there exists an integer m such that u m  À1 mod p if and only if p  1 mod 4. (Thus. and ä ˆ À1 if p  À1 mod 4. By inducing linear characters of this subgroup. 5. sPQ where Q is the set of quadratic residues modulo p (that is. but E has no faithful irreducible representation. where u is an element of order q modulo p. Let E be the group of order 18 which is given by E ˆ ha. X X X . (This justi®es the comment which follows the de®nition of Fp. with p Tˆ 2. (a) Find a group whose irreducible character degrees are .

1. 1. bÀ1 ab ˆ a2 iX Find the character table of G. 3. 3. 3. 6X 8. 3. 3. 2. 2. 1. 3. 3. 1. 3. 1. 3. 6. 1. 1.Characters of groups of order pq 1. Let G be the group of order 54 which is given by G ˆ ha. 3. 3. 297 . 3. 1. 1. 3. 3X (b) Find a group whose irreducible character degrees are 1. b: a9 ˆ b6 ˆ 1. 6. 1. 3X (c) Find a group whose irreducible character degrees are 1. 3. 2. 3. 3. 3. 1. 6. 1. 3.

In particular. Proof (1) Since H v G. Recall that Z(G) denotes the centre of G (see De®nition 9. Z(G) Tˆ {1}. and H ’ Z(G) consists of those conjugacy classes in H which have size 1. then G is abelian. that we have found the character tables of all groups of order less than 32. We later give explicitly the irreducible characters of all groups of order p3 and of all groups of order 16. (2) If K < Z(G) and GaK is cyclic. have an abelian subgroup of index p. H is a union of conjugacy classes of G. In the ®rst lemma we collect several well known properties of pgroups. indeed. (1) If {1} Tˆ H v G then H ’ Z(G) Tˆ {1}. p will be a prime number. 26.1 Lemma Let G be a group of order pn with n > 1. all of which have size a power of p. with references. Elementary properties of p-groups A p-group is a group whose order is a power of the prime number p. we show that all groups of order pn with 1 < n < 4 do. and before explaining the method. At the end of the chapter we point out.15). We shall show how to obtain the character tables of all groups of order pn for n < 4. Therefore 298 . The method consists of examining the characters of those p-groups which contain an abelian subgroup of index p. (3) If n < 2 then G is abelian.26 Characters of some p-groups Throughout this chapter.

Then G contains an abelian subgroup of index p. so suppose that 2 < n < 4.1(2). As K < Z( H) and. Then there exists a normal subgroup K of G such that K < H ’ G9 ’ Z(G) and jKj ˆ pX .3 Lemma Let G be a non-abelian p-group which contains an abelian subgroup H of index p. we deduce that Z( H) ˆ H. Z(G) and kxl are distinct nonidentity subgroups of Z( H). Therefore G is abelian.7). Proof The result is immediate if n ˆ 1. Let x1 . jGa Z(G)j < p nÀ1 . x2 P G. x2 ˆ g j k 2 for some integers i. and H is an abelian subgroup of index p. Ha Z( H) is not of order p. (3) By (1). k2 P K. and so Z( H) > p2 . Then x1 ˆ g i k 1 . j Hj ˆ jGjajx G j ˆ p3 .Characters of some p-groups j Hj ˆ j H ’ Z(G)j ‡ (a multiple of p)X 299 As | H| is a multiple of p and | H ’ Z(G)| Tˆ 0. k2 P Z(G). Then by Theorem 12. Then we can ®nd a subgroup H of G such that K < H and j Hj ˆ p nÀ1 . Now assume that Z(G) has no subgroup of order p nÀ2 . G has an element x whose conjugacy class x G is of size p. Let H ˆ CG (x). Therefore H is an abelian subgroup of index p in G.1(1).1(2).2 Lemma Let G be a group of order pn with 1 < n < 4.7. the only possibility is that |G| ˆ p4 and | Z(G)| ˆ p. Hence again Z( H) ˆ H by Lemma 26. recall that the derived subgroup of G is denoted by G9 (see De®nition 17. j and some k1 . Moreover. generated by gK. by Lemma 26. j For our ®nal result on the structure of p-groups.8. 26. Then by Exercise 12. Assume that Z(G) contains a subgroup K of order p nÀ2 . Since Z(G) Tˆ f1g by Lemma 26. Hence if n < 2 then Ga Z(G) is j cyclic and so G is abelian by (2). (2) Suppose that GaK is cyclic. we deduce that H ’ Z(G) Tˆ {1}. Since k1 . it follows that x1 x2 ˆ x2 x1 . 26.

by Theorem 11. This establishes (26X5) if ÷(1) ˆ p and K T< Ker ÷. again by Theorem 11.1(1). First note that if ÷ is a character of G of degree p. we have KH ˆ H. as G9 is in the kernel of every linear character. Since G is non-abelian and |G: H| ˆ p.8 that all the p nÀ1 irreducible characters of the abelian group H are linear.3.3. and therefore K < H. and hence K < G9 < Ker ÷. for some linear character ø of H which satis®es K T< Ker ø.300 Representations and characters of groups Proof Since G is non-abelian.4 Theorem Assume that G is a non-abelian p-group which contains an abelian subgroup H of index p. Since p nÀ1 ‡ ( p nÀ2 À p nÀ3 ) p2 ˆ pn ˆ jGj. Proof Let |G| ˆ pn . we have G9 < Ker ÷.11. j Characters of p-groups with an abelian subgroup of index p In view of Lemma 26. the irreducible characters of GaK lift to give precisely those irreducible characters of G which have K in their kernel. the degree of every irreducible character of G is a power of p).12.12. we have {1} Tˆ G9 v G. then ÷ is irreducibleX We know by Theorem 9. We shall construct p nÀ2 À p nÀ3 further irreducible characters of G. 26.2. h P H}). In the latter case. The sum of the squares of the degrees of the irreducible characters obtained in this way is jGaKj ˆ p nÀ1 . Then every irreducible character of G is given by either (1) the lift of an irreducible character of GaK. Let K be a subgroup of order p in G9 ’ Z(G). By Theorem 17. and hence G9 ’ Z(G) Tˆ {1} by Lemma 26. Now K < Z(G) implies that K v G and that KH is an abelian subgroup of G (where KH ˆ {kh: k P K. Let K be a normal subgroup of G as in Lemma 26. the next theorem provides us with all the irreducible characters of non-abelian groups of order p3 or p4 . then either ÷ is irreducible or ÷ is a sum of linear characters (since by Theorem 22. or (2) ø 4 G. each of degree p. (à ) we shall then have obtained all the irreducible characters of G. Let Ö denote the set of linear .

4 to give an explicit construction of the irreducible characters of the non-abelian groups of order p3 . Therefore by (26. Now let G be a non-abelian group of order p3 . Choose aZ. Write Z ˆ Z(G). Suppose now that ø1 is a linear character of H such that ø 4 G ˆ ø1 4 G. G has at most p nÀ2 À p nÀ3 such characters. ø1 i H X Since (ø 4 G) 5 H has degree p. Then .5). By Lemma 26. Therefore fø 4 G: ø P Ög consists precisely of the p nÀ2 À p nÀ3 irreducible characters we seek. Hence Ga Z  Cp 3 Cp and Z ˆ kzl  Cp .6. Z Tˆ {1} and Ga Z is non-cyclic.1. ø 4 G is an irreducible character of G. Groups of order p3 By Theorem 9. j We now use Theorem 26.23. It follows that fø 4 G: ø P Ög gives at least jÖja p ˆ ( p nÀ1 À p nÀ2 )a p distinct irreducible characters of G of degree p which do not have K in their kernel. By Proposition 21. (ø 4 G)(k) ˆ pø(k) for all k P KX Thus ø 4 G has degree p and does not have K in its kernel. this implies that there are at most p elements ø1 of Ö such that ø1 4 G ˆ ø 4 G. As we saw in (à ). 1 ˆ hø 4 G. we have jÖj ˆ p nÀ1 À p nÀ2 X Let ø P Ö.8. C p2 3 Cp and Cp 3 Cp 3 Cp X The character tables of these groups are given by Theorem 9. ø1 4 Gi G ˆ h(ø 4 G) 5 H. the abelian groups of order p3 are C p3 . Since the linear characters of H which do have K in their kernel are precisely the lifts of linear characters of HaK. bZi.Characters of some p-groups 301 characters of H which do not have K in their kernel. Then by the Frobenius Reciprocity Theorem 21.16. We shall then illustrate Theorem 26. and the proof is complete. bZ such that Ga Z ˆ haZ.4 further by constructing the character tables of all the non-abelian groups of order 16. since K < Z(G).

v is the linear character ÷ u. Let H ˆ ka.6 Theorem Let G ˆ {ar bs z t : 0 < r.v (ar bs Z) ˆ å ru‡sv X The lift to G of ø u. s. and therefore g ˆ ar z t for some t.v (ar bs z t ) ˆ å ru‡sv . Let r be an integer with 1 < r < p À 1. where ø u. ö u (ar bs z t ) ˆ 0.8. t < p À 1} be a non-abelian group of order p3 . zl. as above. and (1 < u < p À 1). so ar Z ˆ gZ.v (0 < u.v öu where for all r. 0 < s < p À 1g and in particular. Since ar P Z. s.v which appears in the statement of the theorem. If ar is conjugate to an element g of G. t < p À 1. 0 < v < p À 1). every element of G is of the form ar bs z t for some r. på . choose a character ø u of H which satis®es ø u (z t ) ˆ å ut (0 < t < p À 1)X We shall calculate ø u 4 G. v < p À 1).302 Representations and characters of groups Ga Z ˆ far bs Z: 0 < r < p À 1. a the conjugacy class (ar ) G does not have size 1. the irreducible characters of Ga Z are ø u. Then the irreducible characters of G are ÷ u. 26. Write å ˆ e2ðia p . Proof By Theorem 9. s. s. @ ut if r ˆ s ˆ 0. and hence (ar ) G ˆ far z t : 0 < t < p À 1gX . t with 0 < r. then ar Z is conjugate to gZ in the abelian group Ga Z. t. otherwiseX (0 < u < p À 1. so that H is an abelian subgroup of order p2 . For 1 < u < p À 1. ÷ u.

v (0 < u. they are .Characters of some p-groups Then by Proposition 21. If p ˆ 2. 12 ‡ ( p À 1) . v < p À 1) and ö u (1 < u < p À 1) are all distinct. We ®nd that 1 ˆ hö u . and (ø u 4 G)( g) ˆ 0 if g P HX a pÀ1 ˆ sˆ0 pÀ1 ˆ sˆ0 303 ø u (z s ) å us We have now established that if ö u ˆ ø u 4 G. and the sum of the squares of their degrees is p2 . ö u i G ˆ 3 ö u ( g)ö u ( g) p gPG ˆ ˆ 1 ˆ ö u ( g)ö u ( g) p3 gP Z 1 ˆ 2 p p3 gP Z ˆ 1X Therefore ö u is irreducible. j Notice that the calculation in the proof of Theorem 26. (ø u 4 G)(z t ) ˆ pø u (z t ) ˆ på ut .6 is a special case of the proof of Theorem 26. In fact.23. there are precisely two non-abelian groups of order p3 . (ø u 4 G)(ar z t ) ˆ ø u (ar ) ‡ ø u (ar z) ‡ X X X ‡ ø u (ar z pÀ1 ) ˆ ø u (ar ) ˆ ø u (ar ) ˆ 0X Also. And if p is odd. p2 ˆ jGjX Hence we have found all the irreducible characters of G.4 (with K ˆ Z(G)). up to isomorphism. Clearly the irreducible characters ÷ u. they are D8 and Q8. then ö u takes the values stated in the theorem.

304 (26X7) Representations and characters of groups H 1 ˆ ha. there are precisely fourteen groups of order 16 (see p. b chosen in the statement of Theorem 26. C4 3 C2 3 C2 and C2 3 C2 3 C2 3 C2 . and it is not Q8 by Exercise 26. b: a8 ˆ b2 ˆ 1. bÀ1 ab ˆ aÀ1 i. It is not C8 by Lemma 26.6.8.8 The groups of order 16 It is known that.7). and their character tables are given by Theorem 9. . These are G of order 16 with G1 ˆ ha.8. For each of the nine non-abelian groups G of order 16 it is the case that |G9 ’ Z(G)| ˆ 2 (see Exercise 26. 134 of the book by Coxeter and Moser listed in the Bibliography). By Theorem 9.3 is given by K ˆ G9 ’ Z(G)X Now GaK is a group of order 8. b: a8 ˆ 1. Hence GaK  D8 . Our descriptions will be in terms of presentations. using Exercise 26. C4 3 C4 . . bÀ1 ab ˆ aziX 2 We have Z( H1 ) ˆ ka p l. We shall describe all these groups and their character tables. it is possible to see.6. az ˆ za. up to isomorphism. so the subgroup K described in Lemma 26. The elements a. according to these three possibilities for GaK. b2 ˆ a4 . and H 2 ˆ ha. b in H1 and H2 will serve for the elements a. G2 ˆ ha. b: a p ˆ b p ˆ 1. b: a8 ˆ b2 ˆ 1. G3 ˆ ha. (A) There are three non-abelian groups GaK  D8 . . Z( H2 ) ˆ kzl.1(2). bz ˆ zb. G9 given below do indeed have order 16. .5. the abelian groups of order 16 are C16 . bÀ1 ab ˆ a3 iX . bÀ1 ab ˆ a p‡1 i. 26. z: a p ˆ b p ˆ z p ˆ 1. that all the nine groups G1 . C8 3 C2 . bÀ1 ab ˆ aÀ1 i ˆ D16 . C4 3 C2 or C2 3 C2 3 C2 X We shall divide our descriptions into three parts. b.

4. as before. of order 2).6). b2 ˆ z 2 ˆ 1. G6 ˆ ha. a7 a. Note that a is conjugate to aÀ1 in G1 and G2 .4(2) as induced characters ø 4 G. z 2 ˆ 1. C1 G1 . hence the values in the columns C4 and C5 are all real for G1 and G2 . a6 a2 . These are G4 ˆ ha. bÀ1 ab ˆ az. . b2 ˆ z 2 ˆ 1. a6 C4 a. a3 C5 a3 . b. b. . K ˆ G9 ’ Z(G). they can be found by using the column orthogonality relations.Characters of some p-groups 305 In each case K ˆ ka4 l. and these are given in the following table. G2 G3 1 1 C2 a4 a4 C3 a2 . b. . bÀ1 ab ˆ azi. where ø is a linear character of the abelian subgroup kal of index 2. . a7 C6 ai b (i even) ai b (i even) C7 ai b (i odd) ai b (i odd) Using Theorem 26. a5 a5 . C7 . z: a4 ˆ 1. az ˆ za. bÀ1 ab ˆ azi. The last two characters can be obtained as in Theorem 26. p á ˆ i 2 ˆ Àâ for G3 X p The ®rst ®ve characters are the lifts of the irreducible characters of GaK  D8 . b2 ˆ z. Each of the groups has seven conjugacy classes C1 . G2 and G3 : Class Centralizer order C1 16 1 1 1 1 2 2 2 C2 16 1 1 1 1 2 À2 À2 C3 8 1 1 1 1 À2 0 0 C4 8 1 1 À1 À1 0 á â C5 8 1 1 À1 À1 0 â á C6 4 1 À1 1 À1 0 0 0 C7 4 1 À1 À1 1 0 0 0 where ሠ2 ˆ Àâ for G1 . but not for G3 (see Corollary 15. G2 . but not in G3 . we obtain the character tables of G1 . z: a4 ˆ 1. alternatively. (B) There are three non-abelian groups G of order 16 with GaK  C4 3 C2 (where. G5 ˆ ha. z: a4 ˆ z. bz ˆ zbiX .

a2 bz C9 ab. bÀ1 ab ˆ az 2 . az a3 . bz ˆ zbiX . az ˆ za. G5 . a3 z b. G8 ˆ ha. . b. b. G9 ˆ ha. abz C10 a3 b. . . for G5 . G6 X (C) Finally. but they are in a form which allows us to describe the conjugacy classes C1 . bz ˆ zbi  Q8 3 C2 . az ˆ za. C10 of all three groups G4 . These are G7 ˆ ha. G6 simultaneously: C1 1 C2 z C3 a2 C4 a2 z C5 C6 C7 C8 a2 b. there are three non-abelian groups G of order 16 with GaK  C2 3 C2 3 C2 (where K ˆ G9 ’ Z(G). az ˆ za. bÀ1 ab ˆ aÀ1 . bÀ1 ab ˆ aÀ1 . b. since a4 ˆ z in G4 . a2 ˆ b2 . of order 2). bz ˆ zbi  D8 3 C2 .4: Class Centralizer order C1 16 1 1 1 1 1 1 1 1 2 2 C2 16 1 1 1 1 1 1 1 1 À2 À2 C3 16 1 À1 1 À1 1 À1 1 À1 á â C4 16 1 À1 1 À1 1 À1 1 À1 â á C5 8 1 i À1 Ài 1 i À1 Ài 0 0 C6 8 1 Ài À1 i 1 Ài À1 i 0 0 C7 8 1 1 1 1 À1 À1 À1 À1 0 0 C8 8 1 À1 1 À1 À1 1 À1 1 0 0 C9 8 1 i À1 Ài À1 Ài 1 i 0 0 C10 8 1 Ài À1 i À1 i 1 Ài 0 0 where á ˆ 2i ˆ Àâ á ˆ 2 ˆ Àâ for G4 . G5 and G6 can again be found using Theorem 26. .306 Representations and characters of groups The given presentations are somewhat cumbersome (for example. a3 bz a. z: a2 ˆ b2 ˆ z 4 ˆ 1. The character tables of G4 . z: a4 ˆ z 2 ˆ 1. bz In each case. z is redundant). z: a4 ˆ b2 ˆ z 2 ˆ 1. K ˆ kzl.

are as follows: . a3 az. a2 bz ab. az 2 az.3. G8 and G9 . a3 b abz. the groups.4. G8 . az 3 b. G8 G9 1 a2 1 z2 C5 C6 C7 C8 C9 C10 307 z a2 z a. are as follows. abz 2 abz. abz3 We have Kˆ @ ha2 i hz 2 i for G7 . which are given by C1 C2 C3 C4 G7 .8 and Section 18. Class Centralizer order C1 16 1 1 1 1 1 1 1 1 2 2 C2 16 1 1 1 1 1 1 1 1 À2 À2 C3 16 1 1 1 À1 1 À1 À1 À1 á â C4 16 1 1 1 À1 1 À1 À1 À1 â á C5 8 1 1 À1 1 À1 1 À1 À1 0 0 C6 8 1 1 À1 À1 À1 À1 1 1 0 0 C7 8 1 À1 1 1 À1 À1 1 À1 0 0 C8 8 1 À1 1 À1 À1 1 À1 1 0 0 C9 8 1 À1 À1 1 1 À1 À1 1 0 0 C10 8 1 À1 À1 À1 1 1 1 À1 0 0 where á ˆ 2 ˆ Àâ á ˆ 2i ˆ Àâ for G7 . given by Theorem 26. a3 z z z 3 a.Characters of some p-groups Each of these groups has ten conjugacy classes. Apart from abelian groups and dihedral groups. we have in fact found the character tables of all groups of order 31 or less. with references for their character tables. a2 b bz. and the character tables of G7 . for G9 . bz 3 ab. for G9 X 26. a3 bz b.9 The groups of order less than 32 At this point. G8 . bz 2 bz. whose character tables are given by Theorem 9.

A4 3 C2 .8 Theorem 19. 2.308 |G| 8 12 16 18 20 21 24 Representations and characters of groups G Q8 A4 T12 G1 .8: groups of order 16. D6 3 C4 S4 SL (2.1 Section 18. Show that for some integer m > 2.18 Exercise 25.5 Theorem 26. 3) T24 U24 V24 H1 . G has pm linear characters and p nÀ2 À p mÀ2 irreducible characters of degree p.4 F7. H2 T28 D6 3 C5 .6: groups of order p3 .18 Theorem 19. G9 D6 3 C3 E T20 F5.5 Exercise 18. Theorem 26. Q8 3 C3 .4: p-groups which contain an abelian subgroup of index p. n > 2).10 Theorem 19. 3. .3 Theorem 25. . T12 3 C2 D8 3 C3 . Section 26.10 Theorem 25.6 Exercise 18.18 Section 18. 1.18 27 28 30 Summary of Chapter 26 In this chapter. . as follows. D10 3 C3 Reference for character table Exercise 17. we gave the irreducible characters of various nonabelian p-groups G.3 Exercise 18. .2 Exercise 18. . Suppose that G is a group of order pn ( p prime. with an abelian subgroup H of index p.3 Theorem 19.3 D12 3 C2 .4 Exercise 18.2 Exercise 18. Theorem 26.1 Exercise 27. Exercises for Chapter 26 1.3 Section 26.

Let H be the group of order 27 which is given by H ˆ ha. 3. az ˆ za. and deduce that G9 ˆ h Zi. bz ˆ zb. ®nd the character table of G. Dl. b. Write Z ˆ ÀI. (a) Prove that all pairs of generators commute modulo h Zi. .4. G9 be the non-abelian groups of order 16 with presentations as given in the text. b2 ˆ a8 . D H À1 f f 0 f Aˆf f 0 d 0 f fi f Cˆf f0 d 0 H i 0 0 0 be the following 4 3 4 matrices: I H 0 0 0 0 0 i g f f0 1 0 0g 0 0 g f Bˆf g. Let G be the group of order 32 which is given by G ˆ ha. bÀ1 ab ˆ azi 309 (see (26. G4 and G9 . and deduce that G is a 2group of order at most 32. (b) Show that for all g in G. . Find the conjugacy classes of H. 5. (Hint: use Corollary 9. G7 and G8 have no faithful irreducible representations? . or otherwise. (c) Prove that the given representation of G of degree 4 is irreducible. z: a3 ˆ b3 ˆ z 3 ˆ 1.3. (a) Find faithful irreducible representations of degree 2 for G1 . B. Let A. and use Theorem 26. . b: a16 ˆ 1.7)). G2 . C. g 2 P h Zi. 4. Let G1 .Characters of some p-groups 2.6 to write down the character table of H. (b) Why do the remaining groups G5 . G3 . bÀ1 ab ˆ aÀ1 iX Using Theorem 26. 0g e 0 I g 0g g g. and ®nd all the irreducible representations of G. B.) (d) Show that |G| ˆ 32. 0g e 0 and let G ˆ kA. fi 0 1 0g 0 0 e d 0 0 0 0 Ài À1 0 I 0 f f0 f Dˆf f0 d 1 H 0 Ài 0 0 0 1 1 0 0 0 0 1 0 I 0 0 g Ài g g g. G6 . Ài g e 0 g 0g g g. . C.

b 3 d 0 Ài 0 e.) (b) Deduce from the result of Exercise 7 that if G is a group of order 16. 0 0 i 0 0 1 H I À1 0 0 z 3 d 0 À1 0 eX 0 0 1 (d) Find faithful representations of degree 3 for G7  D8 3 C2 and G8  Q8 3 C2 . . Let G be a non-abelian group of order p4 . (Note: This exercise can be used to con®rm that the presentations of G1 . and hence that a2 P Z. . Prove that no two of the groups G1 . G9 given in the text do indeed give groups of order 16. then Ga Z(G) T Q8 . a2  b2 mod Z. . then Ga(G9 ’ Z(G)) T Q8 . (a) Prove that | Z(G)| ˆ p or p2 . . bÀ1 ab  aÀ1 mod Zi.310 (c) Representations and characters of groups Check that the following give faithful representations of G5 and G6 : H I H I 0 1 0 i 0 0 G5 : a 3 d 1 0 0 e. bZ: a4 P Z. Prove that a2 commutes with b. (c) Deduce that |G9 ’ Z(G)| ˆ p. . . and that if | Z(G)| ˆ p2 then G has p3 ‡ p2 À p conjugacy classes. . (Hint: assume that Ga Z ˆ haZ. 7. 0 0 1 H I H I i 0 0 0 1 0 G6 : a 3 d 0 Ài 0 e. G9 are isomorphic. 0 0 i 0 0 1 H I À1 0 0 z 3 d 0 À1 0 e. b 3 d 1 0 0 e. 8. . (a) Prove that if G is any group. (b) Prove that |G9| ˆ p or p2 . and that if |G9| ˆ p2 then G has 2 p2 À 1 conjugacy classes. .) 6.

p) the set of all 2 3 2 matrices M with entries in Z p such that det M ˆ 1. To calculate the order of the group SL (2. and recall that Z p is the ®eld which consists of the numbers 0. The group belongs to a whole family of simple groups. p). Examples of simple groups which we have met so far are cyclic groups of prime order. . . except that a Tˆ 0. d P Z p . Denote by SL (2. is the smallest non-abelian simple group. Special linear groups Let p be a prime number. d which make ad À bc ˆ 1 (since a. c. The next smallest is a certain group of order 168. then there are p( p À 1) choices for a. In fact the group A5 . . of order 60. A5 and A6 . c. with addition and multiplication modulo p. and is called the 2-dimensional special linear group over Z p.27 Character table of the simple group of order 168 Recall that a simple group is a non-trivial group G such that the only normal subgroups of G are f1g and G itself. ad À bc ˆ 1)X c d If c ˆ 0. b. b. and d is determined by a). and we begin with a description of this family. p À 1. b. and in this chapter we shall describe this group and ®nd its character table. Then SL (2. we count the matrices   a b (a. d 311 . . And there are p2 ( p À 1) choices for a. p) is a group under matrix multiplication. b are arbitrary. We discussed brie¯y in Chapter 1 the signi®cance of simple groups in the theory of ®nite groups.

p) is Z ˆ fI. p)| ˆ p( p2 À 1). In the exercises. p)a Z is called the 2-dimensional projective special linear group. and we shall construct the character table of this group. p). The following table records representatives gi (1 < i < 6) for the conjugacy classes. and the size of the conjugacy class containing gi . we indicate other ways of obtaining characters of G. J. Thus PSL (2.1. Rotman listed in the Bibliography). so assume that p is an odd prime. p) has order 6. d may be chosen arbitrarily. The conjugacy classes of PSL (2. p) is simple (see Theorem 8. After ®nding the conjugacy classes of G. 3)  A4 . notably the orthogonality relations (Theorem 16. and then b is determined).312 Representations and characters of groups with c Tˆ 0. and it is easy to see that this group is isomorphic to S3 . p)afÆIgX Since |SL (2. the order of CG ( gi ). we have jPSL(2. the group PSL (2. The power of these techniques is therefore well illustrated. The simple group G ˆ PSL (2.1 Lemma The group PSL (2. 5)  A5 . 7) 27. using information about subgroups.19 of the book by J. PSL (2. p) ˆ SL (2. The factor group SL (2. 7) has order 168. Therefore jSL (2. 7) has exactly six conjugacy classes.4) and congruence properties (Corollary 22. . ÀIg (where I is the 2 3 2 identity matrix). the centre of SL (2. c is any non-zero element of Z p . together with the order of gi . such that ad À bc ˆ 1 (since a.26). By Exercise 27. p)j ˆ p( p À 1) ‡ p2 ( p À 1) ˆ p( p2 À 1)X If p ˆ 2 then SL (2. and that for p > 5. p)j ˆ p( p2 À 1)a2X It is known that PSL (2. we shall ®nd the character table using only numerical calculations. and is written as PSL (2.

. Then     a b 2 0 2 ˆÆ c d 0 4 0 and hence b ˆ c ˆ 0. . . . for example. Among g1 . g6 . and then use direct calculation to ®nd all the elements of G which commute with g i . .  a c  b . we verify that gi has the stated order. Suppose that   a b Z c d commutes with g4 . g4 . d  ' Z X  4 0 0 2        3 À2 3 2 0 À1 1 0 . . Consequently &   1 0 2 Z. X 3 À2 4 À2 À2 2 2 2 Also. . CG ( gi ) ˆ k gi l for i ˆ 3.Character table of the simple group of order 168 Order of gi  g1 ˆ  g2 ˆ  g3 ˆ  g4 ˆ  g5 ˆ  g6 ˆ 1 0 0 1  Z  Z 2 4 3 7 7 8 4 3 7 7 21 42 56 24 24 1 168 |CG ( gi )| | gG | i 1 313 0 1 À1 0 2 2 2 0 1 0 1 0  À2 Z 2  0 Z 4  1 Z 1  À1 Z 1 Proof For each i. . Consider. CG ( g 4 ) ˆ 0 1 0 Similarly CG ( g2 ) ˆ &  0 4 0 4  Z. 5. 6. MZ: M ˆ À2 4 2 4 1 0 0 1        ' 2 3 2 4 2 2 2 À2 . the only elements with the same order are g5 . .

÷6 be the irreducible characters of G. then ÷( gi ) is an integer. (2) For some character ÷ of G. where ÷1 is the . d ˆ aÀ1 and     a a‡b a b À aÀ1 ˆ X 0 aÀ1 0 aÀ1 Therefore a2 ˆ À1. which is impossible for a P Z7 . it also has six irreducible characters.16. so no two of these six elements are conjugate. so g5 is not conjugate to its inverse. j Notice that using Lemma 27. except possibly g5 and g6 . g6 are conjugate. gi is conjugate to ( gi ) k whenever gi and ( gi ) k have the same order. Hence the conclusion follows from Theorem 22. Proof (1) By Lemma 27. these exhaust the conjugacy i classes of G. Let ÷1 . for 1 < i < 4. it is easy to check that G is indeed simple. . 27. 7) Since G has six conjugacy classes. ÷( g5 ) is non-real.1. Since the sum of the sizes of the six conjugacy classes g G (1 < i < 6) is 168. and so    a a‡b aÀc ˆÆ c c‡d c bÀd d  with ad À bc ˆ 1X It follows that c ˆ 0. and we have established that no two of the elements g1 . . j The character table of G PSL (2. Suppose that gÀ1 g6 g ˆ g5 with   a b gˆ Z P GX c d Then gg5 ˆ g6 g. a Tˆ 0.1. . . 5 Therefore (2) follows from Corollary 15.8). . since any normal subgroup is a union of conjugacy classes (see Proposition 12. . The size of the conjugacy class g G is obtained by dividing 168 by i |CG ( gi )| (Theorem 12.6. . Thus g5 is not conjugate to g6 . (2) Notice that g6 ˆ gÀ1 .19).314 Representations and characters of groups and g6 .2 Corollary (1) If 1 < i < 4 and ÷ is a character of G. .

and the congruence properties given by Corollaries 22. 0 in some order. we have by Corollary 22. Æ1. 0. with a suitable ordering of ÷2 .) Similarly the entries in the column of g3 are 1. and the sum of the squares of these integers is equal to |CG ( g4 )| ˆ 3. Theorem 16. ÷( g 2 )  ÷(1) mod 2. for which the character values are known to be integers. Now for all characters ÷ of G. g3 . by Corollary 27. Æ1. Recall that the character table is the 6 3 6 matrix with ij-entry ÷ i ( gj ). . Æ1.Character table of the simple group of order 168 315 trivial character (so that ÷1 ( g) ˆ 1 for all g P G). and so ÷( g 2 )  ÷( g 3 ) mod 2X Since we also know that 6 ˆ iˆ1 ÷ i ( g 3 )÷ i ( g 4 ) ˆ 0. so di is the entry on row i of column 1. . Æ1. Æ2. We shall repeatedly exploit the column orthogonality relations. Æ1. (We know that the entry in the ®rst row is ÷1 ( g4 ) ˆ 1. For the moment we concentrate on the entries in the ®rst column of the character table (i.e. Let di ˆ ÷ i (1). and the entries in column g2 are 1. we see that. .4(2). 0. The entries must therefore be 1. 0 in some order.2. Æ1. 0. and ÷( g 3 )  ÷(1) mod 2. Æ1.27 for the elements g2 . By . . part of the character table of G is as follows: Class representative Centralizer order ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g2 8 1 Æ1 0 Æ1 Æ1 Æ2 g3 4 1 Æ1 0 Æ1 Æ1 0 g4 3 1 Æ1 Æ1 0 0 0 We shall determine the signs later. The entries in the column of g4 are integers.26 and 22. the degrees ÷ i (1)). 0 in some order. Æ1. g4 . ÷6 .27.

we have d 4  0 mod 3. and hence d6 ˆ 6.Representations and characters of groups €6 2 Corollary 22.27. Next. d3 equal to 7. and have the following portion: Class representative Centralizer order ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g1 168 1 7 8 3 3 6 g2 8 1 Æ1 0 Æ1 Æ1 Æ2 g3 4 1 Æ1 0 Æ1 Æ1 0 g4 3 1 Æ1 Æ1 0 0 0 . d 4 divides jGj ˆ 168. 2 Now 1 ‡ d 2 ‡ d 2 ‡ 32 ‡ 32 ‡ 62 ˆ 168. d 2 < 168X 6 Therefore d6 is 6 or 12. so as d 2 < 168. d 6  0 mod 3. 8 in some order. Since d 2  1 mod 2. 2 3 so d 2 ‡ d 2 ˆ 113. d 6  0 mod 2. d 2 < 168X 4 The only positive integer d4 which satis®es these conditions is d4 ˆ 3. d 4  1 mod 2. But 0ˆ 6 ˆ iˆ1 316 and and ÷ i ( g 2 )d i ˆ 1 Æ d 2 Æ 3 Æ 3 Æ 2d 6 . we have d2 ˆ 7 and d3 ˆ 8. Theorem 22. d 6 divides 168. d3 2 3 positive integers have d2. In the same way.11 and the fact that iˆ1 d i ˆ 168. We have now found the ®rst column of the character table. d5 ˆ 3. The only solutions to this equation with d2. we have d6 Tˆ 12.

3. the equation 1 ˆ h÷2 .) Also. g4 . By Corollary 27. Hence ÷4 and ÷5 (being the only two irreducible characters with the same degree) must be complex conjugates of each other. Thus the column for g5 is . 6. For this character ÷.2. (Note that ÷2 ( g5 )  ÷2 (1) mod 7. for j ˆ 5. 0ˆ 6 ˆ iˆ1 ÷ i ( g 4 )÷ i ( g j ) ˆ 1 À ÷3 ( g j ) and so ÷3 ( g5 ) ˆ ÷3 ( g6 ) ˆ 1. and let ÷6 ( g5 ) ˆ t. We obtain Class representative Centralizer order ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g1 168 1 7 8 3 3 6 g2 8 1 À1 0 À1 À1 2 g3 4 1 À1 0 1 1 0 g4 3 1 1 À1 0 0 0 Next. g3 . there is an irreducible character ÷ of G such that ÷( g5 ) is non-real. but we could not use this fact as we were not sure that ÷2 ( g5 ) was an integer. ÷2 i ˆ ˆ 6 ˆ ÷2 ( g i )÷2 ( g i ) iˆ1 jCG ( g i )j 7X7 1 1 1 ÷2 ( g 5 )÷2 ( g 5 ) ÷2 ( g 6 )÷2 ( g 6 ) ‡ ‡ ‡ ‡ ‡ 168 8 4 3 7 7 gives ÷2 ( g5 ) ˆ ÷2 ( g6 ) ˆ 0. Let ÷4 ( g5 ) ˆ ÷5 ( g 5 ) ˆ z. 4 now enable us to determine the signs in the columns for g2 .Character table of the simple group of order 168 The equations 6 ˆ iˆ1 317 ÷ i ( g 1 )÷ i ( g j ) ˆ 0 for j ˆ 2. the complex conjugate ÷ will be a different character of the same degree.

7). we obtain p t ˆ À1. ÷ i ( g 3 )÷ i ( g 5 ) ˆ 1 ‡ z ‡ z. we have ÷( g6 ) ˆ ÷( g 5 ) for all characters ÷ of G. Character table of PSL (2. as shown. 7) Class representative Centralizer order ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 p where á ˆ (À1 ‡ i 7)a2.318 Representations and characters of groups Class representative Centralizer order ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g5 7 1 0 1 z z t Now 0ˆ 0ˆ 7ˆ 6 ˆ iˆ1 6 ˆ iˆ1 6 ˆ iˆ1 ÷ i ( g 2 )÷ i ( g 5 ) ˆ 1 À z À z ‡ 2t. . ÷ i ( g 5 )÷ i ( g 5 ) ˆ 2 ‡ 2zz ‡ ttX Solving these equations. g1 168 1 7 8 3 3 6 g2 8 1 À1 0 À1 À1 2 g3 4 1 À1 0 1 1 0 g4 3 1 1 À1 0 0 0 g5 7 1 0 1 á á À1 g6 7 1 0 1 á á À1 It is known that there are precisely ®ve non-abelian simple groups of order less than 1000. 5 We have now completely determined the character table of G ˆ PSL (2. z ˆ (À1 Æ i 7)a2X Since g6 ˆ gÀ1 . We give you the character tables of all of these.

Character table of the simple group of order 168 as follows: G A5 PSL (2. b. 7) that this group is simple. b P Z7 7 0 aÀ1 (where Z ˆ {ÆI}). given the conjugacy classes of G. 3). 7). 3. (a) De®ne the subgroup T of G. 8) PSL (2. and show that (1 T ) 4 G ˆ 1 G ‡ ÷. p)) ˆ fÆIg. as follows: &  ' a b Tˆ Z : a P Zà . Exercises for Chapter 27 1. 7) A6 PSL (2.6 & 1. PSL (2. 3. SL (2. Deduce directly from the character table of PSL (2.2 Exercise 28. Find the character table of SL (2. 11) Order of G 60 168 360 504 660 319 Reference for character table Example 20. Prove that Z(SL (2. c. p)j ˆ p( p2 À 1).13 This chapter Exercise 20. 7). as in Lemma 27. p)afÆIg. 2.1. p) ˆ a c Summary of Chapter 27  ' b : a.3 Exercise 27. . the simple group of order 168. where ÷ is an irreducible character of G. In this exercise we present an alternative construction of the character table of G ˆ PSL (2. p) ˆ SL (2. d P Z p . of order 21. d jSL (2. p)j ˆ p( p2 À 1)a2 ( p odd). Calculate the values of the induced character (1 T ) 4 G. 2. jPSL (2. 4. We constructed the character table of PSL (2. ad À bc ˆ 1 .

14). Use orthogonality relations to complete the character table of G. The character table of SL (2. 7). (c) By considering ÷ S (see Proposition 19. Calculate the values of ë 4 G and prove that this is an irreducible character of G. (c). (d) From (a). Let G ˆ SL (2. (b). 5. the group of all 2 3 2 matrices of determinant 1.320 Representations and characters of groups (b) Let ë be a non-trivial linear character of T. 7. (a) Show that G has 11 conjugacy classes with representatives gi as follows: gi  g1 ˆ  g2 ˆ 1 0 0 1  1 2 4 8 8 3 6 7 14 7 14 336 336 8 8 8 6 6 14 14 14 14 1 1 42 42 42 56 56 24 24 24 24 Order of gi |CG ( gi )| | gG | i  À1 0 0 À1   0 1 g3 ˆ À1 0   2 À2 g4 ˆ 2 2   À2 2 g5 ˆ À2 À2   2 0 g6 ˆ 0 4   À2 0 g7 ˆ 0 À4   1 1 g8 ˆ 0 1   À1 À1 g9 ˆ 0 À1   1 À1 g 10 ˆ 0 1   À1 1 g 11 ˆ 0 À1 . we now have irreducible characters of G of degrees 1. 7). with entries in the ®eld Z7 . 8 and 6. obtain an irreducible character of G of degree 6.

÷9 . (e) By considering the column of g6 in the character table. prove that ø A is equal to the irreducible character of G of degree 6 whose kernel contains Z. we have ÷ j ( g) ˆ À÷ j (À g). The character table of PSL (2. g8 having orders and centralizer orders as follows: gi Order of gi |CG ( gi )| g1 1 660 g2 2 12 g3 3 6 g4 6 6 g5 5 5 g6 5 5 g7 11 11 g8 11 11 Also. . gÀ1 . 11). . 6. Show that for any j with 7 < j < 11 and any g P G. . g7 . This group has eight conjugacy classes with representatives g1 . 6. 11). Find the character table of G. ÷8 . g2 . respec5 6 7 8 tively. Let G ˆ PSL (2. ÷10 . gÀ1 are conjugate to g5 . g8 .Character table of the simple group of order 168 321 (b) Use the character table of PSL (2. and ®nd ÷ j ( g6 ) for 7 < j < 11. g3 and g6 (see Proposition 19. By considering the values of ø A on g1 . . Deduce the values of the irreducible characters of degree 4 on all gi . ÷11 are 4. and deduce that ÷ j (1) is even. . 4. 8.14). show that the degrees of ÷7 . . (f) Let ø be one of the irreducible characters of degree 4. . (d) Prove that ÷ j ( g3 ) ˆ 0 for 7 < j < 11. gÀ1 . (c) Let ÷7 . 6. gÀ1 .7) to write down the six irreducible characters of G with kernel containing Z ˆ {ÆI}. . and congruences modulo 3. (g) Complete the character table of G. . ÷11 be the remaining irreducible characters of G. g6 .

Fraleigh listed in the Bibliography. Recall that a ®eld (F. At ®rst sight. R. We shall determine the character tables of some matrix groups with entries from an arbitrary ®nite ®eld. (F. 3) is a set F with two binary operations ‡ and 3 such that the following properties hold. However. Finally. B. b. ‡. this is a daunting task. The basic properties of ®nite ®elds which we will use without proof are these: 322 . then (F à . Consequently. In the last chapter and its exercises. since the number of irreducible characters increases with the size of the ®eld. we found the character tables of certain groups of 2 3 2 matrices with entries in Z7 and Z11 . c P F. the distributive law holds. that is (a ‡ b)c ˆ ac ‡ bc for all a. we can display the character values in a 4 3 4 matrix. if we write F à ˆ Fnf0g. with identity element 1. For example. First. ‡) is an abelian group. Secondly. The ®elds F q and F q 2 We consider ®nite ®elds. and one of the exercises will show you how to use the results of this chapter to determine the character tables of in®nitely many simple groups. with the usual de®nitions of ‡ and 3. as do the irreducible characters. and we shall tell you which properties of these ®elds we will use. with identity element 0. if you are unfamiliar with ®nite ®elds then you might like to consult the book by J. we shall see that the conjugacy classes of our groups fall into four families. q) We are now going to calculate the character tables of an important in®nite series of groups. 3) is an abelian group. C and Z p ( p prime) are ®elds.28 Character table of GL(2.

every irreducible character of FÃ2 has the form q r 3 r j for some integer j. r 1‡q P S. (28. so s ‡ t P S. we shall identify the sub®eld S of F q 2 in Proposition 28. For every s P F q the sum of s with itself p times is zero.Character table of GL(2. 3) is cyclic. (1) The set S is a sub®eld of F of order q.1) Let p be a prime and n be a positive integer. Proof (1) Suppose that s. Then there exists a ®eld F q of order q and every ®eld of order q is isomorphic to F q . and write q ˆ pn . t P S. It is now easy to check that (S. 28. and hence S  F q . q) (28. so r ‡ r q . . We may write q r ˆ å m for some m and we let r ˆ ù m . in short. The group (Fà . Moreq over. we see that r q ˆ r for all q q q q q2 q r P F. This implies that (r ‡ r ) ˆ r ‡ r ˆ r ‡ r and (r 1‡q ) q ˆ j r 1‡q . Then (s ‡ t) q ˆ s q ‡ t q ˆ s ‡ t. Suppose that r P FÃ2 .2 Proposition Let F ˆ F q 2 and S ˆ fs P F : s q ˆ sg. 3) are abelian groups. You are now in a position to appreciate the statement of the main result of this chapter. it follows that (s ‡ t) p ˆ s p ‡ t p for all s. Then r 3 r is an irreducible character of FÃ2 . t P F q . namely Theorem 28. We introduce the following useful notation. so S is a ®eld.3) Let å be a generator of the cyclic group FÃ2 and let q 2 ù ˆ e(2ðia(q À1)) . 2 (2) Since FÃ2 is a group of order q 2 À 1. and k k k hence (s ‡ t) p ˆ s p ‡ t p for all positive integers k. Hereafter. ps ˆ 0. q 323 Notice that the binomial coef®cients ( ip ) with 1 < i < p À 1 are all divisible by p. We use this remark in the proof of the next proposition. (2) If r P F then r ‡ r q . ‡) and (Snf0g.5. r 1‡q P S.2 with the ®eld F q .

b). q). Here. jGj ˆ (q 2 À 1)(q 2 À q) ˆ q(q À 1)2 (q ‡ 1)X There are four families of conjugacy classes of G. and we talked about some special linear groups in the last chapter. we are going to calculate the character table of GL(2. The subgroup consisting of all matrices of determinant 1 is the special linear group SL(2. Let G ˆ GL(2. q). q). c9g. of which three are easy to describe. The number of such matrices is found by noting that (a. d) can be any row vector which is not a multiple of (a. since conjugate matrices have the same eigenvalues. First.324 Representations and characters of groups The conjugacy classes of GL(2. and once (a. giving us q 2 À q choices. cg ˆ fa9.   a b 0 c can be conjugate to  a9 0 b9 c9  only if fa. (c. consider the matrices  s us ˆ 0 Let give us q À 1 conjugacy classes of  1 s (s P Fà )X q . The matrices   s 0 sI ˆ (s P Fà ) q 0 s belong to the centre of G. b) can be any non-zero row vector. Next. b) has been chosen. giving us q 2 À 1 choices. and remember that the matrix   a b c d belongs to G if and only if its rows are linearly independent. q) is de®ned to be the group of invertible 2 3 2 matrices with entries in F q . q) The general linear group GL(2. Keep this in mind during the following discussion. Therefore. They size 1.

s X On the other hand. Now. t ˆ d s. by Theorem 12. then we have that gd s.2. Thus. so v r has eigenvalues r and r q.8. The characteristic polynomial of v r is det(xI À v r ) ˆ x(x À (r ‡ r q )) ‡ r 1‡q ˆ (x À r)(x À r q ). the matrices us (s P Fà ) give us q À 1 conjugacy classes. the matrices d s. 2 3 Àbr 1‡q a ‡ b(r ‡ r q ) gv r ˆ and Àdr 1‡q c ‡ d(r ‡ r q ) 2 3 c d vr g ˆ X Àar 1‡q ‡ c(r ‡ r q ) Àbr 1‡q ‡ d(r ‡ r q ) . if s Tˆ t. the q centralizer order is (q À 1)q. t ˆ P G (s. each conjugacy class contains q 2 À 1 elements. the centralizer order is (q À 1)2 . so. let   s 0 d s. Thus. t 0 1 1 0  ˆ d t. t g if and only if b ˆ c ˆ 0. Finally. consider   0 1 vr ˆ (r P F q 2 nF q )X Àr 1‡q r ‡ r q By Proposition 28. q)   a b P GX gˆ c d Then gus ˆ   and us g ˆ   325 as cs a ‡ bs c ‡ ds as cs d ‡ bs ds so g belongs to the centralizer of us if and only if c ˆ 0 and a ˆ d. t P Fà . v r P G. Now. s Tˆ t) give us q (q À 1)(q À 2)a2 conjugacy classes. t P Fà ) q 0 t and note that  0 1 1 0 À1  d s.Character table of GL(2. so each conjugacy class contains q(q ‡ 1) elements. Since r P F q we see that v r lies in a none of the conjugacy classes we have constructed so far. t (s.

so it is not conjugate to v r unless t ˆ r or t ˆ r q. of classes qÀ1 us (q À 1)q qÀ1 d s. 28. We have now found all the conjugacy classes of G. We therefore partition F q 2 nF q into subsets fr. described as follows. b) Tˆ (0. 0) and r. q q Proof The conjugacy classes we have found account for (q À 1) ‡ (q À 1)(q 2 À 1) ‡ (q À 1)(q À 2)q(q ‡ 1)a2 ‡ (q 2 À q)(q 2 À q)a2 elements altogether. Therefore. each subset gives us a conjugacy class representative v r and different subsets give us representatives of different conjugacy classes. But this sum is equal to the order of GL(2. tg of distinct elements of Fà . then ad À bc ˆ a2 ‡ ab(r ‡ r q ) ‡ b2 r 1‡q ˆ (a ‡ br)(a ‡ br q )X Since (a. jCG (v r )j ˆ q 2 À 1. g sI |CG ( g)| (q 2 À 1)(q 2 À q) No. we see that a ‡ br and a ‡ br q a are non-zero. g P CG (v r ) if and only if   a b gˆ X Àbr 1‡q a ‡ b(r ‡ r q ) Thus. r q g of elements of FÃ2 nFà . q). q The family of conjugacy class representatives v r is indexed by unordered pairs fr. r q g. q The family of conjugacy class representatives d s.326 Representations and characters of groups Hence gv r ˆ v r g only if c ˆ Àbr 1‡q and d ˆ a ‡ b(r ‡ r q ). If these conditions hold. so we have found all the conjugacy classes. t (q À 1)2 (q À 1)(q À 2)a2 vr q2 À 1 (q 2 À q)a2 The families of conjugacy class representatives sI and us are indexed by elements s of Fà . r q P F q. j . The matrix v t has eigenvalues t and t q.4 Proposition There are q 2 À 1 conjugacy classes in GL(2. and the conjugacy class containing v r has size q 2 À q. t is indexed by unordered pairs fs. q). Class rep.

5 Theorem Label the conjugacy classes of GL(2.6 Proposition Let K ˆ hvå i. Then the q irreducible characters of GL(2. sI ëi øi ø i. j we have 0 < i . Thus. q) The characters of GL(2. t (st) i (st) i si t j ‡ s j t i 0 vr r i(1‡q) Àr i(1‡q) 0 À(r i ‡ r iq ) Here.q) We are now in a position to describe the character table of G. ø i. q) are given by ë i . there are q À 1 characters ë i . if j1 and j2 belong to this set and j1  j2 q mod (q 2 À 1) then we choose precisely one of j1 and j2 to belong to the indexing set for the characters ÷ i . Then jKj ˆ q 2 À 1. we have the following restrictions on the subscripts.3). there are (q 2 À q)a2 characters ÷ i . we present a proposition which will be useful later. 327 28. there are q À 1 characters ø i . (b) For ø i we have 0 < i < q À 2. 1 X vå ˆ å ‡ åq Àå 1‡q 28. Hence. Thus. each of degree q. (d) For ÷ i.4. there are (q À 1)(q À 2)a2 characters ø i.Character table of GL(2. ÷ i as follows. (a) For ë i we have 0 < i < q À 2. q) as in Proposition 28. and of the remaining q 2 À q elements of K. Recall that å is our chosen generator for FÃ2 and q   0. . Thus. each of degree q ‡ 1. each of degree q À 1. and let r 3 r be the function from FÃ2 to C described in (28. j . j . (c) For ø i. The group K contains the q À 1 scalar matrices sI in G. we ®rst consider the set of integers j with 0 < j < q 2 À 1 and (q ‡ 1) T j j. each of degree 1. ø i . j ÷i s 2i qs 2i (q ‡ 1)s i‡ j (q À 1)s i us s 2i 0 s i‡ j Às i d s. Before we embark upon the task of calculating the irreducible characters of G. j < q À 2.

7 Proposition There are q À 1 linear characters ë i of G. j of G whose values on the conjugacy class representatives. 28.5. i iq The eigenvalues of vå and of vå are å i and å iq . the functions ë i : g 3 (det g) i ( g P G) give q À 1 distinct linear characters of G. Proof The map det : g 3 det g is a homomorphism from G onto Fà . as described in Proposition 28. in turn.4. since vå has eigenvalues in F q and vå is diagonalizable.328 Representations and characters of groups precisely two belong to each of the conjugacy classes represented by v r with r P FÃ2 nFà . i i i If å i ˆ å iq then vå ˆ å i I. j We will see later that the linear characters ë i (0 < i < q À 2) which appear in Proposition 28. q q Proof The eigenvalues of vå are å and å q. ø i.5. If å i Tˆ å iq then i i iq å P F q and vå and vå must be conjugate to vå i . j We shall construct. j there is a character ø i.5. whose values appear in Theorem 28. t si t j ‡ s j t i vr 0 .8 Proposition For all integers i. 28. Hence two elements a of K belong to the conjugacy class of vå i . j and ÷ i which appear in Theorem 28.7 are all the linear characters of G. are as follows. and this case accounts for the q À 1 scalar matrices. sI ø i. and they are given in Theorem 28. q As i varies between 0 and q À 2 inclusive. the irreducible characters ë i . j (q ‡ 1)s i‡ j us s i‡ j d s. ø i . so vå has order q 2 À 1.

t : g ˆ vr : ø i.i i ˆ (q ‡ 1)2 1 (q À 1) (q À 1) ‡ 2 À 1)(q 2 À q) (q À 1)q (q ‡ ˆ 2X Here. ø i. j ( g) ˆ ø i. g ˆ sI : g ˆ us : g ˆ d s.s jC B ( g)j jC B ( g9)j ø i. We use Proposition 21. we calculate hø i. j ( g) ˆ jC G ( g)j ë i.5. Proof We shall demonstrate that the character ø i. Remember that the complex conjugate of s i is s Ài . j ˆ ëi.i ˆ ë i ‡ ø i . j ( g9) ø i.Character table of GL(2. j ( g) jC B ( g)j   ë i. where g9 ˆ d t. We have hø i. there is an irreducible character ø i of G whose values are given in Theorem 28. the ®rst term corresponds to the conjugacy classes of elements 4 (q À 1)(q À 2) (q À 1)2 2 .i i and hø i.23 to calculate ø i. j ( g) jC B ( g)j jC G ( g)j ë i.i which appears in Proposition 28.i . the values of ø i. j 4 G. j ( g) ˆ jC G ( g)j ‡ . j ( g) ˆ 0X j Hence. The characters ø i for 0 < i < q À 2 are all different. ø i. j ( g) ë i.8 gives us ø i. j : B 3 C by   s r ë i. ë i i. as follows.i . To this end. j : 3 s i t jX 0 t Then ë i. q) Proof Let Bˆ &  ' PG X 329 a 0 b c Then B is a subgroup of G with jBj ˆ (q À 1)2 q. j is a character of B. j ( g) for each conjugacy class representative g. We let ø i.9 Proposition For each integer i. j are as stated in the proposition. De®ne ë i. 28.i .

10 Proposition Suppose that 0 < i . (1) ø i. we obtain hø i. ë i i ˆ 1 and hø i. Next.8 is irreducible.i i are calculated in a similar fashion. where (q ‡ 1)2 1 (q À 1) (q À 1). q Hence the characters ø i for 0 < i < q À 2 are all different. Then the character ø i. ø i.i .i (sI)ø i. t P Fà g is an abelian group of q order (q À 1)2. note that fd s.i (sI) ˆ (q ‡ 1)2 .i ˆ ë i ‡ ø i for some irreducible character ø i.i . j .330 Representations and characters of groups sI.i to get the values of ø i as given in Theorem 28. ë i i ˆ (q ‡ 1) 1 (q À 1) (q À 1) ‡ (q 2 À 1)(q 2 À q) (q À 1)q ‡ ˆ 1X The facts that hø i.i i ˆ 2 imply that ø i.i . q To evaluate C. tg of distinct elements of Fà . j which are given in Proposition 28. ø i. Let s be an element of Fà of order q À 1.8. hø i. Proof We shall show that hø i. Subtract ë i from ø i. j i ˆ 1. t 3 s i t j ‡ s j t i then ó is a sum of two . j < q À 2. j which appears in Proposition 28. t : s. Bˆ 2 À 1)(q 2 À q) (q À 1)q (q 1 1ˆ i j Cˆ (s t ‡ s j t i )(s Ài t À j ‡ s À j t Ài )X (q À 1)2 2 sTˆ t Aˆ and 2 (q À 1)(q À 2) 2 (q À 1) 2 The coef®cent 1 appears in C because we have just one conjugacy 2 class for each unordered pair fs. and the calculation of this ®rst term involves the following three observations. j i ˆ A ‡ B ‡ C. Then ø i : d s. and if ó : d s. j 28. j . The remaining terms in hø i.1 3 s i . Using the values of ø i.i .5. ø i. (2) jC G (sI)j ˆ (q 2 À 1)(q 2 À q)X (3) There are q À 1 conjugacy classes with representatives of the forms sI. ø i.

j9 ‡ ë j9. ø i. j ‡ ë j.i : 3 s i t j ‡ s j t iX 0 t Since ë i.11 Corollary The characters ø i. j Tˆ ø i9. there exists a character ö i of G which takes the following values. sI öi q(q À 1)s i us 0 d s.i Tˆ ë i9. j i ˆ 1. In either case. and ø i. j9 . j) Tˆ (i9. 28.Character table of GL(2. hø i. j . j9 on a conjugacy class j of G. for 0 < i . t 0 vr r i ‡ r iq . We have   s b ë i. q) 331 inequivalent irreducible characters of this group. j differs from ø i9. Therefore. j which were used in the proof of Proposition 28. j < q À 2 and 0 < i9 . j9 . 2 3 ˆ 1 i j j i Ài À j À j Ài 4(q À 1) ‡ (s t ‡ s t )(s t ‡ s t ) ˆ 2X (q À 1)2 sTˆ t Hence. j 28. hó . we see that ø i. ø i.8. j is irreducible. there exists complex (q À 1)th roots of unity s and t such that either s Tˆ t and s i t j ‡ s j t i Tˆ s i9 t j9 ‡ s j9 t i9 or s ˆ t and s i‡ j Tˆ s i9‡ j9 . j Tˆ ø i9. and (i. Thus.i9 . Therefore. ó i ˆ 2X That is. j < q À 2 are distinct irreducible Proof Suppose that 0 < i . j characters of G. Cˆ qÀ3 X qÀ1 And now we ®nd that A ‡ B ‡ C ˆ 1. j9). j9 < q À 2. j ‡ ë j. We must prove that ø i. Consider the group B and its linear characters ë i.12 Proposition For each integer i.

332 Representations and characters of groups Proof Let K ˆ hvå i. v r is conjugate to an element of g of K. Then. t (s Tˆ t). Hence á i ( g) ˆ r i or r iq and á i ( g) ‡ á i ( g q ) ˆ r i ‡ r iq X Let ö i ˆ á i 4 G. If g ˆ sI with s P Fà then g P K and q ö i ( g) ˆ jC G ( g)j á i ( g) ˆ q(q À 1)s i X jC K ( g)j Suppose that r P F q 2 nF q . by Proposition 28. ö i has the values stated in the proposition. Then ˆ (r i ‡ r iq )(r Ài ‡ r Àiq ) ˆ 2(q À 1)2 X rPF q 2 nF q Proof Note that & r G1 ˆ 0 0 rq  : r P FÃ2 q ' and G2 ˆ & r 0 0 rq  : r P Fà q ' are abelian groups of orders q 2 À 1 and q À 1.6. Suppose that g P K and g is conjugate in G to v r .6. . as in Proposition 28.   á i ( g) ái( g q) ö i ( g) ˆ jC G ( g)j ‡ jC K ( g)j jC K ( g)j ˆ á i ( g) ‡ á i ( g q ) ˆ r i ‡ r iq X Thus. 28. respectively. we shall the use the following lemma. Also. Thus. ®rst recall that á i 4 G is zero on all elements which are not conjugate to an element of K. In order to calculate ö i .6. and consider the linear character á i of K which sends the generator vå of K to å i . Now. ö i is zero on the elements of the form u s and d s. j To be able to work out certain inner products involving our characters ö i . by Proposition 28. Then g has eigenvalues r and r q .13 Lemma Assume that i is an integer and (q ‡ 1) T j i.

Recall the characters ø i. since (q ‡ 1) T j i implies that å i Tˆ å iq . the character ÷ is twice an irreducible character. ÷ i is the class function on G which is given by ÷ i ˆ ø0. ø i and ö i given in Propositions 28. j . the character ÷ is a sum of two inequivalent irreducible characters. For G1.8. Now. 28. q)   r 0 3 r i ‡ r iq 0 rq 333 gives a character ÷ of degree 2 for each group. and for G2. sI ÷i (q À 1)s i us Às i d s.i À ö i X . Proof We can justify the manoeuvre which we now perform only by saying that it gives the correct answer.9 and 28. Taking the inner product of the q character ÷ of G1 with itself.14 Proposition For each integer i. t 0 vr À(r i ‡ r iq ) If (q ‡ 1) T j i then ÷ i is an irreducible character of G. we get 1 ˆ i (r ‡ r iq )(r Ài ‡ r Àiq ) ˆ 4X qÀ1 à rPF q Hence ˆ rPF q 2 nF q (r i ‡ r iq )(r Ài ‡ r Àiq ) ˆ 2(q 2 À 1) À 4(q À 1) ˆ 2(q À 1)2 X j 28. let ÷ i be the class function on G with the following values.Ài ø i À ø0. since r q ˆ r for r P FÃ. we get 1 ˆ i (r ‡ r iq )(r Ài ‡ r Àiq ) ˆ 2 2À1 q rPFà 2 q and doing the same for the character ÷ of G2 .Character table of GL(2.12.

It is possible to use the character table of GL(2. as we wished to show. ÷ i i ˆ 1 and ÷ i (1) . Proof Let K ˆ hvå i. with integer coef®cients.i öi ÷i (q ‡ 1)s Ài qs 2i q(q ‡ 1)s i (q ‡ 1)s i q(q À 1)s i (q À 1)s i us s Ài 0 0 si 0 Às Ài d s. ÷ i i ˆ (q À 1)2 1 (q À 1)2 (q À 1) ‡ 2 (q À 1) ‡ 2 ˆ 1X q Àq (q 2 À 1)(q 2 À q) q À1 Since ÷ i is a linear combination of irreducible characters of G. iq mod(q 2 À 1). ÷ i Tˆ ÷ j .334 Representations and characters of groups The table below allows us to verify this. Therefore. assume that (q ‡ 1) T j i. iq mod(q 2 À 1). Then the characters ÷ i and ÷ j of G are different. and h÷ i . sI ø0. it follows that ÷ i is an irreducible character of G.13. which is the same as the number of conjugacy classes of G. so either s i Tˆ s j for some s P Fà or q r i ‡ r iq Tˆ r j ‡ r jq for some r P F q 2 nF q . the characters á i ‡ á iq and á j ‡ á jq of K are different. and consider the linear character á i of K which sends the generator vå of K to å i . since we have shown that the class functions given in the theorem are inequivalent irreducible characters.Ài øi ø0.15 Proposition Suppose that i and j are integers with (q ‡ 1) T j i and (q ‡ 1) T j j and j T i. as in Proposition 28. and the number of them is q 2 À 1. j 28. Suppose that g P K. Since j T i. t s Ài ‡ t Ài (st) i i s ‡ ti si ‡ t i 0 0 vr 0 Àr i(1‡q) 0 0 r i ‡ r iq À(r i ‡ r iq ) Next. ÷ i i using Lemma 28.5. We work out h÷ i . q) to ®nd the . If g ˆ sI where s P Fà then (á i ‡ á iq )( g) ˆ 2s i . q If g is conjugate to v r where r P F q 2 nF q then (á i ‡ á iq )( g) ˆ r i ‡ r iq . h÷ i .6. 0.Ài ø i ø0. j We have now completed the proof of Theorem 28.

you are asked to consider the easiest case. though. since most of the irreducible characters remain irreducible when restricted. q) ± compare Chapter 27 ± and so the character table of PSL(2. q) was ®rst given in 1907. in 1955.Character table of GL(2. A challenging exercise is to determine the character table of PSL(2. those with kernel containing the centre of SL(2. (d) There are (q 2 À q)a2 conjugacy classes with representatives of the . q) when q is a power of 2. and there are q À 1 irreducible characters of degree 0 s q. q) was found. and they depend upon whether q is a power of 2 or q  1 mod 4 or q  3 mod 4.2. Then. A. q) has the following properties. In Exercise 28. Green determined the character table of GL(n. Since SL(2. (b) There are q À 1 conjugacy classes with representatives of the form   s 1 us ˆ . and there are (q À 1)(q À 2)a2 0 t irreducible characters of degree q ‡ 1. J. q) 335 character table of SL(2. and there are q À 1 irreducible characters of 0 s degree 1. it was not until the 1950's that the character table of GL(3. q). q) when q  1 mod 4 or q  3 mod 4 from the character table of GL(2. q) is rather easier to ®nd than that of SL(2. q)  PSL(2. Although the character table of GL(2. (a) Thereare q  1 conjugacy classes with representatives of the form À s 0 sI ˆ . q). t ˆ (s Tˆ t). We do not go fully into this. this gives the character tables of an in®nite series of simple groups PSL(2. Among the characters of SL(2. (c) There are (q À 1)(q À 2)a2 conjugacy classes with representatives   s 0 of the form d s. q). q) for all positive integers n. namely that where q is a power of 2. q). q). Summary of Chapter 28 The character table of GL(2. since the answers are quite complicated. q) provide the characters of the groups PSL(2.

Exercises for Chapter 28 1. 3). Use your solution to Exercise 28. Let Z ˆ fsI : s P Fà g. q).336 Representations and characters of groups   0 1 . Use Theorem 28. q)  Z 3 SL(2. and there are (q 2 À q)a2 irreducible form v r ˆ Àr 1‡q r ‡ r q characters of degree q À 1. Prove that q GL(2. q) from that of GL(2. 2.2 to write down explicitly the character table of PSL(2. Suppose that q is a power of 2. 8). q)X Deduce the character table of SL(2. q) is simple. 3. . Prove that if q Tˆ 2 then SL(2.5 to write down explicitly the character table of GL(2.

3g 3 f2. An action of G on Ù is a homomorphism ö: G 3 Sym(Ù).e.) Check that ö is an action of Sn .1 Examples (1) If G < Sn then the identity map is an action of G on f1. 2.29 Permutations and characters We have already seen in Chapter 13 that if G is a permutation group. In particular. ng. and develop some useful results. F F F . jg of elements of f1. ng then Sym(Ù) ˆ Sn . We also say that G acts on Ù (via ö).12 below). if Ù ˆ f1. If Ù is a set. F F F . jgg for all g P Sn and 1 < i . 3g. it is called the action of Sn on pairs. 29. jg( gö) ˆ fig. a fact which proved useful in many of our subsequent character table calculations. j < n. (2) Let G ˆ Sn and let Ù be the set consisting of all pairs fi. (So for example. a subgroup of Sn for some n. then G has a permutation character ð de®ned by ð( g) ˆ jfix( g)j for g P G. denote by Sym(Ù) the group of all permutations of Ù. ng. F F F . Group actions We begin with a more general notion than that of a permutation group. particularly about irreducible characters of symmetric groups (see Theorem 29. (1 2)ö sends f1. De®nition Let G be a group and Ù a set. i. De®ne ö: G 3 Sym(Ù) by setting fi. 337 . In this chapter we take the theory of permutation groups and characters somewhat further.

Let V be the 2-dimensional vector space over F q consisting of all row vectors (a. b)i 3 h(a. b) with a. De®ne ö: G 3 Sym(Ù) by setting hvi( gö) ˆ hv gi for all hvi P Ù and g P G. if   1 1 gˆ 0 1 then gö sends h(a. b P F q . if ö: G 3 Sym(Ù) is an action. generated by x. de®ne a relation $ on Ù as follows: for á. Then by Exercise 9 of Chapter 23. Write orb(G. It is easy to see that $ is an equivalence relation on Ù. in other words. g P G. q). Ù) ˆ 1. the group of invertible 2 3 2 matrices over the ®nite ®eld F q. The group G is said to be transitive on Ù if orb(G. Thus Ù is the disjoint union of the orbits of G. â P Ù. Adopting this notation. there exists g P G such that á g ˆ â. With this notation. and let Ù be the set of all 1-dimensional subspaces hvi of V. we have á $ â if and only if there exists g P G such that á g ˆ â. the fact that ö is a homomorphism simply says that ù( gh) ˆ (ù g)h for all ù P Ù and g. G is transitive if. To simplify notation. The equivalence classes are called the orbits of G on Ù. given any á. a ‡ b)iX Then ö is an action of G on Ù. â P Ù. for ù P Ù and g P G we usually just write ù g instead of ù( gö).2 Examples (1) Let G ˆ C4 . ö is an action of „ G. Ù) for the number of orbits of G on Ù.338 Representations and characters of groups (3) Let G ˆ GL(2. h P G. For example. as de®ned in Chapter 28. 29. De®ne ö: G 3 Sym(Ù) by ( Hx)( gö) ˆ Hxg for all x. and let Ù be the set of all right cosets Hx of H in G (so jÙj ˆ n). (4) Let G be a group with a subgroup H of index n. and let ö: G 3 S8 be the action . and Ker ö ˆ xPG x À1 Hx < H. say.

the size of the orbit ù G is equal to the index of Gù in G.3 Proposition The stabilizer Gù is a subgroup of G. Also gÀ1 P Gù . CÙ consists of all expressions of the form .Permutations and characters 339 de®ned by xö ˆ (1 2 3 4)(5 6)(7 8) (and of course x k ö ˆ ((1 2 3 4)(5 6)(7 8)) k for any k). Hy P Ù. and in Example (4). 3. h P Gù then ù( gh) ˆ (ù g)h ˆ ùh ˆ ù. F F F . For ù P Ù. q) such that vA ˆ w. Moreover. so Gù is a subgroup. that is. 6g and f7. Then G has three orbits on Ù ˆ f1. In other words. namely f1. to verify it for Example (3) you need to convince yourself that for any two non-zero row vectors v. write ù G for the orbit of G which contains ù. 3. Clearly ã is also surjective. 29. 8g. given two right cosets Hx. j Permutation characters Let G be a group acting on a ®nite set Ù. and Gù contains the identity. 4g. and hence jÄj ˆ jù G j. as required. simply observe that. This is clear in Example (2). w P V there is an invertible 2 3 2 matrix A P GL(2.1(2. the element g ˆ x À1 y P G has the property that ( Hx) g ˆ Hy. (2) The group G is transitive on the set Ù in each of Examples 29. hence gh P Gù . and de®ne Gù ˆ f g P G : ù g ˆ ùgX We call Gù the stabilizer of ù in G. f5. 8g. y P G. 2. jù G j ˆ jG : Gù jX Proof If g. so ù G ˆ fù g : g P Gg. Observe that for x. 4). Let G be a group acting on a set Ù. Now let Ä be the set of right cosets Gù x of Gù in G. Denote by CÙ the vector space over C for which Ù is a basis. Gù x ˆ Gù y D xy À1 P Gù D ùxy À1 ˆ ù D ùx ˆ ù yX Hence we can de®ne an injective function ã : Ä 3 ù G by ã(Gù x) ˆ ùx for all x P G.

then for g P G. 1 G i ˆ 1 ˆ 1 ˆ ð( g) ˆ jfixÙ ( g)jX jGj gPG jGj gPG Let Ä1 . for each g. As in Chapter 13. It is often referred to as ``Burnside's Lemma''. ù g ˆ ùg. by de®ning 2 3 ˆ ˆ ëù ù g ˆ ëù (ù g) for all g P G. we can make CÙ into a CG-module. 1 G i ˆ jfixÙ ( g)j ˆ orb(G. We see just as in Chapter 13 (p. pick ù i P Ä i . 29.340 Representations and characters of groups ˆ ëù ù (ëù P C) ùPÙ with the obvious addition and scalar multiplication. but is in fact due to Cauchy and Frobenius. called the permutation module. for 1 < i < t we have jÄ i j ˆ jù G j ˆ jG : Gù i jX i Hence jÄ i j jGù i j ˆ jGj. and provides a basic link between the permutation character and the action of G. and let ð be the permutation character. hence . is rather famous. We calculate jÖj in two different ways. the number of ù P Ù such that ù g ˆ ù is equal to jfixÙ ( g)j. 129) that if ð is the character of this permutation module. We call ð the permutation character of G on Ù. By Proposition 29. and for each i. g) : ù P Ù. F F F .4 Proposition Let G be a group acting on a ®nite set Ù. ð( g) ˆ jfixÙ ( g)j. Ù)X jGj gPG Proof First note that hð.3. though elementary. First. Ä t be the orbits of G on Ù. The next result. g P G. Now de®ne Ö ˆ f(ù. Then 1 ˆ hð. where fixÙ ( g) ˆ fù P Ù : ù g ˆ ùg.

4. jGj gPG jGj gPG j which is equal to orb(G. ð2 i ˆ orb(G. Then we can de®ne an action of G on the Cartesian product Ù1 3 Ù2 by setting (ù1 . 29. Ù1 3 Ù2 ) by Proposition 29. ð2 i ˆ 1 ˆ 1 ˆ jfixÙ1 ( g)kfixÙ2 ( g)j ˆ jfixÙ1 3Ù2 ( g)j. Then hð1 . and the conclusion follows. Hence if ð is the permutation character of G on Ù1 3 Ù2 .5 Corollary G is transitive on Ù if and only if hð. 29. ù2 g) for all ù i P Ù i . g P G. for each ù. It is clear that fixÙ1 3Ù2 ( g) ˆ fixÙ1 ( g) 3 fixÙ2 ( g) for any g P G. the number of g P G such that ù g ˆ ù is equal to jGù j. In the rest of the chapter we apply Proposition 29. with permutation characters ð1 and ð2 respectively. ù2 ) g ˆ (ù1 g. the ®rst being the case where Ù1 ˆ Ù2 . 1 G i ˆ 1. hence jÖj ˆ Therefore € gPG ˆ ùPÙ jGù j ˆ t ˆ iˆ1 jÄ i j jGù i j ˆ t ˆ 1 jGj ˆ tjGjX j jfixÙ ( g)j ˆ tjGj.6 in a number of situations.Permutations and characters ˆ jÖj ˆ jfixÙ ( g)jX gPG 341 Secondly. Ù1 3 Ù2 )X Proof We have hð1 . Now let G be a group. . then ð( g) ˆ ð1 ( g)ð2 ( g) for all g P G.6 Proposition Let G act on Ù1 and Ù2 . with corresponding permutation characters ð1 and ð2 respectively. and suppose that G acts on two sets Ù1 and Ù2 .

29. G is 2-transitive if.8 Proposition Let G act on Ù. Then Ä ˆ f(ù.9 De®nition Let G be transitive on Ù. there exists g P G such that á1 g ˆ â1 and á2 g ˆ â2 . namely (ù1 . j . Then r(G. written r(G. â2 ) in Ù 3 Ù. ù2 g) for all ù1 . Ù 3 Ù)X The next result is immediate from Proposition 29. The result follows. 1.8. ù) : ù P Ùg is an orbit of G on Ù 3 Ù.17.6. Ù) > 2. Then G is said to be 2-transitive on Ù if r(G. and hð. ù2 ) g ˆ (ù1 g. where ÷ is an irreducible character of G. ði ˆ 2 by Proposition 29. Ù) ˆ 2. Ù) ˆ orb(G. for any ordered pairs (á1 . Ù) ˆ hð.10 Corollary If G is 2-transitive on Ù. 29. In other words. and hence certainly r(G. 29. Then G also acts on Ù 3 Ù in the way de®ned above. with permutation character ð. â1 Tˆ â2 . ù2 P Ù. then ð ˆ 1 G ‡ ÷. á2 ) and (â1 . with á1 Tˆ á2 .5. The case where equality holds is of particular interest. Ù). Thus r(G.342 Representations and characters of groups Suppose G acts on Ù. Proof We have hð.7 De®nition The number of orbits of G on Ù 3 Ù is called the rank of G on Ù. 29. with permutation character ð. ðiX Now suppose G is transitive on Ù and jÙj . 1 G i ˆ 1 by Corollary 29. using Theorem 14. g P G.

Also An is 2-transitive. and we call such a sequence a partition of n. Ä1 and Ä2 . v2 3 w2 is therefore invertible.5. q) which sends hv1 i 3 hw1 i. jg ’ fk. fk.3.17). F F F . ði ˆ r(G.15 we know that the conjugacy classes of Sn are in bijective correspondence with the set of all possible cycle-shapes of permutations. f2. the irreducible character ÷ of G given by Corollary 29. there is no element of Sn which sends (f1. Ù) ˆ 3. as claimed.11 Examples (1) The symmetric group Sn is 2-transitive on f1. 4g) to (f1. By Theorem 15. q) given in Example 29. with n > 4.1(3). w2 are both bases of V. Here Ù is the set of all 1-dimensional subspaces of the 2-dimensional vector space V. since. an irreducible character ÷ ë . lgj ˆ 1g. where ÷ and ø are irreducible characters of Sn . 2g. ë s ) of positive integers ë i such that ë1 > ë2 > F F F > ë s and ë1 ‡ F F F ‡ ë s ˆ n. 3g).1. lgj ˆ 0gX Thus hð. hv2 i 3 hw2 i. provided n > 4. for example. Then v1 .1(2). Ä2 ˆ f(fi.16. hv2 i) and (hw1 i. A key aim is therefore to construct. v2 and w1. jg. 19. and so ð ˆ 1 G ‡ ÷ ‡ ø. f3. the irreducible characters of Sn are also in bijective correspondence with the partitions ë of n. To see this. giving an element of GL(2. Some irreducible characters of S n By Theorem 12. F F F . hw2 i) be two pairs of distinct 1-spaces in Ù. 19.10 is the character ø0 of degree q in Theorem 28. for each partition ë. This action is not 2-transitive. let (hv1 i. lg) : jfi. Each cycle-shape (including 1-cycles) is a sequence ë ˆ (ë1 . Hence these groups have an irreducible character ÷ given by ÷( g) ˆ jfix( g)j À 1X We have seen this irreducible character in a number of previous examples (see 18. and Ä1 ˆ f(fi. 2g. Hence G is 2-transitive on Ù. We claim that G is 2-transitive on Ù.Permutations and characters 343 29. fk. The linear transformation from V to V which sends v1 3 w1 . Since jÙj ˆ q ‡ 1 (see Exercise 1 at the end of the chapter). jg ’ fk. (2) Consider the action of G ˆ GL(2. ng. jg. (3) Consider the action of Sn on pairs de®ned in Example 29. lg) : jfi. In fact it is easy to see that the orbits of G ˆ Sn on Ù 3 Ù are Ä. where Ä is as above.

13 Theorem Let m ˆ na2 if n is even. F F F . Observe that   n ð k (1) ˆ jI k j ˆ X k 29. k) ˆ ð k À ð kÀ1 .1) . j 29. Now assume the statement holds for all values less than k. This holds for k ˆ 1 by Corollary 29. F F F . where for 0 < s < l. J s ˆ f(A. ð k ˆ ÷ ( n) ‡ ÷ ( nÀ1. We shall apply the material of this chapter to carry out this aim in the case where ë ˆ (n À k. ð l i ˆ l ‡ 1. we refer you to the book by G. 2. I k 3 I l ) ˆ l ‡ 1. k). ÷ ( nÀ1. The ideas can be developed to carry out the aim in general. by induction on k. but we do not do this.1(2) we can de®ne an action of G on I k as follows: for any subset A ˆ fi1 .2) . ÷ ( nÀ1. ð l i ˆ orb(G. ik g P I k and any g P G.12 Proposition If l < k < na2. m) such that for all k < m. ik ggX Let ð k be the permutation character of G in its action on I k .1) ‡ F F F ‡ ÷ ( nÀ k. a 2-part partition (see Theorem 29. ÷ ( nÀ2. F F F . The orbits of G ˆ Sn on I k 3 I l are easily seen to be J 0 . F F F J l . B) P I k 3 I l : jA ’ Bj ˆ sgX Hence orb(G. ng. k) such that ð k ˆ ÷ ( n) ‡ ÷ ( nÀ1. ÷ ( nÀ k. F F F . Then Sn has distinct irreducible characters ÷ ( n) ˆ 1 G .13 below). then hð k . Proof By Proposition 29. I k 3 I l ).1) ‡ F F F ‡ ÷ ( nÀ k. if you want to see this. Then . James listed in the Bibliography. and m ˆ (n À 1)a2 if n is odd. and much more. let Ag ˆ fi1 g. de®ne I k to be the set consisting of all subsets of I of size k. For an integer k < na2. ÷ ( nÀ k. in a natural way. hð k . giving the conclusion. on the character theory of Sn . Just as in Example 29. k) X In particular. Let G ˆ Sn and I ˆ f1. Proof We prove the existence of irreducible characters ÷ ( n) .6.344 Representations and characters of groups of Sn . J 1 .10. F F F . ÷ ( nÀ m.1) . k) .

÷3 . hð k . k) ÷ (1) ˆ ð k (1) À ð kÀ1 (1) ˆ À X k kÀ1 As another example.1) ‡ F F F ‡ ÷ ( nÀi. and the corresponding character of G is ð. kÀ1) such that ð i ˆ ÷ ( n) ‡ ÷ ( nÀ1. k) . k) corresponding to 2-part partitions have values given by Theorem 29.1) ‡ F F F ‡ ÷ ( nÀ k. 1 G i. ÷ ( nÀ k‡1. An action of G on Ù is a homomorphism G 3 Sym(Ù). and r(G. ÷ (4. 2. as required. hð k . Summary of Chapter 29 1. The number of orbits is equal to hð. The rank r(G. j 29. If G acts on Ù then CÙ is the permutation module.2) on a 3-cycle: ÷ (5. The irreducible characters ÷ ( nÀ k.12. F F F . Ù) ˆ hð. hð k . .1) .1) .2) . If G is 2-transitive then r(G. ÷ (3. 1 G i ˆ 1. ði. 4. k) . Ù) ˆ 2 and ð ˆ 1 G ‡ ÷ with ÷ irreducible. k) ˆ ð k À ð kÀ1 makes it easy to calculate the values of the characters ÷ ( nÀ k.2) (123) ˆ ð2 (123) À ð1 (123) ˆ jfix I 2 (123)j À jfix I 1 (123)j ˆ 6 À 4 ˆ 2X (2) In the character table of S6 given in Example 19. respectively.17.14 Examples (1) The formula ÷ ( nÀ k. ð1 i ˆ 2. k) .13. F F F .i) for all i . Ù) is the number of orbits of G on Ù 3 Ù. the irreducible characters ÷1 . where ð( g) ˆ jfixÙ ( g)j. ÷ (5. the degree is     n n ( nÀ k. hð k . ÷7 . ð k i ˆ k ‡ 1X It follows that ð k ˆ ð kÀ1 ‡ ÷ for some irreducible character ÷.Permutations and characters 345 there exist irreducible characters ÷ ( n) . The irreducible characters of Sn are in bijective correspondence with partitions of n. Now by Proposition 29.3) . ÷ ( nÀ1. ÷9 are equal to ÷ (6) . Writing ÷ ˆ ÷ ( nÀ k. The size of the orbit ù G containing ù is jù G j ˆ jG : Gù j. ð kÀ1 i ˆ k. The orbits are the equivalence classes in Ù of the relation de®ned by á $ â D á g ˆ â for some g P G. k. 3. suppose n ˆ 7 and let us calculate the value of the irreducible character ÷ (5. we have ð k ˆ ÷ ( n) ‡ ÷ ( nÀ1. For example.

For i ˆ 1. and ÷ 3 ÷ is the irreducible character of G 3 G given by Theorem 19.) . 2 de®ne Ù i to be the set of right cosets of H i in G. ( Hint: one way to do this is to write down the values of ð on the conjugacy classes of G. G) is equal to the number of conjugacy classes of G. h)ö) ˆ g À1 xh for all x. (b) Find the stabilizer (G 3 G)1 of the identity 1 P G. Suppose that ð1 ˆ ð2 . Prove that if G is abelian. so that G acts on Ù i as in Example 29.1(2). g.5.18. (a) Show that ö is an action of G 3 G on G. H 2 be subgroups of G.1(4). g P G. and ®nd the kernel of ö. and de®ne an action ö : G 3 Sym(V à ) by v( gö) ˆ v g for v P V Ã. h P G. q) and let V ˆ F2 as in Example 29. Let G be a ®nite group acting transitively on a set Ù of size greater than 1. let ð i be the permutation character of G in the action on Ù i . then H 1 ˆ H 2 X Give an example to show that this need not be the case in general. Decompose ð as a sum of irreducible characters of GL(2. (c) Show that the rank of this action r(G 3 G. Let G be a ®nite group. and let H 1 . 3.) 4. and the permutation character ð is ðˆ ˆ ÷ ÷ 3 ÷. and de®ne a function ö : G 3 G 3 Sym(G) by x(( g.1(3)). Prove that G contains an element g such that jfixÙ ( g)j ˆ 0. then jÙj ˆ q ‡ 1. where the sum is over all irreducible characters ÷ of G. 2. (Such an element is called a ®xed-point-free element of G. Let q V à ˆ V À f0g. Let G ˆ GL(2. which is transitive. Show that if Ù is the set of all 1-dimensional subspaces of a 2dimensional vector space over F q (as in Example 29. Let G be a ®nite group. q) (the latter are given by Theorem 28. Let ð be the permutation character of G in this action. and take inner products with the irreducible characters of G given in 28. 5.346 Representations and characters of groups Exercises for Chapter 29 1.5).

(i. and let the permutation character of Sn in this action be ð( nÀ2.1) .1) . which irreducible character is equal to ÷ (4. and let Ù be the set of all ordered pairs (i.1.1. prove that ð( nÀ2.17. j) g ˆ (ig. where ÷ is an irreducible character. ng and i Tˆ j. calculate the degree of ÷ ( nÀ2.1) .1) ˆ 1 ‡ 2÷ ( nÀ1. Let n be a positive integer.13. j P f1.1.1.1. By considering inner products as in the proof of Theorem 29. jg) for g P Sn ). j) with i. Writing ÷ ˆ ÷ ( nÀ2. In the character table of S6 given in Example 19. F F F . and calculate its value on the elements (12) and (123) of Sn .1) ‡ ÷ ( nÀ2.2) ‡ ÷.1) ? . Let Sn act on Ù in the obvious way (namely.Permutations and characters 347 6.

The second application takes this much further: the Brauer±Fowler Theorem 23.22 that the class sums C1 . . . known as the class algebra constants. C l form a basis for the centre of the group algebra CG € (where C i ˆ gPC i g).30 Applications to group theory There are several ways of using the character theory of a group to determine information about the structure of the group. seeing whether or not the group is simple. Class algebra constants Let G be a ®nite group and let C1 . .1 Proposition There exist non-negative integers aijk such that for 1 < i < l and 1 < j < l. . 348 . In this chapter we present some rather deeper applications. . Cl be the distinct conjugacy classes of G. Using a little group theory and a lot of character theory we shall carry out such a study in the case where C ˆ D8 . and so on ± require little calculation. as we shall demonstrate. These constants carry information about the multiplication in G. . 30.19 motivates the study of simple groups containing an involution with centralizer isomorphic to a given group C. . The ®rst involves doing arithmetic with character values to determine certain numbers. . Recall from Proposition 12. The examples which we have come across so far ± ®nding the centre of the group. and they can be used to investigate the subgroup structure of G. the dihedral group of order 8.

As the centre of the group algebra plays an important role in representation theory. . C l is a basis of Z(CG). b) with a P Ci . 30. 30. C l . The result follows. . b P Cj and ab ˆ g. the numbers aijk carry information about the multiplication in G: (30X3) For all g P Ck and all i. . so it must be a linear combination of C1 .Applications to group theory Ci C j ˆ l ˆ kˆ1 349 aijk C k X Proof For g P Ck the coef®cient of g in the product C i C j is equal to the number of pairs (a. the constants aijk determine the product of any two elements in the centre Z(CG) of the group algebra. b) with a P Ci . . Then for all i. and is independent of the chosen element g of Ck . . we have aijk ˆ ˆ ÷( g i )÷( g j )÷( g k ) jGj jCG ( g i )j jCG ( g j )j ÷ ÷(1) where the sum is over all the irreducible characters ÷ of G.4 Theorem Let gi P Ci for 1 < i < l. b P Cj and ab ˆ gX Also. . . j we have aijk ˆ the number of pairs (a. k.1 is to note that C i C j belongs to Z(CG). since C1 . Our next theorem shows that this is indeed the case. j.2 De®nition The integers aijk in the formula Ci C j ˆ l ˆ kˆ1 aijk C k are the class algebra constants of G. This number is a non-negative integer. . . From their very de®nition. you might suspect that the class algebra constants are determined by the character table of G. j Another way of looking at Proposition 30.

for all u P U we have uC i ˆ Therefore uC i C j ˆ and l ˆ mˆ1 jGj÷( g i ) uX jCG ( g i )j÷(1) ÷( g i )÷( g j ) jGj2 u jCG ( g i )j jCG ( g j )j (÷(1))2 aijm uC m ˆ l ˆ mˆ1 aijm jGj÷( g m ) uX jCG ( g m )j÷(1) Since C i C j ˆ (30X5) € m aijm C m .6 Example In this example we shall use the class algebra constants to prove some facts about the elements and subgroups of the symmetric group S4 . Multiply both sides of equation (30.350 Representations and characters of groups Proof Let ÷ be an irreducible character of G. we deduce that l ˆ mˆ1 aijm ÷( g i )÷( g j ) ÷( g m ) jGj ˆ X jCG ( g m )j jCG ( g i )j jCG ( g j )j ÷(1) Pick k with 1 < k < l.5) by ÷( g k ) and sum over all irreducible characters ÷ of G. this yields aijk ˆ ˆ ÷( g i )÷( g j )÷( g k ) jGj X jCG ( g i )j jCG ( g j )j ÷ ÷(1) j Examples 30.4(2). and let U be a CGmodule with character ÷. Theorem 16. to obtain l ˆ mˆ1 aijm ˆ ÷( g m )÷( g k ) ÷ jCG ( g m )j ˆ ˆ ÷( g i )÷( g j )÷( g k ) jGj X jCG ( g i )j jCG ( g j )j ÷ ÷(1) By the column orthogonality relations.7. the character table of G is as shown: . Then by Lemma 22. By Section 18. but they serve as a useful illustration of the method.1. these results can readily be proved directly. Let G ˆ S4 .

b of order 2 such that ab has order 4. bl  D8. S4 does not possess elements a. (3) Finally. We deduce from this that S4 does not have a subgroup which is isomorphic to the quaternion group Q8 : for Q8 does have two elements of order 4 with product of order 4.1) that S4 has a subgroup which is isomorphic to D8.4 to calculate the class algebra constant a555 :   24 1 À1 0 À1 1 ‡ ‡ ‡ ‡ ˆ 0X a555 ˆ X 44 1 1 2 3 3 Hence.   24 1 1 a245 ˆ X 1‡1‡ ‡ ˆ 2X 48 3 3 Hence S4 has elements a. In fact. 43 so S4 has elements a of order 2 and b of order 3 with ab of order 4. (2) By Theorem 30. aÀ1 xa ˆ ba ˆ (ab)À1 ˆ x À1 . by (30.4. and all products of elements of S4 are determined by the given relations. S4 is generated by a and b. so ka. We deduce the fact (which we already know from Exercise 18. we have x 4 ˆ 1. 24 a235 ˆ X (1 ‡ 1) ˆ 4.Applications to group theory Character table of S4 Class Ci gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 C1 1 24 1 1 2 3 3 C2 (1 2) 4 1 À1 0 1 À1 C3 (1 2 3) 3 1 1 À1 0 0 C4 (1 2)(3 4) 8 1 1 2 À1 À1 351 C5 (1 2 3 4) 4 1 À1 0 À1 1 (1) We use Theorem 30. Writing x ˆ ab. it can be shown that S4 has a presentation as follows: S4 ˆ ha. b: a2 ˆ b3 ˆ (ab)4 ˆ 1iX In other words. We supply a . b of order 4 such that the product ab also has order 4.3).

That such a subgroup exists is not obvious.4 to ®nd a subgroup H of the simple group PSL (2. 30. 7) Class rep. 7) with H isomorphic to S4 . you may wish to puzzle out the relevance of the ®gure above. 7) is as follows. Character table of PSL (2.6 ± in the meantime. g1 1 168 1 7 8 3 3 6 g2 2 8 1 À1 0 À1 À1 2 g3 4 4 1 À1 0 1 1 0 g4 3 3 1 1 À1 0 0 0 g5 7 7 1 0 1 á á À1 g6 7 7 1 0 1 á á À1 .7 Example We use Theorem 30. We found in Chapter 27 that the character table of G ˆ PSL (2. gi Order of gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 p where á ˆ (À1 ‡ i 7)a2.352 Representations and characters of groups proof in the solution to Exercise 30. and it is quite tricky to construct directly.

given a ®nite group C. namely xy. Then G has order 168 or 360.3). Gorenstein listed in the Bibliography). This programme formed an important part of the effort of many mathematicians to classify all the ®nite simple groups.10. The Brauer programme The Brauer±Fowler Theorem 23. yl of G. is {1}. We have chosen to present this result because it provides a wonderful illustration of the use of character theory in the service of group theory. The next result carries out part of Brauer's programme in the case where C ˆ D8 . C2 or {1}. It determines the possible orders of simple groups G having an involution t such that C G (t)  D8 . V4 . By Theorem 1. By Theorem 30. y has order 3 and xy has order 4. we conclude that H  S4 X Thus we have shown that PSL (2. S4 aKer ö  H.   168 1 a243 ˆ X 1 ‡ ‡ 0 ‡ 0 ‡ 0 ‡ 0 ˆ 8X 83 7 353 Hence. 7) has a subgroup which is isomorphic to S4 . all ®nite simple groups G possessing an involution t such that C G (t)  C. Now Ker ö.Applications to group theory We calculate the class algebra constant a243 . G contains elements x and y such that x has order 2.8 Theorem Let G be a ®nite non-abelian simple group which has an involution t such that C G (t)  D8 . S3 . we know that S4 ˆ ha. being a normal subgroup of S4 . by (30.20). b: a2 ˆ b3 ˆ (ab)4 ˆ 1iX Hence there is a homomorphism ö from S4 onto H (ö sends a to x and b to y). 30. Since H has an element of order 4.4. A4 or S4 (see Example 12. . From Example 30. an effort which was ®nally completed in the early 1980s (see the book by D. so H is isomorphic to S4 .6. This fact led Brauer to initiate a programme to ®nd. Let H be the subgroup kx. a dihedral group of order 8.19 states that there are only ®nitely many non-isomorphic ®nite simple groups containing an involution with a given centralizer.

30. contrary to assumption.10 Lemma Let G be a ®nite non-abelian simple group. Suppose Q is a subgroup of P with jP : Qj ˆ 2. This means that in its action on Ù. Proof Suppose u is not conjugate to an element of Q.4 of the book by J.2(4)). such a subgroup is call a Sylow p-subgroup of G. and A6 is a simple group of order 360 with this property (see Exercise 7 at the end of the chapter). we require a couple of preliminary results. Using some rather more sophisticated group theory than that covered in this book.1). Observe that jÙj ˆ 2jG : Pj ˆ 2m.9 Sylow's Theorem Let p be a prime number. and let P be a Sylow 2subgroup of G. Before embarking upon the proof of Theorem 30. g P G (see Example 29.354 Representations and characters of groups Observe that PSL(2. and de®ne an action of G on Ù by (Qx) g ˆ Qxg for x. one can show that PSL(2. b are positive integers and p T j b. (2) all Sylow p-subgroups are conjugate in G (i.8. Let Ù be the set of right cosets Qx of Q in G. 7) and A6 are the only simple groups of order 168 or 360. Then (1) G contains a subgroup of order p a. but refer you to Theorems 18. and let G be a ®nite group of order p a b. (3) if R is a subgroup of G with jRj ˆ p c for some c. Hence the subgroup f g P G : g acts as an even permutation on Ùg . then there exists g P G such that Q ˆ g Pg). where m is odd since P is a Sylow 2-subgroup. We shall not prove this. Fraleigh listed in the Bibliography. The ®rst is Sylow's Theorem. 30. then u is conjugate to an element of Q. the involution u is a product of m disjoint 2-cycles. 7) is a simple group of order 168 having an involution with centralizer D8 (see Lemma 27.e. À1 subgroups. then there is a Sylow p-subgroup of G containing R. Now consider fixÙ (u) ˆ fù P Ù : ùu ˆ ùg If Qx P fixÙ (u). hence is an odd permutation. if P Q are Sylow p. then Qxu ˆ Qx and hence xux À1 P Q.3 and 18. where a. a basic result in ®nite group theory. Hence fixÙ (u) ˆ Æ. If u is an involution in G.

so t P D.12)). and each n÷ P Z. Then Z(P) < C G (t) ˆ D. Therefore P < . but this need not be the case for a generalized character. Proof of Theorem 30. In particular. so Z(P) < Z(D) ˆ hti.8 Let G be a ®nite non-abelian simple group with an involution t such that C G (t) ˆ D  D8 . Certainly t commutes with itself. By Theorem 30. hø. This is impossible since G is non-abelian and simple. Notice also that the orthogonality relations give the usual inner products ˆ hø. we have t P Z(D). we de®ne the induced generalized character ø 4 G by ø 4 G ˆ (á 4 G) À ( â 4 G) where ø ˆ á À â as above. there is a Sylow 2-subgroup P of G such that D < P. the degree ø(1) can be 0 or negative for a generalized character ø. ÷i ˆ n÷ .Applications to group theory 355 is a normal subgroup of index 2 in G. and as t commutes with all elements of D.0 Finally. The centre of D8 is a cyclic group of order 2 (see (12.20 hold for generalized characters ø. If n÷ > 0 for all ÷ then of course ø is a character. and hence Z(D) ˆ hti. the centre of D. ⠈ À n÷ ÷X n÷ >0 n÷ . if H is a subgroup of a group G.19 and Corollary 21. where á and â are characters of H: take ˆ ˆ ሠn÷ ÷. By Lemma 26. This is simply a class function of the form ˆ øˆ n÷ ÷ x where the sum is over all the irreducible characters of H. j We also need to introduce the idea of a generalized character of a group H.9(3).1(1) we have Z(P) Tˆ 1. and hence Z(P) ˆ hti. This contradiction completes the proof. The generalized character ø can be expressed as a difference á À â. øi ˆ n2 ÷ for a generalized character ø as above. It is clear from this de®nition that the formulae for the values of ø 4 G given in Proposition 21.

every involution of G is conjugate to an involution in C. bi where a4 ˆ b2 ˆ 1 and bÀ1 ab ˆ aÀ1 .19 gives 1 ˆ • À1 (è 4 G)(c) ˆ è( y cy)X 8 yPG By (30. D is a Sylow 2-subgroup of G. (In particular. and de®ne è ˆ (1 C 4 D) À (ë 4 D). Let C ˆ hai be the cyclic subgroup of index 2 in D.) Hence hè. if y À1 cy P C then y P D. Also (è 4 G)(1) ˆ 0 (see Corollary 21. Write D ˆ ha. èi ˆ 3. We next establish (30X12) hè 4 G. Referring to the character table of D8 in Example 16. èi Now for 1 Tˆ c P C. observe ®rst that by Frobenius reciprocity. By Lemma 30. This is all the group theory we will need for the proof.356 Representations and characters of groups C G (t) ˆ D. We summarise what we have proved so far: (30.10. The rest is character theory. Since è vanishes on D À C. è(1) ˆ 0. èi ˆ hè. In other words.20). let g P G and suppose that g À1 cg P C for some non-identity element c P C.12) that . t G is the unique conjugacy class of involutions in G. It follows that (è 4 G)(c) ˆ è(c). And if y À1 cy P D À C then è( y À1 cy) ˆ 0. we must have g À1 tg ˆ t. Now hè 4 G. giving (30. hè 4 G. As t ˆ a2 is the only such involution. aÀ1 . we have è ˆ ÷1 ‡ ÷2 À ÷5 . 1 C i ˆ 1. and so it follows from (30. Then è takes the value 2 on a. èi ˆ 3. è 4 Gi ˆ h(è 4 G) 5 D.3(3).12). Since t ˆ c or c2. è 4 Gi ˆ 3X To see this. Then t ˆ a2 . Let ë be the linear character of C such that ë(a) ˆ i. 1 G i ˆ h1 C À ë.11) D is a Sylow 2-subgroup of G. hence g P C G (t) ˆ D and so g À1 Cg ˆ C. and if C ’ g À1 Cg ˆ C then g P D. and 0 elsewhere. we conclude that t G is the unique conjugacy class of involutions in G. whence y À1 cy ˆ cÆ1 and è( y À1 cy) ˆ è(c). and so P ˆ D. a generalized character of D. Next. we therefore have h(è 4 G) 5 D. the value 4 on t. for any g P G we have C ’ gÀ1 Cg ˆ f1g or C.11). Proposition 21.

14) we have 2 3 jGj á(t)2 â(t)2 hè 4 G. For g P G.14) We have ㈠jGj ˆ ÷(t)2 ÷. â are irreducible characters of G. 357 where á. á(t) and â(t) are integers. y P t G . ã 5 Ci ˆ 1 X4X((1 À i) ‡ 2 ‡ (1 ‡ i)) ˆ 4X jCj Hence from (30. where á.4 yields the following. By (30. hè 4 G. Write d ˆ á(1) and e ˆ á(t) P Z. from (30. 1 ‡ á(1) À â(1) ˆ 0 and 1 ‡ á(t) À â(t) ˆ 4. If we write t G ˆ C i and g lies in the conjugacy class C k of G. (30. then x À1 cx ˆ yx ˆ cÀ1 . First. then ã( g) ˆ a iik in the notation of (30. Therefore h1 C À ë. y) P t G 3 t G such that g ˆ xy. we have now proved the following.13) we have .Applications to group theory è 4 G ˆ 1 G ‡ á À â.10. ãi ˆ h1 C À ë. de®ne ã( g) to be number of ordered pairs (x.13) and (30. ãi ˆ 1‡ (30X15) À X 64 á(1) â(1) On the other hand. Consider ã(c) for 1 Tˆ c P C. and hence x P D by (30. Since we have shown that (è 4 G)(t) ˆ è(t) ˆ 4. Now calculation in D8 shows that ã(c) ˆ 4. ã 5 Ci. Note that by Corollary 13. by Frobenius Reciprocity. jDj2 ÷ ÷(1) where the sum is over all irreducible characters ÷ of G. We now introduce another class function of G into the picture. similarly y P D.15) we deduce 2 3 á(t)2 â(t)2 jGj 1 ‡ (30X16) À ˆ 28 X á(1) â(1) This equation gives us enough number-theoretic information about jGj to ®nish the proof fairly quickly.11). Hence Theorem 30. If c ˆ xy with x.13) We have è 4 G ˆ 1 G ‡ á À â. We shall calculate the inner product of ã and è 4 G in two ways. (30.3). â are irreducible.

â(t) ˆ e À 3X From the column orthogonality relations 16. suppose that e ˆ 2. The class algebra constants aijk are given by ˆ Ci C j ˆ aijk C k X k They can be calculated from the character table. a Sylow 2-subgroup of G has order 8.8. giving jGj ˆ 360. This completes the proof of Theorem 30. we deduce that d ‡ 2 ˆ 23 . and hcf (d À 1. Hence (d À 1)2 must divide 210 . . d d‡1 whence jGj ˆ 28 d(d ‡ 1) X (d À 1)2 Now the highest common factor hcf (d À 1. Finally. j Summary of Chapter 30 1. Then (30. Suppose now that e ˆ 1. by using the formula aijk ˆ ˆ ÷( g i )÷( g j )÷( g k ) jGj X jCG ( g i )j jCG ( g j )j ÷ ÷(1) 2.16) gives   1 4 jGj 1 ‡ À ˆ 28 . Given groups G and H.4(2). It follows that r ˆ 3 and d ˆ 9. the class algebra constants of G can sometimes be used to determine whether or not G has a subgroup which isomorphic to H.8.358 Representations and characters of groups â(1) ˆ d ‡ 1.16) yields jGj ˆ 28 d(d ‡ 1) X (d ‡ 2)2 Reasoning as above. d ‡ 1) is 1 or 2. one of the possibilities in the conclusion of Theorem 30. we have 8 ˆ jC G (t)j > 1 ‡ á(t)2 ‡ â(t)2 ˆ 1 ‡ e 2 ‡ (e À 3)2 . giving d ˆ 6 and jGj ˆ 168. Moreover. d) ˆ 1. and so d À 1 ˆ 2 r with r < 5. Then (30. from which it follows that e ˆ 1 or 2.

b: a2 ˆ b2 ˆ 1. (ab)7 ˆ 1l. g4 1 2 À1 À1 À1 0 1 g5 1 À1 2 À1 À1 0 1 g6 1 0 0 á â À1 0 g7 1 0 0 â á À1 0 where á ˆ (1 ‡ 5)a2. 11) contain a subgroup which is isomorphic to A5 ? 4.Applications to group theory 359 3.6. Exercises for Chapter 30 1. ⠈ (1 À 5. 7) contains elements a and b such that a has order 2. g1 ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 1 5 5 8 8 9 10 p g2 1 1 1 0 0 1 À2 g3 1 À1 À1 0 0 1 0 p 5)a2. and that A5 has the following presentation: A5 ˆ ha. and that G has the character table shown. given at the end of Chapter 27. 11) is given in the solution to Exercise 27. 2. 7) contain a subgroup isomorphic to D14 ? (Hint: D14 ˆ ka. 7). Does PSL (2.13). (b) Use the Frobenius±Schur Count of Involutions to obtain an . Using Sylow's Theorem. you may assume that A5 is a simple group. to prove that PSL (2. then G  A5 . Suppose that G is a group. (a) Show that G is a simple group of order 360. if G is a group with the same character table as A5 (see Example 20. together with lots of ingenious character theory. Prove that A5 is characterized by its character table ± that is. b has order 3 and ab has order 7.) For the next three exercises. The character table of PSL (2. b: a2 ˆ b3 ˆ (ab)5 ˆ 1iX 3. Use the character table of PSL (2. it can be shown that any simple group possessing an involution with centralizer isomorphic to D8 must have order 168 or 360. Does PSL (2.

360 Representations and characters of groups upper bound for the number of involutions in G.9. p). (c) Prove that G has a subgroup H which is isomorphic to A5 . (d) Using Exercise 23. both of which contain an involution with centralizer isomorphic to D8 . (Hint: look for a suitable simple group PSL(2. 8. 7. 6. Prove that PSL(2. and deduce that g2 has order 2 and g3 has order 4. 7) and A6 are simple groups of order 168. 360 respectively.6(3) to show that every group G which is generated by two elements a and b which satisfy a2 ˆ b3 ˆ (ab)4 ˆ 1 has order at most 24. Find a simple group G having an involution C G (t)  D16 . Use the ®gure which appears in Example 30.) t such that . show that G  A6 .

then no group of order pa q b is simple. it is called the minimal polynomial of á. In the ®rst edition of his book Theory of groups of ®nite order (1897). until H. Bender found one in 1972. which we now describe. We omit proofs of these ± for a good account.2) which is concerned with character values. An algebraic number is a complex number which is a root of some non-zero polynomial over Q. but it was only after studying Frobenius's new theory of group representations that he was able to prove the theorem in general. Let á be an algebraic number. if ù is an nth root of unity then the minimal poly361 . For example. which states that if p and q are prime numbers and a and b are positive integers. A preliminary lemma We prepare for the proof of Burnside's Theorem with a lemma (31. many later attempts to ®nd a proof which does not use representation theory were unsuccessful. see for instance the book by Pollard and Diamond listed in the Bibliography.31 Burnside's Theorem One of the most famous applications of representation theory is Burnside's Theorem. In order to establish this lemma we require some basic facts about algebraic integers and algebraic numbers. and let p(x) be a monic polynomial over Q of smallest possible degree having á as a root. b. The roots of p(x) are called the conjugates of á. Then p(x) is unique and irreducible. We call a polynomial in x monic if the coef®cient of the highest power of x in it is 1. Burnside presented group-theoretic arguments which proved the theorem for many special choices of the integers a. Indeed.

. By Proposition 13.362 Representations and characters of groups nomial of ù divides x n À 1. ‡ |ù d | ˆ d. .1) can be proved easily using some Galois theory. where á9 is a conjugate of á and â9 is a conjugate of â. . and if 0 . where each ù i is a root of unity. Hence each conjugate of ã has 1 d . Then j÷( g)a÷(1)j < 1.1). We prove that ÷( g) ˆ 0. . 1. (31. Moreover. each conjugate of ã is of the form (ù9 ‡ X X X ‡ ù9 )ad 1 d where ù9 . Then every conjugate of á ‡ â is of the form á9 ‡ â9. j÷( g)a÷(1)j . ‡ ù d | < |ù1 | ‡ . For an elementary proof of this. so ÷( g)a÷(1) ˆ (ù1 ‡ X X X ‡ ù d )adX Since |÷( g)| ˆ |ù1 ‡ . .9 we have ÷( g) ˆ ù1 ‡ . . . . We shall require the following fact about conjugates: (31. Alternatively. ù9 are roots of unity. If á is an algebraic integer. Write 㠈 ÷( g)a÷(1). Now suppose that ÷( g)a÷(1) is an algebraic integer and j÷( g)a÷(1)j .1) Let á and â be algebraic numbers. Section 3. Chapter V. where á9 is a conjugate of á. and let p(x) be the minimal polynomial of ã. and let g P G. and so every conjugate of ù is also an nth root of unity. then á is a root of a monic polynomial with integer coef®cients (see Chapter 22). 31.2 Lemma Let ÷ be a character of a ®nite group G. . so that p(x) ˆ x n ‡ a nÀ1 x nÀ1 ‡ X X X ‡ a1 x ‡ a0 where ai P Z for all i. it follows that j÷( g)a÷(1)j < 1. if r P Q then every conjugate of rá is of the form rá9. 1 then ÷( g)a÷(1) is not an algebraic integer. . By (31. ‡ ù d . see Pollard and Diamond. Proof Let ÷(1) ˆ d. and it turns out that the minimal polynomial of á also has integer coef®cients.

Burnside's Theorem

363

modulus at most 1. It follows that if ë is the product of all the conjugates of ã (including ã), then jëj , 1X But the conjugates of ã are, by de®nition, the roots of the polynomial p(x), and the product of all these roots is equal to Æa0 . Thus ë ˆ Æa0 X Since a0 P Z and |ë| , 1, it follows that a0 ˆ 0. As p(x) is irreducible, this implies that p(x) ˆ x, which in turn forces 㠈 0. Thus ÷( g) ˆ 0, and the proof is complete.
j

Burnside's p a q b Theorem We deduce the main result, Theorem 31.4, from another interesting theorem of Burnside. 31.3 Theorem Let p be a prime number and let r be an integer with r > 1. Suppose that G is a ®nite group with a conjugacy class of size pr . Then G is not simple. Proof Let g P G with | gG | ˆ pr . Since pr . 1, G is not abelian and g Tˆ 1. As usual, denote the irreducible characters of G by ÷1 , . . . , ÷ k , and take ÷1 to be the trivial character. The column orthogonality relations, Theorem 16.4(2), applied to the columns corresponding to 1 and g in the character table of G, give 1‡ Therefore
k ˆ iˆ2 k ˆ iˆ2

÷ i ( g)÷ i (1) ˆ 0X

÷ i ( g) .

÷ i (1) 1 ˆÀ X p p

Now À1a p is not an algebraic integer, by Proposition 22.5. Therefore, for some i > 2, ÷ i ( g)÷ i (1)a p is not an algebraic integer (see Theorem 22.3). Since ÷ i ( g) is an algebraic integer (Corollary 22.4), it follows

364

Representations and characters of groups

that ÷ i (1)a p is not an algebraic integer; in other words, p does not divide ÷ i (1). Thus ÷ i ( g) Tˆ 0 and p T j ÷ i (1)X As | gG | ˆ pr , this means that ÷ i (1) and | gG | are coprime integers, and so there are integers a and b such that ajG:CG ( g)j ‡ b÷ i (1) ˆ 1X Hence a jGj÷ i ( g) ÷ i ( g) ‡ b÷ i ( g) ˆ X jCG ( g)j÷ i (1) ÷ i (1)

By Corollaries 22.10 and 22.4, the left-hand side of this equation is an algebraic integer; and since ÷ i ( g) Tˆ 0, it is non-zero. Now Lemma 31.2 implies that j÷ i ( g)a÷ i (1)j ˆ 1X Let r be a representation of G with character ÷ i . By Theorem 13.11(1), there exists ë P C such that gr ˆ ëIX Let K ˆ Ker r, so that K is a normal subgroup of G. Since ÷ i is not the trivial character, K Tˆ G. If K Tˆ {1} then G is not simple, as required; so assume that K ˆ {1}, that is, r is a faithful representation. Since gr is a scalar multiple of the identity, gr commutes with hr for all h P G. As r is faithful, it follows that g commutes with all h P G; in other words, g P Z(G)X Therefore Z(G) Tˆ {1}. As Z(G) is a normal subgroup of G and Z(G) Tˆ G, we conclude that G is not simple. j We now come to the main result of the chapter, Burnside's Theorem. 31.4 Burnside's paqb Theorem Let p and q be prime numbers, and let a and b be non-negative integers with a ‡ b > 2. If G is a group of order pa q b , then G is not simple. Proof First suppose that either a ˆ 0 or b ˆ 0. Then the order of G is a power of a prime, so by Lemma 26.1(1) we have Z(G) Tˆ {1}.

Burnside's Theorem

365

Choose g P Z(G) of prime order. Then k gl v G and k gl is not equal to {1} or G. Hence G is not simple. Now assume that a . 0 and b . 0. By Sylow's Theorem 30.9, G has a subgroup Q of order q b . We have Z(Q) Tˆ {1} by Lemma 26.1(1). Let g P Z(Q) with g Tˆ 1. Then Q < CG ( g), so j g G j ˆ jG:CG ( g)j ˆ pr for some r. If pr ˆ 1 then g P Z(G), so Z(G) Tˆ {1} and G is not simple as before. And if pr . 1 then G is not simple, by Theorem 31.3. j In fact Burnside's pa q b Theorem leads to a somewhat more informative result about groups of order pa q b : (31.5) Every group of order p a q b is soluble. Here, by a soluble group we mean a group G which has subgroups G0 , G1 , . . . , Gr with 1 ˆ G 0 , G1 , X X X , G r ˆ G such that for 1 < i < r, GiÀ1 v Gi and the factor group Gi aG iÀ1 is cyclic of prime order. We sketch a proof of (31.5), using induction on a ‡ b. The result is clear if a ‡ b < 1, so assume that a ‡ b > 2 and let G be a group of order pa q b . By Burnside's Theorem 31.4, G has a normal subgroup H such that H is not {1} or G. Both H and the factor group Ga H have order equal to a product of powers of p and q, and these orders are less than pa q b . Hence by induction, H and Ga H are both soluble. Therefore there are subgroups 1 ˆ G0 v G1 v X X X v Gs ˆ H, 1 ˆ Gs a H v G s‡1 a H v X X X v Gr a H ˆ Ga H with all factor groups Gi aG iÀ1 of prime order. Then the series 1 ˆ G0 v G 1 v X X X v G r ˆ G shows that G is soluble. Summary of Chapter 31 1. If G has a conjugacy class of size pr ( p prime, r > 1), then G is not simple.

366

Representations and characters of groups

2. If |G| ˆ pa q b ( p, q primes, a ‡ b > 2), then G is not simple. Exercises for Chapter 31 1. Show that a non-abelian simple group cannot have an abelian subgroup of prime power index. 2. Prove that if G is a non-abelian simple group of order less than 80, then |G| ˆ 60. (Hint: use Exercise 13.8.)

32 An application of representation theory to molecular vibration

Representation theory is used extensively in many of the physical sciences. Such applications come about because every physical system has a symmetry group G, and certain vector spaces associated with the system turn out to be RG-modules. For example, the vibration of a molecule is governed by various differential equations, and the symmetry group of the molecule acts on the space of solutions of these equations. It is on this application ± the theory of molecular vibrations ± that we concentrate in this ®nal chapter. In order to keep our treatment elementary, we stay within the framework of classical mechanics throughout. (Quantum mechanical effects can be incorporated subsequently, but we shall not go into this ± for an account, consult the book by D. S. Schonland listed in the Bibliography.)

Symmetry groups Let V be R2 or R3, and for v, w P V, let d(v, w) denote the distance between v and w ± in other words, if v ˆ (x1 , x2 , . . .) and w ˆ ( y1, y2, . . .), then 3 r2ˆ 2 d(v, w) ˆ (xi À yi ) X An isometry of V is an invertible endomorphism W of V such that d(vW, wW) ˆ d(v, w) for all v, w P V X The set of all isometries of V forms a group under composition, called the orthogonal group of V, and denoted by O(V). Any rotation of R3 about an axis through the origin is an example 367

368

Representations and characters of groups

of an isometry; so is any re¯ection in a plane through the origin. The endomorphism À1R3 which sends every vector v to Àv is another example of an element in the orthogonal group O(R3 ). It turns out that the composition of two rotations is again a rotation, and that for every isometry g in O(R3 ), either g or À g is a rotation (see Exercise 32.1). The orthogonal group O(R3 ) therefore contains a subgroup of index 2 which consists of the rotations. The same is true of the group O(R2 ). If Ä is a subset of V , where V ˆ R2 or R3, then we de®ne G(Ä) to be the set of isometries which leave Ä invariant ± that is, G(Ä) ˆ f g P O(V ): Ä g ˆ Äg (where Ä g ˆ {v g: v P Ä}). Then G(Ä) is a subgroup of called the symmetry group of Ä. The subgroup of G(Ä) the rotations in G(Ä) is called the rotation group of Ä. the rotation group of Ä in the symmetry group G(Ä) is 1 O(V ), and is consisting of The index of or 2.

32.1 Example Let V ˆ R2, and let Ä be a regular n-sided polygon, with n > 3, centred at the origin. The symmetry group of Ä is easily seen to be the dihedral group D2n , which was de®ned as a group of n rotations and n re¯ections preserving Ä (see Example 1.1(3)). Now let V ˆ R3, and again let Ä be a regular n-sided polygon (n > 3) centred at the origin. This time, G(Ä) ˆ D2 n 3 C2 ; the extra elements arise because there is an isometry which ®xes all points of Ä, namely the re¯ection in the plane of Ä. 32.2 Example Let Ä be a regular tetrahedron in R3 centred at the origin:

An application of representation theory to molecular vibration 369 Label the corners of the tetrahedron 1, 2, 3, 4. We claim that each permutation of the numbers 1, 2, 3, 4 corresponds to an isometry of Ä. For example, the 2-cycle (1 2) corresponds to a re¯ection in the plane which contains the origin and the edge 34; similarly each 2-cycle corresponds to a re¯ection. Since S4 is generated by the 2-cycles, each of the 24 permutations of 1, 2, 3, 4 corresponds to an isometry, as claimed. No non-identity endomorphism of R3 ®xes all the corners of Ä, since Ä contains three linearly independent vectors. Therefore we have found all the isometries, and G(Ä)  S4 . Notice that the rotation group of Ä is isomorphic to A4 ; for example, (1 2)(3 4) corresponds to a rotation through ð about the axis through the mid-points of the edges 12 and 34, and (1 2 3) corresponds to a rotation through 2ða3 about the axis through the origin and the corner 4. Finally, observe that the group G(Ä) is unchanged if we take Ä to consist of just the four corners of the tetrahedron. 32.3 Example In this example we describe the symmetry groups of the molecules H2 O (water), CH3 Cl (methyl chloride) and CH4 (methane). The symmetry group of a molecule is de®ned to be the group of isometries which not only preserve the position of the molecule in space, but also send each atom to an atom of the same kind. The shapes of the three molecules are as follows.

370

Representations and characters of groups

We always assume that the centroid of our molecule lies at the origin in R3 . The CH4 molecule has four hydrogen atoms at the corners of a regular tetrahedron, and a carbon atom at the centre of the tetrahedron. So the symmetry group of the molecule CH4 is equal to the symmetry group of the tetrahedron, as given in Example 32.2. This group is isomorphic to S4 , permuting the four hydrogen atoms among themselves and ®xing the carbon atom. As for the CH3 Cl molecule, this possesses a rotation symmetry a of order 3 about the vertical axis, and three re¯ection symmetries in the planes containing the C, Cl and one of the H atoms. If b is one of these re¯ections, then the symmetry group is f1, a, a2 , b, ab, a2 bg and is isomorphic to S3 , permuting the three H atoms and ®xing the C and Cl atoms. Finally, the H2 O molecule possesses two re¯ection symmetries, one in the plane of the molecule, and one in a plane perpendicular to this one passing through the O atom; and it has a rotation symmetry of order 2. Hence the symmetry group is isomorphic to C2 3 C2 . Vibration of a physical system We prepare for a description of the general problem with an example. 32.4 Example Suppose we have a spring stretched between two points P and Q on a smooth horizontal table, with equal masses m attached at the points of trisection of the spring:

The masses are displaced slightly, and released. What can we say about the subsequent motion of the system? To investigate this problem, we let x1 and x2 be the displacements of the two masses at time t. We measure x1 from left to right, and x2 from right to left, as indicated in the ®gure above. Let k be the

An application of representation theory to molecular vibration 371 stiffness of the spring ± in other words, if the extension in the spring is x, then the restoring force is kx. The spring pulls the left-hand mass towards P with force kx1 and towards Q with force Àk(x1 ‡ x2 ). By dealing with the right-hand mass similarly, we obtain the following equations of motion of the system: m1 ˆ Àkx1 À k(x1 ‡ x2 ) ˆ À2kx1 À kx2 , x m2 ˆ Àkx2 À k(x1 ‡ x2 ) ˆ Àkx1 À 2kx2 , x  where xi denotes the second derivative of xi with respect to t. These are second order linear differential equations in two unknowns x1 and x2 , so the general solution involves four arbitrary constants. We shall ®nd the general solution, using a method which can be applied in a much wider context.  Write x ˆ (x1 , x2 ), x ˆ (1 , x2 ) and q ˆ kam. Then the equations of x  motion are equivalent to the matrix equation   À2q Àq  x ˆ xA, where A ˆ (32X5) X Àq À2q Notice that A is symmetric. Hence the eigenvalues of A are real, and A has two linearly independent eigenvectors. It is this property which we wish to emphasize and exploit in the present example. Before we explicitly ®nd the eigenvectors of A, let us explain why they allow us to solve the equation of motion (32.5). Suppose that u is an eigenvector of A, with eigenvalue Àù2 . For an arbitrary constant â, let x ˆ sin (ùt ‡ â) uX Then  x ˆ Àù2 sin (ùt ‡ â) u ˆ sin (ùt ‡ â) uA ˆ xAX Thus x is a solution of the equation of motion. If u1 and u2 are linearly independent eigenvectors of A, with eigenvalues Àù2 and 1 Àù2 , respectively, then 2 á1 sin (ù1 t ‡ â1 ) u1 ‡ á2 sin (ù2 t ‡ â2 ) u2 (since uA ˆ Àù2 u)

À1). We assume that the internal forces are linear functions of the displacements. we obtain equations which may be expressed in the form (32X6) (Compare (32. t ‡ â2 ) (1. 1) and (1. â1 . x1 ˆ Àx2 ˆ sin ( q . À1)X The solutions which involve just one eigenvector of A are called the normal modes of vibration.5) is p p á1 sin ( (3q) t ‡ â1 ) (1. at each atom the three coordinate axes . â2 . which we use to measure the displacement of the atom. so it is the general solution. It follows that when we apply Newton's Second Law of Motion. À1) The general molecular vibration problem Suppose we have a molecule consisting of n atoms which vibrate under internal forces. á2 .) Here x is the displacements of all the atoms. Thus. 1) ‡ á2 sin ( q . t ‡ â2 ) (1. with corresponding eigenvectors (1. We now adopt this line of attack in the problem to hand. 1) Mode 1: p Here. x1 ˆ x2 ˆ sin ( (3q) t ‡ â1 ) and the vibration is p Here. for the moment. At the equilibrium position of each atom. p sin ( (3q) t ‡ â1 ) (1. we assign three coordinate axes. the state of the molecule at a given time is described by a vector in the 3n-dimensional vector space R3 n .372 Representations and characters of groups is a solution of the equation of motion which involves four arbitrary constants á1 . entries which are determined by Assume.5). Therefore the general solution of the equation of motion (32. For the matrix given in (32. the eigenvalues are À3q and Àq.5). t ‡ â2 ) and the vibration is Mode 2: p sin ( q . They are as follows. that  x ˆ xAX row vector in R3 n which measures the and A is a 3n 3 3n matrix with real the internal forces.

7 Proposition All the eigenvalues of A are real.8 De®nition A normal mode of vibration for our molecule is a vector in R3 n of one of the following forms: (1) sin (ùt ‡ â) u ( â constant) where Àù2 is a non-zero eigenvalue of A and u is a corresponding eigenvector. which we de®ne next. the effect of changing coordinate axes is merely to replace A by a matrix which is conjugate to A. and A has 3n linearly independent eigenvectors.9 Proposition Each normal mode of vibration is a solution of the equation of motion (32. The general solution of the equation of motion is a linear combination of the normal modes of vibration. and for this solution all the atoms vibrate with the same frequency.6). (2) (t ‡ â) u ( â constant) where u is an eigenvector of A corresponding to the eigenvalue 0. It can be shown. then  x ˆ Àù2 sin (ùt ‡ â) u ˆ sin (ùt ‡ â) uA ˆ xAX If uA ˆ 0 and x ˆ (t ‡ â)u. from physical considerations. To solve the equation of motion (32. Now. A has real eigenvalues. where our chosen coordinate axes are not necessarily at right angles to each other. then  x ˆ 0 ˆ (t ‡ â)uA ˆ xAX This proves that the normal modes of vibration are solutions of the . 32. for the general case. Proof If uA ˆ Àù2 A and x ˆ sin (ùt ‡ â) u.6). 32. we look for normal modes of the system. and A has 3n linearly independent eigenvectors. 32.An application of representation theory to molecular vibration 373 which we have chosen are at right angles to each other. In particular. that in this special case the matrix A is symmetric. Therefore we have the following proposition.

9 reduces the problem of solving the equations of motion to that of ®nding all the eigenvalues and eigenvectors of the 3n 3 3n matrix A. and we shall describe a method for doing this. Since each normal mode involves an arbitrary constant. negates v2 and v3. R3 n is an RG-module. By construction.6) consists of second order differential equations in 3n unknowns). by Proposition 32. However. Thus.6). Then g ®xes v1 . Use of the symmetry group We continue the discussion of the previous section. Since G permutes the atoms among themselves. j Proposition 32. and . ù or 0) in a normal mode. and for 1 < i < 9. Let G be the symmetry group of the molecule in question. this can be a huge and unwieldy task if it is attempted directly ± even writing down the matrix A for a given molecule can be a painful operation! The symmetry group of the molecule and its representation theory can often be used to simplify greatly the calculation of the eigenvectors of A. with eigenvector u. 32. let v i denote a unit vector along coordinate axis i. Therefore there exist 3n linearly independent normal modes. the general linear combination of normal modes involves 6n arbitrary constants. which is nonsense. then x ˆ e ë t u would be a solution to the equation of motion. each element of G acts as an endomorphism of the space R3 n of displacement vectors. interchanges v4 and v7. for if ë were p such an eigenvalue.10 Example Let g be the rotation of order 2 of the H2 O molecule: Assign coordinate axes at the initial positions of each atom as shown. all the atoms vibrate with the same frequency (namely.374 Representations and characters of groups equation of motion (32.6) (as (32.7. so it is the general solution to the equation of motion (32. Note that A can have no strictly positive eigenvalue.

7). Àx3 . vA ˆ ëv for some ë P R. A similar argument applies to the eigenspace for the eigenvalue 0. and the eigenspaces of A are RG-submodules of R3 n . and let v be a corresponding eigenvector. x3 . (xg)A ˆ (xA) g. Choose a basis of R3 n which consists of eigenvectors of A (see Proposition 32. Àx5 .11 Proposition For all g P G and x P R3 n . This shows that the eigenspace for Àù2 is an RG-submodule of R3 n . v g must also specify the directions and relative magnitudes of displacements in a normal mode of frequency ù. since the relative con®guration of the atoms is unaltered by applying g. x4 . In effect. j The idea now is to use representation theory to express the RGmodule R3n as a direct sum of irreducible RG-submodules. Àx8 . v g is an eigenvector of A. x7 . with eigenvalue Àù2 . x7 . The eigenspace for the eigenvalue ë of A is. x9 ) g ˆ (x1 . x2 . Proof Let Àù2 be a non-zero eigenvalue of A. Àx6 )X  We return to the general set-up. fx P R3 n : xA ˆ ëxgX We can now present the crucial proposition which allows us to exploit the symmetry group G of our molecule. Therefore. x6 . The equations of motion are x ˆ xA. and let g P G. and (v g)A ˆ ë(v g) ˆ (ëv) g ˆ (vA) gX Hence (xg)A ˆ (xA) g for all x P R3 n . it tells us that the function x 3 xA (x P R3 n ) is an RG-homomorphism from R3 n to itself.An application of representation theory to molecular vibration 375 interchanges v5 and v6 with the negatives of v8 and v9 . and we are trying to ®nd the eigenvectors of A. x5 . Àx2 . 32. x8 . x4 . Àx9 . For all vectors v in the basis. Then v speci®es the directions and relative magnitudes of the displacements of the atoms from the equilibrium position when the molecule is vibrating in a normal mode of frequency ù. Therefore g acts on R9 as follows: (x1 . For all g in G. by de®nition. and hence .

Therefore. The problem of ®nding the eigenspaces of A is considerably simpli®ed as a consequence of our next proposition. The function å: v ‡ w 3 w (v P V÷ .11. We can use character theory to see which irreducible RG-modules are contained in R3 n .) j . We call V÷ a homogeneous component of R3 n . the function x 3 (xA)å is an RG-homomorphism from V÷ into W. (This procedure sometimes needs to be modi®ed. w P W ) (x P V÷ ) is an RG-homomorphism. since the character of the RG-module R3 n might contain an irreducible character which cannot be realized over R ± but in practice. so xA P V÷ for all x P V÷ . then the element ˆ ÷( g À1 ) g gPG sends R3 n onto the sum of those irreducible RG-submodules of R3 n which have character ÷ (see (14. and the normal modes of the molecule. problems like this are uncommon. By Proposition 11.13 Proposition Each homogeneous component V÷ of R3 n is A-invariant ± that is. xA P V÷ for all x P V÷ X Proof By Maschke's Theorem we may write R3n ˆ V÷ È W for some RG-module W.3 is stated in terms of CG-modules.27)). Let V÷ denote the sum of those irreducible RG-submodules of R3 n which have character ÷. 32. by Proposition 32.376 Representations and characters of groups to determine the eigenspaces of A.12 De®nition Suppose that ÷ is the character of an irreducible RG-module. (Although Proposition 11.8. and no RG-submodule of W has character ÷.) 32.3. this function is zero. its proof works equally well for RG-modules ± compare Exercise 23. and if ÷ is the character of an irreducible RGmodule which occurs.

then all the non-zero vectors in V÷ are eigenvectors of A. (3) Calculate the character ÷ of the RG-module R3 n and express ÷ as a linear combination of the irreducible characters of G.13). then see Remark 32.) Since V÷ is A-invariant. 32. each homogeneous component V÷ i of R3 n . with eigenvalue ë. j We now summarize the steps in the procedure for ®nding the normal modes of vibration of a given molecule. (2) Calculate the symmetry group G of the molecule. Then R3 n is an RG-module.15 Summary (1) Assign three coordinate axes at each of the n atoms of the molecule.7. or by some other method. This programme can often be successfully completed. This involves no extra work if V÷ i is irreducible (see Corollary 32. we may choose v P V÷ such that v is an eigenvector of A. in turn. then sin (ùt ‡ â) v (or (t ‡ â)v if ù ˆ 0) is a normal mode. If V÷ i is reducible.14). say.19 below. Proof (Compare the proof of Schur's Lemma. with eigenvalue Àù2 . (6) If v is an eigenvector of A. . to make further progress. This can € À1 3n be done by applying the element for each gPG ÷ i ( g ) g to R irreducible character ÷ i of G which appears in ÷. (5) Consider. where â is an arbitrary constant. as we shall illustrate in the examples which make up the rest of this chapter.14 Corollary If V÷ is an irreducible RG-module. (4) Express R3 n as a direct sum of homogeneous components. or Exercise 32. so it must equal V÷ . to obtain R3n . Then the intersection of V÷ with the eigenspace for ë is a non-zero RG-submodule of V÷ . and ®nd the eigenvectors of A in V÷ i (see Proposition 32.An application of representation theory to molecular vibration 377 32. It is usually necessary to know the equations of motion in order to determine the frequency ù.

The symmetry group (in two dimensions) of the molecule is the dihedral group D6. 1). where â1 .17 Example Consider a hypothetical triatomic molecule. 32. The displacement vectors (x1 . sin (ù2 t ‡ â2 )(1. . . where xi is the displacement along axis i (1 < i < 6).16 Example We ®rst return to Example 32.4. with the spring and two vibrating masses: The symmetry group of this system is G ˆ h g: g 2 ˆ 1i. where g is the re¯ection in the mid-point of PQ. ù2 are the frequencies. À1)X It follows that the normal modes of the system are given by sin (ù1 t ‡ â1 )(1. We choose to take our axes along the edges of the triangle. generated by a rotation a of order 3 and a re¯ection . x6 ) in R6 . Notice that we have determined the normal modes of vibration (but not their frequencies) using the symmetry group alone. where u1 ˆ (1.4. Since (x1 . where the three identical atoms are at the corners of an equilateral triangle. .378 Representations and characters of groups 32. x2 ) form an RG-module R2 . as this eases the calculations: Thus the position of the molecule is given by a vector (x1 . This agrees with the conclusion of Example 32. x2 ) g ˆ (x2 . u2 ˆ (1. À1). so we assign two displacement coordinates to each atom. â2 are constants and ù1 . we consider only vibrations of the molecule in the plane. the RG-submodules of R2 are sp (u1 ) and sp (u2 ). . 1). For simplicity. as shown. x1 ).

It is easy to work out the action of each element of D6 on R6 . These include the rotation and translation modes. x4 . v3 are 2-dimensional displacement vectors for the three atoms. v3 ) P R6 pictorially by the diagram We ®rst calculate the normal modes of the form (t ‡ â)v. then (x1 . x4 . x3 )X We want to express the RD6 -module R6 as a direct sum of irreducible RD6 -modules. ÷2 . if b is the re¯ection which ®xes the top atom. we seek to express R6 as a direct sum of RD6 -submodules with characters ÷1 . x2 .An application of representation theory to molecular vibration 379 b (see Example 32. x3 . which occur for every molecule. And from the action of b given above. we see that ÷(b) ˆ 0. ÷(a) ˆ 0. v2 . v2 . the character table of D6 is 1 ÷1 ÷2 ÷3 1 1 2 a 1 1 À1 b 1 À1 0 Hence ÷ ˆ ÷1 ‡ ÷2 ‡ 2÷3 .1). x5 . if v1 . As a matter of notation. x1 . we ®rst calculate the character ÷ of the module. x6 )b ˆ (x2 . then we represent the displacement vector (v1 . Thus the values of ÷ are 1 ÷ 6 a 0 b 0 By Section 18. For example. ÷3 and ÷3 .3. corresponding to the eigenspace of A for eigenvalue 0. x6 . Thus. Since the rotation a does not ®x any of the atoms. x5 . . To do this.

so the character must be ÷3 . v2 and v3. the molecule rotates with constant angular velocity about the centre. ÷ R (b) ˆ À1. 1. v2 ˆ (1. À1. 0.380 Representations and characters of groups Rotation mode In this mode. The mode is given by (t ‡ â)v. 1. the subspace sp (v1 . À1). v3 ) has dimension 2. 1. À1. sp (v) ˆ R6 å2 . where v is a vector in the span of v1 . We call sp (v) the rotation submodule of R6 . and so ÷ R ˆ ÷2 . The modes are of the form (t ‡ â)v. pictorially. where ˆ å2 ˆ ÷2 ( g À1 ) g ˆ 1 ‡ a ‡ a2 À b À ab À a2 b gP D6 (compare (14. v3 ˆ (0. If ÷ R is the character of sp (v). Thus the character of the translation submodule is part of ÷ À ÷2 ˆ ÷1 ‡ 2÷3 . It is clearly an RD6 -submodule of R6 . and is called the translation submodule. where v ˆ (1. 0. 0. Translation modes These are modes in which all atoms move in the same constant direction with the same constant speed. 0. 1)). . it does not contain the rotation submodule. À1). À1. these vectors being given pictorially by (thus v1 ˆ (À1. À1. Indeed. Since v1 ‡ v2 ‡ v3 ˆ 0. 1. 0). 1.27)). v2 . À1. 1. ÷ R (a) ˆ 1. then ÷ R (1) ˆ 1. À1.

u1 À u3 ) is an RD6 -submodule of R6 . with character ÷vib . so that total moment of each vector in R6 about the vib centre is zero. u2 . every vector in R6 which satis®es these equations lies in R6 . if w P R6 then the total component of w vib in each direction is zero. R6 does not contain the rotation vib submodule. Our calculation of the normal modes is now complete.18(3)) below). . it is easy to see that sp (u1 À u2 . Therefore a basis of R6 is u1 . moreover. and we summarize our ®ndings below.11). Finally. The vibratory mode given by u1 ‡ u2 ‡ u3 is sometimes called the expansion±contraction mode (you will see the reason for this name in the picture in (32. The sum of these eigenspaces forms an RD6 -submodule R6 of R6 (by Proposition 32.An application of representation theory to molecular vibration 381 Vibratory modes The remaining normal modes correspond to eigenspaces of the matrix A with non-zero eigenvalues. These constraints imply three independent linear equations in the coordinates of the vectors in R6 . u3 . where vib ÷vib ˆ ÷ À (÷2 ‡ ÷3 ) ˆ ÷1 ‡ ÷3 X In particular. u2 . and are called vibratory modes. u3 among themselves. and since R6 has vib vib dimension 3. since D6 permutes the vectors u1 . Since no mode in R6 can have vib vib any translation component. Its character is ÷3 and it gives us the last eigenspace for the vib matrix A. R6 has dimension 3. where vib vib Clearly sp (u1 ‡ u2 ‡ u3 ) is an RD6 -submodule of R6 with characvib ter ÷1 .

382 (32. In order to .) We emphasize that we have found the normal modes of vibration without explicit knowledge of the equations of motion.18) Representations and characters of groups (1) Rotation mode: (2) Translation modes: linear combinations of (3) Vibratory mode: expansion±contraction mode (4) Vibratory modes: linear combinations of (We have chosen 2u2 À u1 À u3 . u1 À u3 pictorially. 2u1 À u2 À u3 as the basis for the vibratory modes in (4) merely because these modes look simpler than u1 À u2 .

the difference in length between QR and Q9R9 is (x4 ‡ x5 ) ‡ 1(x3 ‡ x6 )X 2 (We always assume that x1 . . For a general displacement (x1 .) Similarly. R9. x5 . x6 are small compared with the distance between the atoms. x6 ). Let m be the mass of each atom. X X X . m  x1 ˆ À(x1 ‡ x6 ) À 1(x2 ‡ x5 )X 2 k m  x2 ˆ À(x2 ‡ x3 ) À 1(x1 ‡ x4 ). so that we may ignore second order terms. x3 . Q9. x2 . and assume that the magnitude of the force between two atoms is k times the decrease in distance between them. x4 .An application of representation theory to molecular vibration 383 check our results. From the diagram. PR À P9R9 ˆ (x1 ‡ x6 ) ‡ 1(x2 ‡ x5 ). denote the new positions of the atoms by P9. we now calculate the equations of motion. 2 PQ À P9Q9 ˆ (x2 ‡ x3 ) ‡ 1(x1 ‡ x4 )X 2 Hence the force on the molecule at P in the direction of the ®rst coordinate axis is Àk(PR À P9R9) ˆ Àk(x1 ‡ x6 ) À 1 k(x2 ‡ x5 )X 2 Therefore. 2 k In the same way.

.18)(2) and (4)) because V÷3 ’ R6 vib was an A-invariant RG-submodule of V÷3 different from {0} and V÷3 .2. where we found that G is isomorphic to S4 . 32.20 Example We analyse the normal modes for the methane molecule CH4 . Label the corners of . The matrix A for  which x ˆ xA is therefore given by H I 1 1a2 1a2 0 0 1 f 1a2 1 1 0 0 1a2 g f g Àk f 0 1 1 1a2 1a2 0 g f gX Aˆ 1a2 1a2 1 1 0 g m f 0 f g d 1a2 0 0 1 1 1a2 e 1 0 0 1a2 1a2 1 You should check that the vectors which we gave in (32.14).17. . x6 . since these homogenous components were irreducible (see Corollary 32. We determined the symmetry group G of this molecule in Example 32. In our next example. . . the character ÷ of the RG-module R6 was given by ÷ ˆ ÷1 ‡ ÷2 ‡ 2÷3 X All the non-zero vectors in the homogeneous components for ÷1 and ÷2 gave normal modes. but we were able to write it as a sum of two subspaces of eigenvectors (those appearing in (32. 32.384 Representations and characters of groups   and we obtain similar equations for x3.19 Remark In Example 32. This illustrates a method which sometimes helps to deal with reducible homogeneous components. the situation is more complicated. The homogeneous component V÷3 for ÷3 was reducible.18) are indeed eigenvectors of A.

v42 . jX Thus. v43 . v41 . w2. v14 be unit vectors at corner 1 in the directions of the edges 12. for example. and let W be the vector space over R spanned by w1. v13 . 2. v13 . similarly. Let v12 . and we can quickly calculate the character ÷ of V: 1 ÷ 12 (1 2) 2 (1 2 3) 0 (1 2)(3 4) 0 (1 2 3 4) 0 . 24. and identify G with S4 . let v21 . We now introduce a new idea. 13. (2 4 3)X In order fully to exploit the symmetry of the methane molecule. with w i pointing towards corner i (1 < i < 4). and so on. 4. by taking four unit vectors w1. G permutes our twelve basis vectors of V. the rotations about the vertical axis through 1 are written as 1. v21 . w2. v32 . v23 . v31 . these four vectors span a 3-dimensional space. (2 3 4). 14. as shown below. Then V  R12. v23 . W  R3 and V and W are RG-modules. The action of G on V is easy to describe. w3. v14 . Since w1 ‡ w2 ‡ w3 ‡ w4 ˆ 0. Our main task is to ®nd RG-submodules of R15 ˆ V È W. we have v ij g ˆ v ig. w4. v34 . at each hydrogen atom we choose displacement axes along the edges of the tetrahedron. respectively. v24 . giving twelve vectors v ij .jg for all i. Let V be the vector space over R with basis v12 . 3.An application of representation theory to molecular vibration 385 the tetrahedron 1. for g in G. in all. w3 and w4 at the carbon atom. v24 be unit vectors at corner 2 in the directions of the edges 21. 23. thus.

27)). p2 ˆ (v31 À v13 ) ‡ (v14 À v41 ) ‡ (v43 À v34 ). ÷5 and 3÷4 (see (14. . Let p1 ˆ (v23 À v32 ) ‡ (v34 À v43 ) ‡ (v42 À v24 ). it is easy to calculate the character ö of W ± it is 1 ö 3 (1 2) 1 (1 2 3) 0 (1 2)(3 4) À1 (1 2 3 4) À1 By Section 18. 387. for g in G. . the character table of S4 is as shown at the top of p. . ÷3 . 5. we have wi g ˆ wig (1 < i < 4)X After recalling that w1 ‡ . The group G acts on W as follows.1. we can ®nd RG-submodules with characters ÷1 . all the basis vectors are moved by (1 2 3). ö ˆ ÷4 X By applying the elements ˆ ÷ i ( g À1 ) g gPG (i ˆ 1. j and this gives the expansion±contraction normal mode: We next describe the RG-submodule W5 with character ÷5 . The RG-submodule W1 with character ÷1 is spanned by ˆ v ij i. 3. and so on.386 Representations and characters of groups For example. We ®nd that ÷ ˆ ÷1 ‡ ÷3 ‡ 2÷4 ‡ ÷5 . 4) to R15 . (1 2) ®xes the basis vectors v34 and v43 only. ‡ w4 ˆ 0.

p4 ˆ (v21 À v12 ) ‡ (v13 À v31 ) ‡ (v32 À v23 )X The vector pi gives a rotation about the axis through the corner i and the centroid of the tetrahedron. It should be clear from the pictures that for all i with 1 < i < 4 and .An application of representation theory to molecular vibration 387 Character table of S4 1 ÷1 ÷2 ÷3 ÷4 ÷5 1 1 2 3 3 (1 2) 1 À1 0 1 À1 (1 2 3) 1 1 À1 0 0 (1 2)(3 4) 1 1 2 À1 À1 (1 2 3 4) 1 À1 0 À1 1 p3 ˆ (v12 À v21 ) ‡ (v41 À v14 ) ‡ (v24 À v42 ).

(Compare.17. its character is ÷3 . p3 . Therefore. q3 ). the dimension of W3 is 2. Let q1 ˆ (v12 ‡ v21 ) ‡ (v34 ‡ v43 ) À (v13 ‡ v31 ) À (v24 ‡ v42 ). Check that the character of W5 is ÷5 . p2 . Since q1 ‡ q2 ‡ q3 ˆ 0.) Now we construct the RG-submodule W3 of V with character ÷3 .388 Representations and characters of groups all g in G. we have pi g ˆ Æ p j for some j. q2 . Then W3 is an RG-submodule of V. Now p1 ‡ p2 ‡ p3 ‡ p4 ˆ 0. The RG-module W5 is the rotation submodule. q3 ˆ (v14 ‡ v41 ) ‡ (v23 ‡ v32 ) À (v12 ‡ v21 ) À (v34 ‡ v43 )X (Each q i is associated with an `opposite pair of edges'. the picture for p4 with the picture for the rotation vector v in Example 32. for example. so dim W5 ˆ 3. p4 ).) For all i with 1 < i < 4 and g in G. q2 ˆ (v13 ‡ v31 ) ‡ (v24 ‡ v42 ) À (v14 ‡ v41 ) À (v23 ‡ v32 ). we have q i g ˆ Æq j for some j. . if we let W 5 ˆ sp ( p1 . Let W3 ˆ sp (q1 . then W5 is an RG-submodule of V.

De®ne the vectors r1. r2 ˆ (v12 ‡ v21 ) ‡ (v23 ‡ v32 ) ‡ (v24 ‡ v42 ) À (v13 ‡ v31 ) À (v14 ‡ v41 ) À (v34 ‡ v43 ). r3. r2. by Corollary 32. r3 ˆ (v13 ‡ v31 ) ‡ (v23 ‡ v32 ) ‡ (v34 ‡ v43 ) À (v12 ‡ v21 ) À (v14 ‡ v41 ) À (v24 ‡ v42 ). r4 by r1 ˆ (v12 ‡ v21 ) ‡ (v13 ‡ v31 ) ‡ (v14 ‡ v41 ) À (v23 ‡ v32 ) À (v24 ‡ v42 ) À (v34 ‡ v43 ).14. We now come to the homogeneous component (V È W)÷4 of R15 . r4 ˆ (v14 ‡ v41 ) ‡ (v24 ‡ v42 ) ‡ (v34 ‡ v43 ) À (v13 ‡ v31 ) À (v12 ‡ v21 ) À (v23 ‡ v32 )X (The vector ri is associated with corner i.An application of representation theory to molecular vibration 389 In the RG-submodules W 1 . W 5 and W3 which we have found so far. all the non-zero vectors are eigenvectors of A.) .

390 Representations and characters of groups For each g in G and i with 1 < i < 4. r3. s4 ˆ (v41 ‡ v42 ‡ v43 ) À (v14 ‡ v24 ‡ v34 )X We have si g ˆ sig ( g P G. r3. s4 by s1 ˆ (v12 ‡ v13 ‡ v14 ) À (v21 ‡ v31 ‡ v41 ). s3 ˆ (v31 ‡ v32 ‡ v34 ) À (v13 ‡ v23 ‡ v43 ).24). The character of W4 is ÷4 (see Proposition 13. Note that r1 ‡ r2 ‡ r3 ‡ r4 ˆ 0. Next. 1 < i < 4). r2. so r1. r2. r4 among themselves. s1 ‡ s2 ‡ s3 ‡ s4 ˆ 0. we have ri g ˆ rig. Thus G permutes the vectors r1. s2 . s2 ˆ (v21 ‡ v23 ‡ v24 ) À (v12 ‡ v32 ‡ v42 ). s3 . de®ne the vectors s1 . r4 span a 3-dimensional RGsubmodule W4 of V. .

In this case. w3 . and so they give the ®nal 3-dimenvib sional space of eigenvectors (see Remark 32. and we can decompose V÷4 ˆ W 4 È W 4 in the following way. 1 < i < 4). r3 À 2s3 . (2) The vectors r1. in order to deal with the easier case of a molecule with identical atoms at the corners of the tetrahedron. r3. s4 span a 3-dimensional RG-submodule W9 of V with 4 character ÷4 . w2 .21) (1) The vectors r1 À 2s1 . W 4 and W is direct. r2 À 2s2 . Since dim (V È W )÷4 ˆ 9. w1 ‡ w2 ‡ w3 ‡ w4 ˆ 0. and let U1 ˆ fv P (V È W )÷4 : vh ˆ v for all h P HgX Since (vh)A ˆ (vA)h for all v P V÷4 and all h P H. 9 (32. The sum of W 4 . we have wi g ˆ wig ( g P G. it is very dif®cult to calculate the eigenvectors of A in (V È W )÷4 directly. r4 span the subspace V÷4 ’ R12 of V÷4 . upon the constants which appear in the equations of motion. so 9 (V È W )÷4 ˆ W 4 È W 4 È W X 9 We now break off temporarily from studying the methane molecule. We shall therefore press on and explain how to reduce the dif®culty to that of calculating the eigenvectors of a 3 3 3 matrix. and no central atom. s2 . To see why. it follows that U1 is A-invariant.An application of representation theory to molecular vibration 391 and s1 . and the character of W is ÷4 .19). . in fact. look at the picture of the vector r1 ! We return now to the methane molecule. so we cannot complete the work using only representation theory. The task in front of us is to ®nd the eigenvectors of A which lie in (V È W )÷4 ˆ W 4 È W 4 È W X 9 The solution of this problem depends. Let H be the subgroup of S4 consisting of those permutations which ®x the number 1. Now recall that w1 . r2. w4 span W. s3 . the space W does not enter our calculations. r4 À 2s4 span the 3-dimensional space of translation modes. The normal mode (sin ùt)r1 is sometimes called an `umbrella mode'.

r1 (1 2) ˆ r2 . s1 . r2 À 2s2 ‡ 3 cos Ww2 . A similar remark applies to U3. By means of representation theory. w1 (1 2) ˆ w2 . the space U 2 . s2 . and hence the matrix A. namely the translation vector r1 À 2s1 ‡ 3 cos Ww1 . w1 )X Once the equations of motion. r3 À 2s3 ‡ 3 cos Ww3 ). The eigenvectors of B then give three eigenvectors of A. But for all h P H. Therefore U 1 ˆ sp (r1 . s2 . the process of calculating the eigenvectors of the 3 3 3 matrix B gives nine eigenvectors of A which form a basis of (V È W )÷4 . w1 is easy to ®nd. and so dim U1 ˆ 3. and the matrix of A acting on r2.5). Better still. 1 H i H ˆ 3. r1 h ˆ r1. where U 3 ˆ sp (r3 . It is hard to imagine a more spectacular application of representation theory with which to conclude our text. s1 (1 2) ˆ s2 . have been calculated. w3 )X Therefore. s1 .392 We ®nd that Representations and characters of groups h3÷4 5 H. where W is the angle between an edge of the tetrahedron and the line joining a corner on the edge to the centroid. s1 h ˆ s1 and w1 h ˆ w1 . w2 is again B. . s3 . Thus we obtain the translation submodule sp (r1 À 2s1 ‡ 3 cos Ww1 . and since A commutes with the action of G. w2 ) is A-invariant. One eigenvector of A acting on r1. de®ned by U2 ˆ sp (r2 . w1 (see Exercise 32. it is possible to calculate the 3 3 3 matrix B of the action of A on r1. we have therefore reduced the initial problem of ®nding the eigenvectors of a 15 3 15 matrix A to that of calculating two of the eigenvectors of a 3 3 3 matrix. s1 .

(a) Show that the matrix B of b with respect to the basis e1 . and is called a normal mode. 0). (iii) 1 is an eigenvalue of C. (c) Deduce from (a) and (b) that b is a rotation if det B ˆ 1. The equations of motion of the molecule have the form  x ˆ xA. where x P R3 n and the 3n 3 3n matrix A has 3n linearly independent eigenvectors. Deduce that det B ˆ Æ1. If V÷ i is irreducible. e2 ˆ (0. Suppose that b P O(R3 ). (d) Show that if b is a rotation through an angle ö about some axis. 2. 3. The action of any g P G on R3 n commutes with A. 1. (b) Let C ˆ (det B)B. Prove successively that (i) C has a real eigenvalue. 1). To determine the eigenvectors of A (and hence all solutions of the equations of motion). and Àb is a rotation otherwise. e3 of R3 satis®es BBt ˆ I (where Bt denotes the transpose of B). 0). 0. (ii) C has a positive eigenvalue. e2 . and let e1 ˆ (1. All solutions are linear combinations of normal modes. The space R3n of displacement vectors is an RG-module. 2. . then x ˆ sin (ùt ‡ â)u (or x ˆ (t ‡ â)u if ù ˆ 0) is a solution of the equations of motion. If u is an eigenvector of A. Exercises for Chapter 32 1. 0. The symmetry group G of a molecule with n atoms consists of those distance-preserving endomorphisms of R3 which send each atom to an atom of the same kind. it is suf®cient to ®nd the eigenvectors of A restricted to each homogeneous component V÷ i of the RG-module R3 n . then all non-zero vectors in V÷ i are eigenvectors of A. 4. 5. then tr B ˆ 1 ‡ 2 cos ö. Suppose that G is the symmetry group of some molecule in R3 . e3 ˆ (0.An application of representation theory to molecular vibration 393 Summary of Chapter 32 1. with eigenvalue Àù2 .

) 4.17.20. r2. v13 . Consider the triatomic molecule studied in Example 32.20. 3. r3. r2. What property of r1. Consider the space spanned by the vectors r1. and verify that A is symmetric. Label the corners of the tetrahedron 1. w2 . b b X 0. 4 and let 0 denote the centroid of the tetrahedron. X X X . and let the position vector of the molecule be ˆ iTˆ j xij v ij ‡ p 3 ˆ iˆ1 yi wi X (a) Prove that cos (/ 012) ˆ (2a3) and cos (/ 102) ˆ À1a3.20. 2. if g is not a rotationX 3. r3. w1 . Work with 15 unit displacement vectors v12 . . v43 . r4 prompted us to use these vectors? 5. r4 given in Example 32.394 Representations and characters of groups Show that the sum of the character ÷ T of the translation submodule and the character ÷ R of the rotation submodule has value at g P G given by V b 2(1 ‡ 2 cos ö). if g is a rotation b ` through angle ö (÷ T ‡ ÷ R )( g) ˆ about some axis. (See the paragraph before Proposition 32.7. The purpose of this exercise is to derive the equations of motion of the methane molecule. Find a basis for this space which is simpler than the one which we have used. and so ®nd explicitly the 3 3 3 matrix B which appears at the end of Example 32. Take the axes for the displacement coordinates as shown below:  Calculate the equations of motion x ˆ xA with respect to these axes. w3 as described in Example 32.

j . 3   with similar expressions for y2 and y3 . q2 . show  m1 x41 ˆ À k 1 [x14 ‡ x41 ‡ 1(x42 ‡ x43 ‡ x12 ‡ x13 )] 2 p 1 À 3 k 2 [x41 ‡ x42 ‡ x43 À 1 (3a2)( y1 ‡ y2 ‡ y3 )]. Prove that  m1 x12 ˆ À k 1 [x12 ‡ x21 ‡ 1(x13 ‡ x14 ‡ x23 ‡ x24 )] 2 p 1 À 3 k 2 [x12 ‡ x13 ‡ x14 ‡ (3a2)( y1 À 1 y2 À 1 y3 )]. Assume that the magnitude of the force between hydrogen atoms is k1 times the decrease in distance between them. 3 3         with similar expressions for x13. i. p2 . x32 . x14 . Finally. and the magnitude of the force between a hydrogen atom and a carbon atom is k2 times the decrease in distance between them. 2 with similar expressions for the edges 13. Also. 3 with similar expressions for the edges 02. x31 . p1 . x21 . 23. 24. p3 . 14. x23 . 34. 03. Also. q1 . and m2 the mass of a carbon atom. show p  m2 y1 ˆ Àk 2 [ (2a3)(x12 ‡ x13 ‡ x14 À x41 À x42 À x43 ) ‡ 4 y1 ]. show that the length of the edge 01 has decreased by p (2a3)(x12 ‡ x13 ‡ x14 ) ‡ y1 À 1( y2 ‡ y3 ). x24 . (d) The equations in part (c) determine the 15 3 15 matrix A in the  equations of motion x ˆ xA. show that the length of the edge 04 has decreased by p (2a3)(x41 ‡ x42 ‡ x43 ) À 1( y1 ‡ y2 ‡ y3 )X 3 (c) Let m1 denote the mass of a hydrogen atom. Finally.An application of representation theory to molecular vibration 395 (b) Show that the decrease in the length of the edge 12 from its original length is x12 ‡ x21 ‡ 1(x13 ‡ x14 ‡ x23 ‡ x24 ). Verify that the vectors ˆ v ij . 3   with similar expressions for x42 and x43 . x34 .

. . . and use Exercise 23. È Um. s1 .8. Prove that if u and v are non-zero elements of U1. . a sum of m isomorphic irreducible RG-modules. (See 32.  (b) Calculate the equations of motion.) We assume that ÷ i is the character of an irreducible RG-module which remains irreducible as a CGmodule. Consider a hypothetical molecule in which there are four identical atoms at the corners of a square. (c) Assume that the eigenvectors of the m 3 m matrix A u are known. . s1 A ˆ b21 r1 ‡ b22 s1 ‡ b23 w1 . are eigenvectors of A. indeed. (f) Verify that p (1. Show how to ®nd the eigenvectors of A in V÷ i . then Au ˆ Av . uW m . . Suppose that V÷ i ˆ U1 È . sp (uW1 .20. In this exercise. (Hint: compose the function w 3 wA with a projection. .) (b) Let A u denote the matrix of the endomorphism w 3 wA of sp (uW1 . uW m ) is an A-invariant vector space of dimension m. x ˆ xA. 6) is an eigenvector of B. (a) Find the normal modes of the molecule. . 7. w1 are as in Example 32. where the vectors r1. . (e) Find the entries b ij in the 3 3 3 matrix B which are given by r1 A ˆ b11 r1 ‡ b12 s1 ‡ b13 w1 .15(5). (a) Prove that for all non-zero u in U1. . w1 A ˆ b31 r1 ‡ b32 s1 ‡ b33 w1 . we derive a method for simplifying the problem of ®nding the eigenvectors of A when the homogeneous component V÷ i is reducible. . À2. We reduce our problem to that of ®nding the eigenvectors of an m 3 m matrix.396 Representations and characters of groups which appear in Example 32.20. X X X . eigenvectors of A. uW m ) with respect to the basis uW1 . For 1 < i < m. Assume that the only internal forces are along the sides of the square. and check that the vectors you found in part (a) are. 6. let W i be an RG-isomorphism from U1 to Ui .

First. Ker ö ˆ {1. If Ker W ˆ f1g then W is an isomorphism. b: a2 m ˆ b2 ˆ 1. so G ’ An v G. 0 < j < 1. Since G is simple and Ker W v G. 0 < k < 1. À1) and y ˆ (d. a2 } and Ker ø ˆ {1. then h g 2 i would be a normal subgroup different from G and {1}. 2. where m is odd. 5. 1). d À1 cd ˆ cÀ1 i. since G is abelian. Since G ’ An Tˆ G. The elements of D2 m 3 C2 are (c i d j . G ’ An ˆ f g P G: g is even}. Therefore g p ˆ 1.4 again that ì is a homomorphism. and G Tˆ {1} since G is simple. Check using the method of Example 1. so ì is an isomorphism. Let p be a prime number which divides |G|. Then kgl is a normal subgroup of G. Check that x 2 m ˆ y 2 ˆ 1. and D2 m ˆ hc. If G were in®nite. c2 }. y À1 xy ˆ x À1 X By Example 1. Therefore G ’ An and (G ’ An )h are the only right cosets of G ’ An in G. hence G is ®nite. d: cm ˆ d 2 ˆ 1. (a) Using the method of Example 1. Also Ker ì ˆ {1} and Im ì ˆ L. the function W: D4 m 3 D2 m 3 C2 de®ned by 397 . we have g ˆ a (ghÀ1 )h P (G ’ An )h. and if Ker W ˆ G then H ˆ f1g. For all odd g in G.4. (b) Since b2 ë ˆ I but (bë)2 ˆ Y 2 ˆ ÀI. it follows that ë is not a homomorphism. bÀ1 ab ˆ aÀ1 i. so kgl ˆ G. and Ga(G ’ A n )  C2 . Let g be a non-identity element of G. Then h g p i is a normal subgroup of G which is not equal to G. and so G is cyclic of prime order. Let x ˆ (c( m‡1)a2 . (À1) k ) for 0 < i < m À 1.4. 4. Let D4 m ˆ ha. it is routine to verify that ö and ø are homomorphisms. 3. Note that all subgroups of G are normal.Solutions to exercises Chapter 1 1. either Ker W ˆ f1g or Ker W ˆ G. we may choose h P G with h P An .

and gÀ1 g ˆ ggÀ1 ˆ 1. Hence. by Lagrange's Theorem). then g ˆ a j for some integer j and dnj jn. À1) and hence Im W ˆ D2 m 3 C2 . g À1 g ( g P G). These relations show that every element of the group kA. (b) Assume that G ˆ hai and jGj ˆ dn. if jGj is even then there exists g P G such that g Tˆ 1 and the subset f g. and the identity element is in a subset of size 1. k. Since Im W ˆ kx. by part (b). Moreover. If g P G and g n ˆ 1. r with 0 < r . 6. thus H is cyclic. Since r . then h n ˆ 1 for all h P H (since the order of h divides n. we have r ˆ 0. which is a cyclic group of order n. Now kxl and k yl have order n. we conclude that W is an isomorphism. kA. k P G then (gh)k ˆ g(hk). Let G be the set of non-zero complex numbers. Bl has order 16. so gh P G. Finally. If H is a subgroup of G of order n. hence g P kad l. it follows that H ˆ ke2ðia n l. If g. 9. k. (a) Let G ˆ kal and suppose that 1 Tˆ H < G. If g. where H ˆ { g P G: gn ˆ 1}. are all distinct. 8. Therefore H < f g P G: g n ˆ 1g ˆ he2ðia n iX Since j Hj ˆ n ˆ jhe2ðia n ij. 0 eÀiða4  Bˆ  1 X 0 0 À1 Check that A8 ˆ I.398 Representations and characters of groups W: a i b j 3 x i y j (0 < i < 2m À 1. Each such subset has size 1 or 2. so g ˆ gÀ1 and g has order 2. 0 < j < 1) is a homomorphism. As |D4 m | ˆ |D2 m 3 C2 |. 1) and x m ˆ (1. Bl has the form A j B k with 0 < j < 7. g À1 g has size 1. Therefore a j ˆ a kq and so H ˆ kak l. y P H. B2 ˆ A4 and BÀ1 AB ˆ AÀ1 . and so x is a power of y. 0 and a k P H. h P G then gh Tˆ 0.     ei jða4 0 0 ei jða4 j j . A Bˆ A ˆ X 0 eÀi jða4 ÀeÀi jða4 0 Since these matrices. (c) If x and y are elements of order n in the ®nite cyclic group G. it contains x 2 ˆ (c. also 1 P G and 1g ˆ g1 ˆ g for all g P G. De®ne matrices A. yl. 0 such that ai P H. B as follows:   eiða4 0 Aˆ . Hence a r ˆ a j aÀqk ˆ a j (a k )Àq P H. It follows that f g P G: g n ˆ 1g ˆ had i. Choose k as small as possible such that k . h. First observe that there exists i . also H has order n. If 1 Tˆ a j P H then j ˆ qk ‡ r for some integers q. . 7. with 0 < j < 7. 0 < k < 1. if g P G then g À1 ˆ 1a g P G. then x. Thus G is a group under multiplication. We deduce that hxi ˆ H ˆ h yiX Thus x P h yi. so dj j. Partition G into subsets f g.

hence they form a basis of V. . (2) A (3): If Ker W ˆ {0} then dim (Im W) ˆ dim V (by (2. F F F . it follows that v is a linear combination of u1 . gH are a the two left cosets. 2. Thus W is injective. If g P H then gÀ1 Hg ˆ H. 3. . F F F . v P V and uW ˆ vW then (u À v)W ˆ 0. w1. Then V ˆ U ‡ W. Thus U ’ W ˆ {0}. ì j P F. 4. . (ë(wWÀ1 ))W ˆ ë(wWÀ1 )W ˆ ëwX Hence (u ‡ w)WÀ1 ˆ uWÀ1 ‡ wWÀ1 and (ëw)WÀ1 ˆ ë(wWÀ1 ). this gives ë i ˆ ì j ˆ 0 for all i. Ker W ˆ {0}. . . . ws is a basis of V. ur .12)). (a) Assume ®rst that V ˆ U1 È U2 È U3. . ur are linearly independent. ì j in F. If v P V then v ˆ u ‡ w for some u P U and w P W. while H. . . w2 P W. It is easy to see that V ˆ U ‡ W. Thus v ˆ 0 and so U ’ W ˆ {0}. w s . 5. so u À v P Ker W ˆ {0}. . Then v ˆ v ‡ 0 ˆ 0 ‡ v and this gives us two expressions for v as the sum of an element in U and an element in W. Suppose jG: Hj ˆ 2 and let g P G. ws are linearly independent. Since such expressions are unique. Conversely. we have (uWÀ1 ‡ wWÀ1 )W ˆ (uWÀ1 )W ‡ (wWÀ1 )W ˆ u ‡ w. . Hg are the two right cosets of H in G. we have u1 ˆ . . .12). ur . ‡ ì s ws for some ë i . the expression 0 ˆ 0 ‡ 0 is the unique expression for 0 as the sum of vectors in U and W. V ˆ U È W. v ˆ 0. . and so u ˆ v. . . Suppose that ë1 u1 ‡ X X X ‡ ë r ur ‡ ì1 w1 ‡ X X X ‡ ì s ws ˆ 0 with all ë i . . Let v P U ’ W. If u1 ‡ w1 ˆ u2 ‡ w2 with u1 . and so ë1 u1 ‡ X X X ‡ ë r ur ˆ ì1 w1 ‡ X X X ‡ ì s ws ˆ 0X As u1 . . . Therefore u1 . (3) A (1): Assume that Im W ˆ V. w s . Therefore u1 . As W is surjective and injective. . . . so by Exercise 3. By (2. u2 P U and w1.7)). This shows that V ˆ U È W. ur . this forces ë i ˆ 0 for all i. Then u ˆ u1 ˆ u2 ‡ u3 for some ui P Ui (1 < i < 3). . W is invertible. ws are linearly independent. F F F . Let u P U1 ’ (U2 ‡ U3 ). w1. . . If u. If v P U ’ W then v ˆ ë1 u1 ‡ . w1. First suppose that V ˆ U È W. . And if g P H then H. so W is surjective. Since W is a linear transformation. similarly ì i ˆ 0 for all i. . Chapter 2 1. ur and w is a linear combination of w1 . w P W and ë P F. Assume ®rst that V ˆ U È W. w1. so Im W ˆ V (by (2. ws span V. . so Ker W ˆ {0}. since u is a linear combination of u1 . . . . suppose that u1 . . . Therefore Hg ˆ gH. Since V ˆ U È W. ‡ ë r ur ˆ ì1 w1 ‡ . Let u. hence u1 ˆ u2 and w1 ˆ w2. w1 . so WÀ1 is a linear transformation. . ur . u r . . then u1 À u2 ˆ w2 À w1 P U ’ W ˆ {0}. since u1 . Since u1 ‡ 0 ‡ 0 ˆ 0 ‡ u2 ‡ u3 and the sum U1 ‡ U2 ‡ U3 is direct. . . . . . and so gÀ1 Hg ˆ H again. (1) A (2): If W is invertible then W is injective. . Now suppose that V ˆ U ‡ W and U ’ W ˆ {0}. . j. . .Chapter 2 399 10. Hence H v G.

. ur . .32. . . The representations r2 and r3 are faithful. ur for Im W and a basis w1. Then (ai )r ˆ Ai for all integers i (including i . each r j is a representation. . . ö: V 3 V by W: (x. . the matrix [W]B is diagonal. so 1(v ‡ vW) P U. by Exercise 3. . ‡ dim Ur. . . Suppose ®rst that W is a projection. if [W]B has the given form. by induction on r. so V Tˆ Im ö È Ker ö. m À 1 and i . say B. 1)). 2 3 (b) Let V ˆ R2. . U2 ’ (U1 ‡ U3 ) ˆ U 3 ’ (U1 ‡ U2 ) ˆ f0g. . Now suppose that U1 ’ (U2 ‡ U3 ) ˆ U2 ’ (U1 ‡ U3 ) ˆ U3 ’ (U1 ‡ U2 ) ˆ {0}. . De®ne W. and so r is a representation. 2. Let V ˆ R2. Check that A3 ˆ B3 ˆ C 3 ˆ I. u2 ˆ u9 and u3 ˆ u9 . assume that Am ˆ I. È Ur ) (see (2. . Ker W ˆ sp ((0. the diagonal entries being r 1's followed by s 0's. ws is a basis. (a i a j )r ˆ (a i‡ j )r ˆ A i‡ j ˆ A i A j ˆ (a i r)(a j r). Assume that ui . then clearly W2 ˆ W. 9. 0) and ö: (x. Hence by Exercise 1. Then V ˆ Im W È Ker W by Proposition 2. but r1 is not. Therefore for all integers i. ‡ dim Ur. U2 ˆ sp ((0. . Therefore V ˆ U È W. 1 2 3 1 Similarly. The construction of the basis B is similar to that in Solution 8. . . w1. Conversely. Then u1 . if V ˆ U1 È . . 0)). Similarly. First. . y) 3 ( y. j. Therefore V ˆ U1 È U2 È U3. . 2 2 2 À vW) P W. and Im ö ˆ Ker ö ˆ sp ((1. Let v P V. Similarly. 1)). . 0)X Then Im W ˆ sp ((1. Then v ˆ 1(v ‡ vW) ‡ 1(v À vW)X 2 2 Observe that 1(v ‡ vW)W ˆ 1(vW ‡ v). so V ˆ Im W È Ker W. If v P U ’ W then v ˆ vW ˆ Àv. if V ˆ U È W then dim V ˆ dim U ‡ dim W. u9 P Ui (1 < i < 3) and i u1 ‡ u2 ‡ u3 ˆ u9 ‡ u9 ‡ u9 . and U1 ˆ sp ((1. U3 ˆ sp ((1. . so v ˆ 0. of V. 0)). 0)). 0). .10)). . È Ur then V ˆ U1 È (U2 È . Since ui W ˆ ui for all i and wj W ˆ 0 for all j. 8. Take a basis u1 . dim (U2 È . . By Exercise 4. suppose that r is a representation of G. so dim V ˆ dim U1 ‡ . Therefore U1 ’ (U2 ‡ U3 ) ˆ {0}. ws for Ker W. by Exercise 4.400 Representations and characters of groups u2 ˆ u3 ˆ 0. 1)). Then 1 2 3 u1 À u9 ˆ (u9 À u2 ) ‡ (u9 À u3 ) P U1 ’ (U2 ‡ U3 ) ˆ {0}. so u1 ˆ u9 . Then I ˆ 1r ˆ (am )r ˆ (ar) m ˆ Am X Conversely. y) 3 (x. 1 2(v Chapter 3 1. . Thus V ˆ U ‡ W. . È Ur ) ˆ dim U2 ‡ . so W is a projection. 7. 6. More generally.

then a3 r j Tˆ I. hence r is equivalent to r. 0 < s < 1) are all different. Similarly r4 is faithful. let  Tˆ  1 1 X Ài i Then T À1 (gr4 )T ˆ gr1 for all g P G. 8. The representations r1 and r4 are equivalent: to see this. 0 1 1 B ˆ d0 0 H I 0 0 À1 0 eX 0 1 Then the function r: a r b s 3 A r Bs (0 < r < 3.4). since a2 r2 ˆ I and a3 r3 ˆ I. hence r2 is not equivalent to any of the others.) If j Tˆ 2. 6. so ó is equivalent to r. F) is abelian. then there are invertible matrices S and T such that gó ˆ SÀ1 ( gr)S and gô ˆ T À1 (gó)T for all g P G. And if j Tˆ 3. then a2 r j Tˆ I. 4. we have S 6 ˆ T 2 ˆ I. T ˆ D. then gô ˆ (ST)À1 (gr)(ST). 0 < s < 1) is a faithful representation of D8 ˆ ka. then gr ˆ (T À1 )À1 (gó)T À1. 7. By Theorem 1. F) and GL (1. b: a4 ˆ b2 ˆ 1. (1) For all g P G.2(1). so r1 is faithful. T ˆ ÀB. No: let G be any non-abelian group and let r be the trivial representation. (4) S ˆ C. But r2 and r3 are not faithful. . Check that in each of the cases (1) S ˆ A. Compare Example 3. (2) If r is equivalent to ó then there is an invertible matrix T such that gó ˆ T À1 ( gr)T for all g P G. T À1 ST ˆ S À1 X It follows that each r k is a representation (see Example 1. (To ®nd T. T ˆ B. But Im r < GL (1. (3) If r is equivalent to ó and ó is equivalent to ô. so r is equivalent to ô. De®ne the matrices A and B by 0 A ˆ d À1 0 H I 1 0 0 0 e. 0 < j < 1)X 401 It is easy to check that r is a representation of G. ®rst work out the eigenvectors of C.Chapter 3 3. De®ne r by (a i b j )r ˆ (À1) j (0 < i < n À 1. 5. hence r3 is not equivalent to any of the others. IÀ1 (gr)I ˆ gr. T ˆ B. (3) S ˆ ÀA. (2) S ˆ A3 .10. The matrices A r Bs (0 < r < 5. bÀ1 ab ˆ aÀ1 l. GaKer r  Im r. Therefore GaKer r is abelian.

since one of g. Next. v in V and ë in F. Let V ˆ R4. g H [ g]B 1 1 d0 0 H 1 d0 0 1 0 0 1 0e 0 1 I 0 0 1 0e 0 1 I 0 d1 0 H 1 d0 0 H (1 2) 1 0 0 0e 0 1 I 0 0 À1 0 e À1 1 I 0 d0 1 H 1 d0 0 H (1 3) I 0 1 1 0e 0 0 I 0 0 1 À1 e 0 À1 [ g]B 2 g H [ g]B 1 1 d0 0 H 1 d0 0 (2 3) 0 0 0 1e 1 0 I 0 0 0 1e 1 0 I H (1 2 3) 0 d0 1 H 1 d0 0 1 0 0 1e 0 0 I 0 0 À1 1 e À1 0 I H (1 3 2) 0 d1 0 H 1 d0 0 I 0 1 0 0e 1 0 I 0 0 0 À1 e 1 À1 [ g]B 2 All the matrices [g]B 2 have the form H I 1 0 0 d 0 j j eX 0 j j 2.402 Representations and characters of groups Chapter 4 1.2. and (vg)h ˆ Àv. For all u. h P Sn . since either vg ˆ v ˆ vh (if g. a a Then v(gh) ˆ Àv. 0 < j < 1) is a representation of Q8 over R. g P Q8. so V is an FG-module. h P An ) or vg ˆ Àv ˆ vh (if g. 3. Let v P V and g. assume that gh P An . We have now checked all the conditions in De®nition 4.4(1). If we put . we have v g P V . H I H I 0 0 1 0 0 1 0 0 f À1 0 0 f 0 0g 0 0 1g g f g. and (vg)h ˆ v. h P An ). B2 ˆ A2 . By Theorem 4. Assume ®rst that gh P An . Let A ˆ f d 0 0 0 À1 e and B ˆ d À1 0 0 0e 0 0 1 0 0 À1 0 0 Check that A4 ˆ I. Let g P Sn .2. (ëv) g ˆ ë(v g). Then v(gh) ˆ v. V becomes an RQ8 -module if we de®ne vg ˆ v( gr) for all v P V. v1 ˆ v. BÀ1 AB ˆ AÀ1 X Hence r: a i b j 3 A i B j (0 < i < 3. h is in An and the other is not. (u ‡ v) g ˆ ug ‡ v gX It remains to check (2) of De®nition 4.

where vg ˆ v(gr1 ) for . â)a ˆ (á ‡ â. À1)) and V. If C is a matrix obtained from A by permuting the columns. v3 ˆ (0. . 4. 0). 0. 2. since ó is equivalent to r. Then (á. . k . First consider the FG-module V ˆ F 2. You may ®nd it helpful ®rst to check that if H 0 1 f1 0 f f 1 M ˆf FF f d F 403 v1 ˆ (1. 0. â À á) P U. Therefore ó is reducible. 0. â)a ˆ (á À â.5. 3. 0. Let U be a non-zero FGsubmodule of V. then C ˆ AQ for some permutation matrix Q. and AM is obtained from A by swapping the ®rst two columns. Then ó is equivalent to ô. 1. 0). 0. v2 ˆ (0. To solve the exercise. â) ‡ (á. j X kˆ1 Hence PA ˆ B. á ‡ â) P U. Then r is equivalent to a representation ô of the form H I Xg 0 e ( g P G) ô: g 3 d Yg Zg where X g is a k 3 k matrix and 0 . 0. 1). ` 1. â) P U with (á. we deduce that (1. 1) or (1. sp ((1. Let G ˆ D12 and let r1 . â) À (á. n. 0g g g g g e 1 I 0 then MA is obtained from A by swapping the ®rst two rows. v4 ˆ (0. Hence the FG-submodules of V are {0}. Since at least one of á ‡ â and á À â is non-zero. It is easy to verify that V is an FG-module. and let (á. 0). pij ˆ X 0. r4 be the representations of G de®ned in Exercise 3. the proof is similar to that for the rows. let g be the permutation in Sn which has the property that for all i.Chapter 5 then for all i. row i of B ˆ row ig of AX Let P be the n 3 n matrix ( pij ) de®ned by V if j ˆ ig. . and the ij-entry of PA is n ˆ pik akj ˆ a ig. if j Tˆ igX Then P is a permutation matrix. v i a and v i b are as required in the question. â) Tˆ (0. 1. Suppose that r has degree n and r is reducible. . À1) belongs to U. 0). and (á. Chapter 5 1. 1)). sp ((1.

1) or (1. bl Tˆ G. Since (1. C À1 AC ˆ AÀ1 and C À1 BC ˆ BÀ1 . Then (1. AB ˆ BA. 1) or (1. By the solution to Exercise 1. it follows that dim U > 2.) Therefore r1 is irreducible. On the other hand. (a) It is easy to check the given relations. if å ˆ ç ˆ 1 then sp ((1. also r4 is irreducible. Using the relations. jGj Therefore jGj ˆ 18. either (1. Finally. r3 is irreducible. Hence either (1. let u be (1. g P G.404 Representations and characters of groups v P V.Cˆ çÀ1 1  1 X 0 Thus jGj < 18. Hence. (d) Let V ˆ C2 be the CG-module obtained by de®ning vg ˆ v(gr) for all v P C2 . Suppose that U is a non-zero FG-submodule of V. there exists a cube root î of unity such that   î 0 gr ˆ À1 X 0 î But there are only three distinct cube roots of unity. where H is the subgroup {1. g P G. Now u and ua are linearly independent unless å ˆ 1. 9. by an argument similar to that for r1. where H is the subgroup {1. so r is reducible. and also (1. then U is a CHsubmodule. Bˆ 0 å À1 0 Check that A3 ˆ B3 ˆ C 2 ˆ I. x 2 ˆ 4X1 ‡ a2 ‡ 4a3 X . If U is a non-zero CG-submodule of V. is a multiple of 9 and jGj . Then U is an FH-module. bl with g1 r ˆ g2 r. Consequently U ˆ V and so V is irreducible. Now let V ˆ F 2 with vg ˆ v(gr2 ) for v P V.4). g2 P ka. 1)) is an FG-submodule of V and r2 is reducible. 1)b. accordingly. Let V ˆ {0} and let 0 g ˆ 0 for all g P G. Then V is neither reducible nor irreducible. Therefore r is never faithful. 4. (b) Let    å 0 ç Aˆ . 1) ˆ (1. Hence r is a representation (compare Example 1. Chapter 6 1X (a) xy ˆ À2X1 À a3 ‡ ab ‡ 3a2 b ‡ 2a3 b. À1) lies in U. 1)) is a CG-submodule of V. since r1 and r4 are equivalent. À1) lies in U. c}. (See Example 5. bl. À1) and (1. 0 < k < 1)X that |ka. À1) (so that u P U). it is clear Hence. 1)a are linearly independent. by the solution to Exercise 1. 1) and (1. However. 5. 1)a ˆ À(1.5(2) for an alternative argument. we may write every element of G in the form ai b j c k (0 < i < 2. if either å Tˆ 1 or ç Tˆ 1 then dim U ˆ 2 and so r is irreducible. bl| ˆ 9 and ka. 0 < j < 2. by Lagrange's Theorem.   0 0 . (c) For every element g of ka. À1)a are linearly independent. and u and ub are linearly independent unless ç ˆ 1. b}. g P G. Hence sp ((1. so there exist distinct g1 . 1) or (1. yx ˆ À2X1 À a3 ‡ b ‡ 2a2 b ‡ 3a3 b.

4. Hence ch ˆ hc ˆ c. . ë P F and g P G. there exists a unique h in G such that gi h ˆ gj . u2 P U. and take r ˆ 1 ‡ a. . and bz ˆ 1 ‡ a2 ˆ zb. j. g. hence 0r ˆ v0 ˆ 0. Note ®rst that if u is an element of a vector space. v2 a ˆ ù2 v2 . (ab)r ˆ f d0 1 0 0e d1 0 0 0e 1 0 0 0 0 1 0 0 3. € € (b) c2 ˆ c hPG h ˆ hPG ch ˆ jGjc. then vr ˆ 0 and neither v nor r is 0.) 3. so do gh and hg. If r P CG then ˆ € € rˆ ë g gz ˆ ë g zg ˆ zr. v5 be the natural basis for the permutation module for G over F. Relative to the basis 1. 5. (ëu1 )Wö ˆ (ë(u1 W))ö ˆ ë(u1 (Wö)). (u1 g)Wö ˆ ((u1 W) g)ö ˆ ((u1 W)ö) g ˆ (u1 (Wö)) gX 2. 6. and u ‡ u ˆ u. a. Let C2 3 C2 ˆ ka. Hence W is a CG-submodule of CG. . No: let G ˆ ka: a2 ˆ 1l.5(2) or the solution to Exercise 5. so rz ˆ 2. ab of F(C2 3 C2 ). Hence a i b j z ˆ za i b j for all i. It is easy to show that V0 is an FG-submodule of V.Chapter 7 405 (b) az ˆ ab ‡ a3 b ˆ a2 ba ‡ ba ˆ za. Now 0r ˆ (0 ‡ 0)r ˆ 0r ‡ 0r. Use the argument of either Example 5. the regular representation r is given by H I H I 0 1 0 0 1 0 0 0 f1 0 0 0g f0 1 0 0g f g f g 1r ˆ f g. Then . then u ˆ 0. and v0 ˆ v(0 ‡ 0) ˆ v0 ‡ v0. j and so gz€ zg for all g P G. ar ˆ f d0 0 0 1e d0 0 1 0e 0 0 1 0 0 0 0 1 H I H I 0 0 0 1 0 0 1 0 f0 0 1 0g f0 0 0 1g f g f g br ˆ f gX g.3 to prove that W is irreducible. For all u1 . g ë g g with ë g P C. Let v1 ˆ 1 ‡ ù2 a ‡ ùa2 and v2 ˆ b ‡ ù2 ab ‡ ùa2 b. Let a ˆ (1 2 3 4 5) and let v1 . b: a2 ˆ b2 ˆ 1. ab ˆ bal. The reason is that for all i. . Check that v1 a ˆ ùv1 . Then W: ë1 v1 ‡ X X X ‡ ë5 v5 3 ë1 1 ‡ ë2 a ‡ ë3 a2 ‡ ë4 a3 ‡ ë5 a4 is the required FG-isomorphism. v1 b ˆ v2 and v2 b ˆ v1. hence W is an FG-homomorphism. Let x P G. If r ˆ 1 À g. b. we have (u1 ‡ u2 )Wö ˆ (u1 W ‡ u2 W)ö ˆ u1 (Wö) ‡ u2 (Wö). Let V be the trivial FG-module and let 0 Tˆ v P V and 1 Tˆ g P G. (c) All the entries in [W]B are 1 (compare the solution to Exercise 2). so (v i a)W ˆ v i‡1 W ˆ a i‡1 ˆ (v i W)a. s ˆ 1 À a. (Note that v i W ˆ ai . Chapter 7 1. (a) As g runs through G.

(vö)g ˆ (vg)ö ˆ vö. If v P V0 then (vajGj)W ˆ v. (a) W is easily seen to be a linear transformation. Also (á1 ‡ âx)xW ˆ (â1 ‡ áx)W ˆ (â À á)(1 À x) ˆ (á À â)(1 À x)x ˆ (á1 ‡ âx)WxX Hence W is an FG-homomorphism. since all diagonal matrices . Chapter 8 1. (wöÀ1 )g ˆ (wg)öÀ1 ˆ wöÀ1 . 2ˆ 3 V0 ˆ sp (v1 ‡ v2 . (Find eigenvectors for x. (b) (á À â)(1 À x)W ˆ ((á À â) À (â À á))(1 À x) ˆ 2(á À â)(1 À x). so W 0 öÀ1  V0 . Then RG ˆ sp (1 ‡ a ‡ b ‡ ab) È sp (1 ‡ a À b À ab) È sp (1 À a ‡ b À ab) È sp (1 À a À b ‡ ab)X 3. and let V be a 2-dimensional vector space over C with basis v1 . Suppose r is reducible. . Then by Maschke's Theorem. Let g P G. is an FG-isomorphism from V0 to W 0 . ab ˆ ba). 6. For all w P W 0 . noting that V W  V0 . v4 be the natural basis of the permutation module V for G over F. a. hence W is surjective. this makes V into a CGmodule. For all v P V0 . (c) Let B be the basis 1 À x. we see that W is an FGhomomorphism from V to V0 . v3 ‡ v4 ) and (FG)0 ˆ sp g X gPG Since V0 and (FG)0 have different dimensions. where ù ˆ e2ðia3 . Hence the function ö. it follows from Exercise 4 that V and FG are not isomorphic FG-modules. 1 ‡ x. 4. . In the notation of Exercise 3. V ˆ sp (Àùv1 ‡ v2 ) È sp (Àù2 v1 ‡ v2 ). v2 . b. restricted to V0 . ì g P C)X 0 ìg Then (gó)(hó) ˆ (hó)(gó) for all g. Let G ˆ {1. r is equivalent to a representation ó of the form   ëg 0 gó ˆ (ë g . . and Ker W ˆ Im W ˆ sp (v2 ). . Suppose that ö: V 3 W is an FG-isomorphism.406 Representations and characters of groups ˆ gPG vxg ˆ ˆ gPG vg ˆ ˆ gPG v gxX Hence (vx)W ˆ vW ˆ (vW)x. ì P C) then W is a CG-homomorphism from V to V. and so V0 ö  W 0 .) 2. g P G. 5. Let G be any group. De®ne vg ˆ v for all v P V. If we let W: ëv1 ‡ ìv2 3 ëv2 (ë. No: let v1 . ab}  C2 3 C2 (so a2 ˆ b2 ˆ 1. Hence W2 ˆ 2W. h P G. 4.

ar2 ˆ (À1)X Let C3 ˆ kb: b ˆ 1l and let ù ˆ e2ðia3 . 0 for all x P G. . . Also ˆ ˆ [ug. u] . r2 . 0)) is the only 1-dimensional CG-submodule of V. . r3 . v gg À1 ] by part (1) ˆ [ug À1 . (3) Let W ˆ U c. (1. 5. . Irreducible representations r1 . r3 : 1r1 ˆ br1 ˆ b2 r1 ˆ (1). [u. y). h P G. where U1. Then there exist i P {1. Irreducible representations r1 . 6. Let v P U c and g P G. 7. . v]X xPG xPG (2) It is easy to prove that U c is a subspace of V. v gx) ˆ (ux. v g] ˆ [ug À1 . U ˆ sp ((1. r} and g P G such that ug Tˆ u for some u P Ui (otherwise vg ˆ v for all v P CG). r2 : 1r1 ˆ ar1 ˆ (1). Then V ˆ U È W.6). where x 2 ˆ y 2 ˆ 1. This means that Ui is a faithful irreducible CG-module. (x. (1) It is straightforward to verify for [ . r2 . De®ne K ˆ fx P G: vx ˆ v for all v P Ui gX Check that K is a normal subgroup of G. y)}. Chapter 9 1. bi r2 ˆ (ù i ). also K Tˆ G since g P K. Ur are irreducible CG-submodules of CG. v g] ˆ (ugx. Therefore r is irreducible. This is a contradiction. ux) . 1). so [u. Let CG ˆ U1 È . hence also (gr)(hr) ˆ (hr)(gr) for all g. and so U c is a CG-submodule of V. . For example.Chapter 9 407 commute with each other. vx) ˆ [u. (x. Since G a is simple. . . . and W is a CG-submodule of V by part (2). . 1). we must therefore have K ˆ {1}. Let C2 ˆ ka: a2 ˆ 1l. Irreducible representations r1 . 1r2 ˆ (1). È Ur. r4 : 3 . if u Tˆ 0 then (ux. v] ˆ 0 since ug À1 P U X Therefore vg P U c. We know that the regular CG-module CG is faithful (Proposition 6. Then for all u P U. ] the axioms of a complex inner product. bi r3 ˆ (ù2i )X Let C2 3 C2 ˆ {(1. so there is no CG-submodule W of V with V ˆ U È W. 0.

Yes: if r ˆ 2. let gj generate Cn j . Let z ˆ gPG g. 1). .408 Representations and characters of groups gr1 ˆ (1) for all g P C2 3 C2 . Check that A4 ˆ B2 ˆ I. . For 1 < j < r. y j )r2 ˆ (À1) j . g2 and g1 g2 all have order 2. n1 ˆ 2. Also bÀ1 (a ‡ aÀ1 )b ˆ aÀ1 ‡ a. Hence z P Z(CG). 3. n2 ˆ 3. (d) Let C3 ˆ kx: x 3 ˆ 1l and let ù ˆ e2ðia3 . y): x 4 ˆ y 4 ˆ 1l.3). hence ó is reducible (Corollary 9. 7. then M ˆ ëI for some ë P C. so is not cyclic. BÀ1 AB ˆ AÀ1 when A ˆ ar and B ˆ br. (a) Let Cn ˆ kx: x n ˆ 1l. (1. y j ) 3 (À1) i . (x. r. b) 3 Àù 0 0 Àù . Check that     0 ù ù 0 r: (x. b3 (b) r: a 3 À1 0 0 À1 gives a faithful irreducible representation of D8 (see Example 5. g 22 ) 3 (å11 å22 ) is a faithful representation of degree 1 . (a) Clearly a commutes with a ‡ aÀ1 .3). y j )r3 ˆ (À1) i . € 5. (x i . 4. 3 C n r of degree r. y 2 ) then g1 . Let C4 3 C4 ˆ k(x. (c) The centre of C2 3 D8 is isomorphic to C2 3 C2 .     0 1 1 0 . Therefore Proposition 9. and let å j ˆ e2ðia n j . similarly for ó. 6. Notice that the matrix   5 À6 4 À5 commutes with gó for all g P G. Then xz ˆ z ˆ zx for all x P G.16 shows that C2 3 D8 has no faithful irreducible representation. then i i i i ó : ( g 11 . Hence r gives a representation. so b commutes with a ‡ aÀ1 . Then r: x j 3 (e2ði ja n ) is a faithful irreducible representation of Cn . a) 3 . (x i . y j )r4 ˆ (À1) i‡ j X 2. If M(gr) ˆ (gr)M for g ˆ a and for g ˆ b.5(2)). and the result follows from Proposition 9. g irr ) 3 f d e F ir år 0 0 is a faithful representation of C n1 3 . (b) If g1 ˆ (x 2 . Then H i I å11 f g i FF g r: ( g 11 . 1) and g2 ˆ (1. X X X . (a) r: (x i . we cannot have (g1 g2 )ó ˆ g1 ó ˆ g2 ó ˆ (À1). Hence r is irreducible (Corollary 9. Since (g1 g2 )ó ˆ (g1 ó)(g2 ó) for all representations ó.14. (x i . (b) Check that w(a ‡ aÀ1 ) ˆ Àw for all w P W.

w0 ). w1 ) are CG-submodules of CG. sp (v2 . b 3 (1) r1 : a 3 (1). Moreover U4  U5. U3 and U4 . b 3 (1) r3 : a 3 (À1). It is irreducible (see for example Exercise 8. as in Example 10. w3 ). As in Example 5. Let U4 ˆ sp (v1 . v3 ˆ 1 À ia À a2 ‡ ia3 Chapter 10 (compare the solution to Exercise 2).b3 X 0 Ài 1 0 . 409 € 1. so |G|u ˆ u( gPG g) ˆ € ( gPG g)u P V. U5 ˆ sp (v3 . where Ui ˆ sp (ui ) (0 <€ 3) and u1 . Let u1 ˆ 1 ‡ a ‡ a2 ‡ a3 À b À ab À a2 b À a3 b. so U ˆ sp (u) of for some u. Then V is a trivial CG-submodule of CG. b 3 (À1) r2 : a 3 (À1). Let G ˆ kx: x 4 ˆ 1l. We decompose CG as a direct sum of irreducible CG-submodules. Theorem 10. namely U0 . v1 ˆ 1 ‡ ia À a2 À ia3 . b 3 (À1) 2 3 2 3 i 0 0 1 r4 : a 3 . For 0 < j < 3. u2 ˆ 1 À a ‡ a2 À a3 ‡ b À ab ‡ a2 b À a3 b. w1 ). Then. u3 ˆ 1 À a ‡ a2 À a3 À b ‡ ab À a2 b ‡ a3 bX 4. 2. sp (v2 . w0 ) ˆ U0 È U1 .Chapter 10 gives a representation of C3 3 D8. We have sp (v0 . v2 ˆ 1 À a ‡ a2 À a3 . U1 . w2 ) and sp (v3 . let wj ˆ bv j. U2 . Let V ˆ sp ( gPG g). u2 . Let v0 ˆ 1 ‡ a ‡ a2 ‡ a3 . namely V. Thus U ˆ V. w3 3 v3.4). while u0 ˆ gPG g. U4 and U5 are irreducible CG-modules.4) and faithful. w2 ) ˆ U2 È U3 . w3 ). u3 are as in the solution to i< Exercise 3.8(2). Now suppose that U is an arbitrary trivial CG-submodule € CG. since there is a CG-isomorphism sending v1 3 w1. sp (v1 . the subspaces sp (v0 . Then ug ˆ u for all g P G. and so CG has exactly one trivial CGsubmodule.5 now shows that there are exactly ®ve non-isomorphic irreducible CG-modules.5(2) (or see Exercise 8. Therefore every irreducible representation of D8 over C is equivalent to precisely one of the following: r0 : a 3 (1). Then CG ˆ sp (1 ‡ x ‡ x 2 ‡ x 3 ) È sp (1 ‡ ix À x 2 À ix 3 ) È sp (1 À x ‡ x 2 À x 3 ) È sp (1 À ix À x 2 ‡ ix 3 )X 3.

17.8(2). Therefore we have constructed in®nitely many CG-submodules Im öë of the required form. 1. uö2 ˆ bu (u P U3 ). by Corollary 11. . the dimensions are 1. dim (HomCG (V.2. . 14 22 and 13 3 (where 112 means twelve 1s. Let u1 ˆ 1 À ia À a2 ‡ ia3 . For each g P G. moreover. by Theorem 11. Let W: U1 3 U2 be a CG-isomorphism. Then ö1 . b) such that X a  Y b . dim (HomCG (V.5(2) or by Exercise 8. Hence the answer for D12 is 14 22 . W)) ˆ iˆ1 d i e i . 2. . v2 3 u2 . and similarly the number of integers b with Y b  V i is €k e i . v n be the natural basis of V. Let v1 . Hence by Corollary 11. since the sum U1 ‡ U2 is direct. CG). Thus U1  Im öë . This. Then {ö g : g P G} gives a basis of HomCG (CG. For ë P C. 4.12.6. 18 2 means eight 1s and one 2. in turn. V is irreducible. not all the dimensions are 1 (see Proposition 9. È Ys. Also. CG) (compare the proof of Proposition 11. Chapter 11 1. 2. 5. . u2 ) is a CG-submodule of CG which is isomorphic to V. 6. . Then sp (v1 ‡ . De®ne W1 and W2 by rW1 ˆ v1 r. . W)) is equal to the number of ordered pairs (a. V i )) ˆ d i . 18 2.8). Then sp (u1 .18). either by the method of Example 5.4. A CG-isomorphism is given by v1 3 u1 .8. . by the proof of Proposition 11. Therefore. ‡ v n ) is the unique trivial CG-submodule of V (compare Exercise 10. ö2 by uö1 ˆ u. Let V ˆ X1 È .6. . Let v1 . de®ne ö g : CG 3 CG by rö g ˆ gr (r P CG).5)(3) and Proposition 11. w2 be the basis of U3 described in Example 10. u P Ker öë D u ‡ ëuW ˆ 0 D u ˆ 0.4.1). D12 has at least two inequivalent irreducible representations of degree 2. Since G is non-abelian. and so on). rW2 ˆ w2 r (r P CG). Compare Example 11.13 to see that the possible answers are 112 . de®ne the function öë : U1 3 V by öë : u 3 u ‡ ëuW (u P U 1 )X Then öë is easily seen to be a CG-homomorphism. By Exercise 5. Then by (11.410 Representations and characters of groups 5. U)) ˆ 1. U3 ). where each Xa and each Yb is an irreducible CG-module. 6. It will be shown later (Exercises 15. dim (HomCG (V. equals k ˆ iˆ1 jf(a. È Xr and W ˆ Y1 È . u2 ˆ b À iab À a2 b ‡ ia3 b. . ö2 is a basis of HomCG (U3. 3. It is easy to check that if ë Tˆ ì then Im öë Tˆ Im ö ì . Then W1 . de®ne ö1 . b): X a  Yb  Vi gjX Now the number of integers a with X a  V i is dim (HomCG (V . Hence. W2 is a basis of HomCG (CG.3. .3) that 18 2 cannot occur. .

thus ghÀ1 P CG (x). a ‡ a3 . a3 g. l (unordered). a2 bg. There are ®ve choices for i. We have Q8 ˆ ka. An element x of cycle-shape (5) has CS6 (x) ˆ kxl (note that |x S6 | ˆ 144 and use Theorem 12.8). k (unordered). Then gx ˆ xg and hx ˆ xh. fa2 g. h P CG (x). (i k)( j l ) and (i l )( j k). then we can make three permutations ( j k)(l m). ab ‡ a3 bX 7. (c) Every element of (1 2 3)(4 5 6) G has the form (1 i j)(k l m). x A6 Tˆ x S6 . b2 ˆ a2 . the conjugacy classes of A5 have sizes 1. Each choice gives exactly two 3-cycles. j. For elements g of other cycle-shapes. and a basis of Z(CQ8 ) is 1. 2) (5) 90 120 144 (23 ) 15 (32 ) (4. l distinct. 2) (6) 40 90 120 4. The centralizer CG ((1 2)) consists of all elements x and (1 2)x. Hence by Proposition 12.18(2). 12. The elements of (1 2)(3 4)(5 6) G have the form (1 i)( j k)(l m).8. fab. in agreement with Theorem n 12. with i. 4 . Therefore CG (x) is a subgroup of G. g A6 ˆ g S6 . k.Chapter 12 Chapter 12 411 1. 3 ˆ 15. then four choices for j. fa. 2. 20. The sizes of the conjugacy classes of S6 are given in the following table: Cycle-shape (1) Class size 1 (2) 15 (3) 40 (22 ) 45 (4) (3. If H is a normal subgroup of A5 then j Hj divides 60. The class equation gives . (1 2)(3 4) G consists of all permutations of the form (i j)(k l ) with i. (a) (1 2) G ˆ {(i j): 1 < i . therefore A5 is simple. This gives 5 . and 1 P H. b: a4 ˆ 1. namely (i j)(k l ). j < n} and this set has size (2 ). Assume that g. so hÀ1 x ˆ xhÀ1 and ghÀ1 x ˆ gxhÀ1 ˆ xghÀ1 . Also 1x ˆ x1. By Example 12. n k. 2) from the remaining numbers. n G (b) (1 2 3) consists of all 3-cycles (i j k). If z P Z(G) then zg ˆ gz for all g P G. l. where x is a permutation ®xing 1 and 2. a3 bg. Thus |CG ((1 2))| ˆ 2´(n À 2)!. There are (3 ) choices for the numbers i. Hence |(1 2)(3 4)(5 6) G | ˆ 5 . There are (4 ) choices for the numbers i. Now the required result follows from Theorem 12.8 (since (2 ) ˆ n3a(2X(n À 2)3)). 12. bÀ1 ab ˆ aÀ1 l. fb. k. and three permutations for each choice. ( j l )(k m) and ( j m)(k l ) of cycle shape (2. m distinct. Note that x P CG (g) D x P CG ( gz). a2 . b ‡ a2 b. then we can make two different 3-cycles (k l m) and (k m l ) from the remaining numbers. Hence j Hj ˆ 1 or 60. The conjugacy classes of Q8 are f1g. so 1 P CG (x). 5. n 3. 6. j. 2 ˆ 40 elements in all. 15. namely (i j k) and (i k j).17. j. There are ®ve choices for i. so zx ˆ xz and z P CG (x). j. and H is a union of conjugacy classes of A5 .

a4 À1 2 b. Then zr ˆ ëI for some ë P C. by Proposition 9. The characters ÷ i of r i (i ˆ 1. Thus.412 Representations and characters of groups jGj ˆ j Z(G)j ‡ ˆ xiP Z(G) a jx G jX i (a) For xi P Z(G). (a) For all g. a2 b. a5 À1 0 a2 . by Proposition 9. Let r be a representation with character ÷. we have ÷((1 2)) ˆ 5 and ÷((1 6)(2 3 5)) ˆ 2. 3. . Let r be a representation with character ÷. and Im ä is a subgroup of the . We have now proved that gr ˆ ëI for some ë P C if and only if g P Z(G). and hence ÷(zg) ˆ ë÷(g). 2. The irreducible characters ÷1 . 2) are as follows: 1 ÷1 ÷2 2 2 a3 2 0 a. det ((gh)r) ˆ det ((gr)(hr)) ˆ det (gr) det (hr). h PG. If ÷ is a non-zero character which is a homomorphism. p2 divides a |Z(G)|. |G| > p3 . and hence g P Z(G) since r is faithful. for all g in G. a3 b. Let C4 ˆ kx: x 4 ˆ 1l. Conversely. Moreover.8 and (12. a5 b 0 0 0 À2 Also Ker r1 ˆ {1. then ÷(1) ˆ ÷(12 ) ˆ (÷(1))2 . (b) GaKer ä  Im ä by Theorem 1. then p2 divides |x G | for all i xi P Z(G).14. Chapter 13 1. then ( gr)(hr) ˆ (hr)(gr) for all h P G. Hence p divides |Z(G)|.11(1). This is a contradiction. . so Z(G) Tˆ f1g. in addition. a2 . Hence g 3 (det (gr)) is a representation of G over C of degree 1. so ÷(1) ˆ 1.10. 7. a3 } and Ker r2 ˆ {1. Since ÷(g) ˆ |®x (g)|. (zg)r ˆ (zr)(gr) ˆ ë(gr). so ë m ˆ 1. If g P Z(G) then gr ˆ ëI for some ë P C. i (b) If no conjugacy class of G has size p. a4 b ab.14. ÷4 of C4 are as follows: 1 ÷1 ÷2 ÷3 ÷4 1 1 1 1 x 1 i À1 Ài x2 1 À1 1 À1 x3 1 Ài À1 i We have ÷reg ˆ ÷1 ‡ ÷2 ‡ ÷3 ‡ ÷4 . then by the class equation. I ˆ 1r ˆ z m r ˆ (zr) m ˆ ë m I. a i i Therefore p divides |x G |. . . 5. |x G | divides pn and |x G | Tˆ 1 by Theorem 12. and so ä is a linear character of G.9). If. 4. The required result now follows from Theorem 13. if gr ˆ ëI for some ë P C. 6. a4 }.

. Using Proposition 14. (c) Im ä is a ®nite subgroup of Cà . g2 k of CG so as to obtain a basis B in which g and gx are adjacent for all g P G. øi ˆ ‡ ‡0‡ ‡ ˆ 1X 24 4 8 4 Hence ø is irreducible by Theorem 14. . a2 b 0 0 0 ab. so ÷ i (x) Tˆ ÷ i (1) for some irreducible character ÷ i of G.3 1. The required result now follows from Exercise 7.19. By Exercise 1. 1 3 . Say there are r entries 1 and s entries À1.7.1 (À1)(À1) (À1)(À1) hø. so ÷( g) ˆ r À s  r ‡ s ˆ ÷(1) mod 4. 8. ÷i ˆ ‡ ‡0‡ ‡ ˆ 2. det ([x]B ) ˆ (À1) k ˆ À1. If s is odd then ä( g) ˆ À1. 2. (À1) (À1)(À1) h÷. 3 (À1) . 3). As x Tˆ 1. Also À1 P Im ä. And if s is even then Às  s mod 4. Then I H 0 1 g f1 0 g f g f 0 1 [x]B ˆ f gX g f 1 0 e d FF F 0 0 There are k blocks and since k is odd. Thus ä(x) ˆ À1. The values of these characters are as follows: Conjugacy class ÷1 ÷2 ÷3 1 2 2 2 a2 À2 À2 2 a. De®ne ä as in Exercise 7.Chapter 14 413 multiplicative group Cà of non-zero complex numbers. 3 (À1)(À1) 3 . a3 b 0 0 À2 . 10. Let r be the regular representation of G. Therefore GaKer ä is abelian. Let ÷ i be the character of r i (i ˆ 1. which is abelian. It is easy to check that {g P G: ä(g) P H} is a normal subgroup of G of index 2. we obtain 3 . We may choose a basis B of V so that [g]B is diagonal with all diagonal entries Æ1 (see (9. G has an element x of order 2. øi ˆ ‡ ‡0‡ ‡ ˆ 0. 2. and G has a normal subgroup of index 2 by Exercise 7.20).20 (but ÷ is not).8. (0 1 ). . we have ÷reg (x) Tˆ ÷reg (1) (see Proposition 13. by Exercise 1. 10 Chapter 14 1. Hence Im ä contains a subgroup H of index 2. 9. 24 4 8 4 3.5(2). . Order the natural basis g1 . a3 0 0 0 b. 24 4 8 4 3 . and de®ne the character ä as in Exercise 7. Let V be a CG-module with character ÷. hence is cyclic. 3 (À1)(À1) h÷. so Im ä has even order. by Theorem 13.10)).

by Proposition 13. (À1) . ÷1 l Tˆ 0. øl ˆ 4. ÷l ˆ ÷(1). ÷3 i ˆ 1(19 . 1 ‡ 3 . we have k÷.21. As ÷ Tˆ ÷1 . ø is not a character of G. 1 ‡ 2 . No: let G ˆ C2 and ÷ ˆ ÷reg . or exactly one of the di is 2. Then 1 ˆ h÷. 6 h÷. then either exactly four of the di are 1. the regular character of C2 . (À2) . Chapter 15 1X h÷. 2 ‡ 0 ‡ 2 . By the method used in the solution to Exercise 1. hence r and ó are equivalent. by Theorem 14. If kø. If |G| ˆ 12 . the rest are 0.17 shows that ÷ is reducible. if G ˆ Z(G) ‘ Z(G)x then G ˆ Z(G). 5. we obtain ø 1 ˆ 1 ÷1 ‡ 1 ÷2 ‡ 1 ÷3 . Since the coef®cient of ÷2 is a negative integer. The representations r and ó have the same character. if x P G then the subgroup generated by x and the elements of Z(G) is abelian (since the elements of Z(G) commute with powers of x). ÷i ˆ 1 ˆ ÷reg ( g)÷( g)X jGj gPG But ÷reg (g) is |G| if g ˆ 1 and is 0 if g Tˆ 1. 6. Hence. (À2)(À1)) ˆ 7X 6 h÷. 4. Recall that hø. ÷2 i ˆ 1(19 . Therefore the centre of a group G never has index 2 in G. Theorem 14. 1) ˆ 2. r1 and r2 are equivalent. (À1)(À1) ‡ 2 . 6 6 3 ø 2 ˆ 1 ÷1 À 1 ÷2 . 2 or 3.4 and Theorem 14. ÷1 i ˆ ÷( g)X jGj gPG Since ÷(1) . and this gives the required matrix T.414 Representations and characters of groups By Theorem 14. 4. This follows at once from Exercise 11. Hence k÷reg . €k 7. ÷1 i ˆ 1(19 . 6 Hence ÷ ˆ 2÷1 ‡ 3÷2 ‡ 7÷3 . (À2) . 1) ˆ 3. 2 2 ø 3 ˆ 1 ÷1 ‡ 1 ÷2 À 1 ÷3 X 3 3 3 3. 1 ‡ 3 . Every abelian group of order 12 has 12 conjugacy classes. 3.21. Let ÷1 be the trivial character of G.24. Since all the coef®cients here are non-negative integers. 0 and by hypothesis ÷(g) > 0 for all g P G. We have h÷reg .2. We ®nd that ø ˆ À÷2 ‡ ÷3 ‡ ÷5 ‡ 2÷6 . øi ˆ iˆ1 d 2 . but r3 is not equivalent to r1 or r2. it follows that ÷ is a character of S3 . 2. 8. (a) For all groups G. Hence if kø. then i exactly a of the integers di are 1 and the rest are 0. øl ˆ a where a ˆ 1.

(b) Since the number of irreducible representations is equal to the number of conjugacy classes. Then iˆ1 ÷ i ( g 2 )÷ i ( g 4 ) ˆ 0 € gives ÷3 ( g2 ) ˆ 1. similarly ÷4 ( g3 ) ˆ (À1 ‡ 5)a2. together with the relation €4 €4 iˆ1 ÷ i (1)÷ i ( g 4 ) ˆ 0. so there cannot be as many as 9 conjugacy classes in total. and if G ˆ A4 then G has 4 conjugacy classes (see Example 12.Chapter 16 415 and G is non-abelian. 1). (x. if G ˆ D12 then G has 6 conjugacy classes (see (12. the remaining conjugacy classes have size at least 2.1) (1. 4 ÷ i (1)÷ i ( g 2 ) ˆ 0 gives iˆ1 p p ÷4 ( g 2 ) ˆ (À1 À 5)a2. similarly ÷3 (g3 ) ˆ 1. iˆ1 Because g4 has order 2. (1. at most 4 conjugacy classes of G have size 1 (see (12. The last row of the character table is (cf.g.18(1)). Exercise 9.5(2)) g1 ÷5 2 g2 À2 g3 0 g4 0 g5 0 3. If G is abelian (e. Therefore. 1) 1 1 À1 À1 (1. 1) ÷1 ÷2 ÷3 ÷4 1 1 1 1 (x. Corollary 13. Let C2 3 C2 ˆ {(1.9)). it follows from the solution to Exercise 11. 2 since 4 ( ÷ i (1))2 ˆ 10. ÷4 (1) are 1. y) 1 À1 1 À1 (x.2 and part (a) that G has 4. y): x 2 ˆ y 2 ˆ 1}. (x.10. Finally. . so |Z(G)| < 4. 6 or 12 conjugacy classes. y). then |Z(G)| divides 12 and |Z(G)| Tˆ 6 or 12. G ˆ C4 3 C3 ) then G has 12 conjugacy classes. The complete character table of G is gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 g1 10 1 2 1 2 g2 5 1p (À1 ‡ 5)a2 1p (À1 À 5)a2 g3 5 1p (À1 À 5)a2 1p (À1 ‡ 5)a2 g4 2 1 0 À1 0 € The two unknown degrees ÷3 (1). Example 16. 1). The character table of C2 3 C2 is (cf.12)). y) 1 À1 À1 1 2. gives the values on g4 . Chapter 16 1.

p Hence æ ˆ (À1 Æ i 7)a2. (b) The column which corresponds to the conjugacy class containing gÀ1 2 has values which are the complex conjugates of those in the column of g2 (see Proposition 13.9(3)). fa2 g. (b) G9 ˆ f1. 5.) Chapter 17 1. and if det C ˆ Àdet C then det C is purely imaginary. G9abg  C2 3 C2 . Let g P G. C t C is the k 3 k diagonal matrix whose diagonal entries are |CG ( gi )| (1 < i < k). a2 bg and fab.13 to see quickly that bÀ1 ab ˆ a2 . Hence the linear characters of G are gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 1 8 1 1 1 1 a2 8 1 1 1 1 a 4 1 1 À1 À1 b 4 1 À1 1 À1 ab 4 1 À1 À1 1 (c) Using the column orthogonality relations. Therefore det C ˆ Ædet C. fa. this is a different column of the character table of G.9(3)). Use Proposition 12. p If G ˆ C3 then det C ˆ Æi3 3. (a) €5 Representations and characters of groups iˆ1 ÷ i ( g 1 )÷ i ( g 2 ) € ˆ 0 gives 3 ‡ 3æ ‡ 3æ ˆ 0. we have iˆ1 ÷ i ( g)÷ i ( g) ˆ jC G ( g)j. By the column orthogonality relations. G9b. since æ is non-real. if det C ˆ det C then det C is real. 2. The matrix C is obtained from C by rearranging the columns (see Proposition 13. This number is equal to |G| if and only if CG (g) ˆ G. (The sign depends upon the ordering of rows and columns. a3 g. Hence  jdet Cj2 ˆ jC G ( g i )j. 6. which occurs if and only if g P Z(G). By the column orthogonality relations applied to the column €k corresponding to g.416 4.1. and 5 ÷ i ( g 2 )÷ i ( g2 ) ˆ 7 iˆ1 gives 3 ‡ 2ææ ˆ 7. a3 bg. fb. (a) The conjugacy classes of Q8 are f1g. (a) Using the relations. G9a. every element of G has the form am bn with . It is easy to see that a7 ˆ b3 ˆ 1. the last irreducible character of G is 1 ÷5 2 a2 À2 a 0 b 0 ab 0 The character table of Q8 is the same as that of D8. a2 g and GaG9 ˆ fG9. The character table of C2 3 C2 is given in the solution to Exercise 16.

Now consult the solution to Exercise 11. all of these are irreducible characters by Proposition 17. a4 g. so G is certainly not simple.2 to see that there are 3. ÷3 ˆ ÷ 2 . ÷6 ˆ ö÷. (b) The conjugacy classes of G are f1g. and the class sizes j g G j come from the equations i jGj ˆ jC G ( g i )jj g G j (Theorem 12. In the character table below. If there are 12. we note that a is not conjugate to aÀ1 . hence jGj < 21. we obtain 1 ÷4 ÷5 3 3 a á á a3 á á b 0 0 b2 0 0 p where á ˆ (À1 ‡ i 7)a2. (c) First. But a has order 7 and b has order 3.11. 4. The number of linear characters of G divides jGj by Theorem 17. fa m b: 0 < m < 6g and fa m b2 : 0 < m < 6g.8). so 21 divides jGj by Lagrange's Theorem. If G has 3 or 4 linear characters then jGaG9j ˆ 3 or 4 and again G is not simple as G9 v G. fa3 . then G is abelian (see Proposition 9. a5 . 3. therefore for some irreducible character ÷. i gi |CG ( gi )| | gG | i ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 g1 12 1 1 1 1 1 2 2 g2 4 3 1 Ài À1 i 0 0 g3 4 3 1 i À1 Ài 0 0 g4 6 2 1 1 1 1 À1 À1 g5 6 2 1 À1 1 À1 À1 1 g6 12 1 1 À1 1 À1 2 À2 . we have ÷(a) Tˆ ÷(a) (see Corollary 15. 0 < n < 2. ÷4 ˆ ÷2 ÷3 . a6 g. fa.18).6). Hence ÷4 and ÷5 must be complex conjugates of each other. G9 ˆ kal.14. Applying the column orthogonality relations. The centralizer orders are obtained by using the orthogonality relations. Therefore jGj ˆ 21. ÷5 ˆ ö. so we get three linear characters of G: gi |CG ( gi )| ÷1 ÷2 ÷3 1 21 1 1 1 a 7 1 1 1 a3 7 1 1 1 b 3 1 ù ù2 b2 3 1 ù2 ù where ù ˆ e2ðia3 . 4 or 12 linear characters. ÷2 ˆ ÷. we have ÷1 ˆ 1 G . a2 .Chapter 17 417 0 < m < 6. To ®nd the two remaining irreducible characters ÷4 and ÷5 .

and inequivalent (consider the character values on a2 ). 22 ‡ 4 . (a) Check that the given matrices satisfy the relevant relations: 2 32 n 2 3n 2 3 2 32 å 0 1 0 å 0 0 1 ˆ . ha2 . (b) The given representations. for å ˆ e2ði ka2 n with 0 < k < n À 1.11). For å ˆ e2ði ra2 n .3. Now the sum of the squares of the degrees of the irreducible representations we have found so far is (n À 1) . We get further representations by     ç 0 0 1 a3 . ha2 . they are not equivalent to any representation found earlier. . . so jGaG9j ˆ 4 and there are four representations of degree 1 (see Theorem 17.b3 .4. by Exercise 8. (a) Check that the given matrices satisfy the relevant relations.418 Representations and characters of groups D8 ˆ Ker ÷1 . Also G9 ˆ kbl. (b) The given representations. (a) Check that the given matrices satisfy the relevant relations. Note that b2 does not belong to the kernel of any of these representations. . are irreducible (by Exercise 8. these representations are irreducible and inequivalent. 2. ha2 i ˆ Ker ÷2 ’ Ker ÷3 . are irreducible (by Exercise 8. . since b2 is in the kernel of each of these representations. note that the structure of GaG9 depends upon whether n is even or odd. so we have obtained all the irreducible representations. ˆ .4). we get n À 1 irreducible representations. 22 ‡ 2n . 0 çÀ1 1 0 where ç is any (2n)th root of unity in C. Moreover.4). . (For further details on the representations of degree 1. by Theorem 11. The sum of the squares of the degrees of the irreducible representations we have found is n . For ç ˆ e2ði ja2 n with 1 < j < n À 1. hai ˆ Ker ÷2 . so jGaG9j ˆ 2n and there are 2n representations of degree 1. f1g ˆ Ker ÷5 X 5. (b) The representations in part (a) are irreducible unless å ˆ Æ1. for å ˆ e2ði ja n with 0 < j < n À 1. n À 1. abi ˆ Ker ÷4 . bi ˆ Ker ÷3 . 12 ˆ 6n. no two of which are equivalent. with r ˆ 1. 8. since they have distinct characters.12. Moreover G9 ˆ ka2 l.) 7. The normal subgroups of D8 are 6. Example 1. 12 ˆ 4nX Hence we have found all the irreducible representations. 0 å À1 0 1 0 å À1 ån 0 32 3À1 2 3À1 2 3 2 0 1 å 0 0 1 å 0 X ˆ ån 0 0 å À1 ån 0 0 å À1 Hence we have representations of T4 n (cf.4) and inequivalent (their characters are distinct). see the solution to Exercise 18.

The character table of D8 is as shown.28(2). Then ð takes the following values: 1 ð 4 a2 0 a 0 b 2 ab 0 Hence ð ˆ ÷1 ‡ ÷3 ‡ ÷5 .) 2. ù2 ‡ ùÀ2 ˆ ù4 ‡ ùÀ4 ˆ À1. G9 ˆ ka2 . We have now found all the irreducible representations. Let ù ˆ e2ðia6 . so we get four representations of degree 1. Character table of D12 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 1 12 1 1 1 1 2 2 a3 12 1 1 À1 À1 À2 2 a 6 1 1 À1 À1 1 À1 a2 6 1 1 1 1 À1 À1 b 4 1 À1 1 À1 0 0 ab 4 1 À1 À1 1 0 0 . where we took b to be a different re¯ection.3(3) or Section 18. 12 ˆ 8nX Chapter 18 1. using Section 18. the character table of D12 is as shown. since the sum of the squares of the irreducible representations given above is n .3. b2 l and GaG9  C2 3 C2.) Regarding D8 as the symmetry group of a square. Hence. Then ù ‡ ùÀ1 ˆ 1. (Compare Example 14. Character table of D8 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 1 8 1 1 1 1 2 a2 8 1 1 1 1 À2 a 4 1 1 À1 À1 0 b 4 1 À1 1 À1 0 ab 4 1 À1 À1 1 0 (See Example 16. 22 ‡ (n À 1) .Chapter 18 419 Finally.3. 22 ‡ 4 . take b to be a re¯ection in a diagonal of the square.

fa2 r g. ka3 l ˆ Ker ÷6 and {1} ˆ Ker ÷5 . kal ˆ Ker ÷2 . The remaining four irreducible characters of G are linear. abl ˆ Ker ÷4 . far . aÀ r g(1 < r < n À 1). a2 r b2 g. fa2 r b. 4. for 0 < r < n À 1. fa2 j b: 0 < j < n À 1g. bl ˆ Ker ÷3 . then GaG9  C2 3 C2 and the linear characters are gi ÷1 ÷2 ÷3 ÷4 1 1 1 1 1 an 1 1 1 1 ar (1 < r < n À 1) 1 1 (À1) r (À1) r b 1 À1 1 À1 ab 1 À1 À1 1 Note that T4  C4 . The n ‡ 3 conjugacy classes of G are f1g.6 as follows: gi |CG ( gi )| øj 1 4n 2 an 4n 2(À1) j ar (1 < r < n À 1) 2n ù rj ‡ ùÀ rj b 4 0 ab 4 0 where ù ˆ e2ðia2 n . ka2 . If n is odd. fa2 j‡1 b: 0 < j < n À 1gX We get n À 1 irreducible characters ø j (1 < j < n À 1) of G from Exercise 17. 3.7 gives us n irreducible characters ø k (0 < k < n À 1). a2 r‡1 b2 gX We have G9 ˆ hbi and GaG9 ˆ hG9ai  C2 n . T8  Q8 and T12 is the Example in Section 18.4. fa2 r‡1 . Exercise 17. fan g. as shown. The 3n conjugacy classes of G are. ka2 l ˆ Ker ÷3 ’ Ker ÷4 . ka2 . a2 r‡1 b. . Hence we get 2n linear characters ÷ j (0 < j < 2n À 1). then GaG9 ˆ hG9bi  C4 and the linear characters are gi ÷1 ÷2 ÷3 ÷4 1 1 1 1 1 an 1 À1 1 À1 ar (1 < r < n À 1) 1 (À1) r 1 (À1) r b 1 i À1 Ài ab 1 Ài À1 i If n is even.420 Representations and characters of groups Seven normal subgroups of D12 are G ˆ Ker ÷1 .

Chapter 18 Character table of U6 n gi |CG ( gi )| ÷j (0 < j < 2n À 1) øk (0 < k < n À 1) Note: ù ˆ e2ðia2 n . Character table of V8 n gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 øj (0 < j < n À 1) öj (1 < j < n À 1) 1 8n 1 1 1 1 2 2 a2 r‡1 (0 < r < n À 1) 8n 4n 1 1 1 1 À2 2 1 1 À1 À1 b2 a2s a2s b2 (1 < s < (n À 1)a2) 4n 4n 1 1 1 1 ù4 js ‡ùÀ4 js ù2 js ‡ùÀ2 js 1 1 1 1 Àù4 js ÀùÀ4 js ù2 js ‡ùÀ2 js b 4 1 À1 1 À1 0 0 ab 4 1 À1 À1 1 0 0 ù2 j(2 r‡1) ÀùÀ2 j(2 r‡1) ù j(2 r‡1) ‡ùÀ j(2 r‡1) Note: ù ˆ e2ðia2 n . aÀ2s b2 g(1 < s < (n À 1)a2). 5. . For example. aÀ2 rÀ1 b2 g(0 < r < n À 1). n characters ø j (0 < j < n À 1) of degree 2.8. we get four linear characters ÷1 . k ˆ 1 or 3gX a2 r 6n ù2 jr 2ù2 kr a2 r b 3n ù2 jr Àù2 kr a2 r‡1 2n ù j(2 r‡1) 0 421 Using Exercise 17. fa2s b2 . . 422. . fa2s . and a further n À 1 characters ö j (1 < j < n À 1) of degree 2. fa2 r‡1 . and faj bk : j odd. as shown below. ÷4 . fb2 g. . k ˆ 1 or 3g. aÀ2s g. faj bk : j even. The 2n ‡ 3 conjugacy classes of G are f1g. the character table of V24 is given at the top of p. U12  T12 and U18  D6 3 C3 . Observe that U6  D6.

15 and (14. öl ˆ kø. By Proposition 15. The result now follows from Proposition 13. Therefore k÷ n . k÷ø. k÷ø. Hence ö( g) ˆ ö(1) for all irreducible characters ö for which k÷ n .422 Representations and characters of groups Character table of V24 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ø0 ø1 ø2 ö1 ö2 1 24 1 1 1 1 2 2 2 2 2 b2 24 1 1 1 1 À2 À2 À2 2 2 a 12 1 1 À1 À1 p0 ip 3 Ài 3 1 À1 a3 12 1 1 À1 À1 0 0 0 À2 2 a5 12 1 1 À1 À1 0 p Ài 3 p i 3 1 À1 a2 12 1 1 1 1 2 À1 À1 À1 À1 a2 b2 12 1 1 1 1 À2 1 1 À1 À1 b 4 1 À1 1 À1 0 0 0 0 0 ab 4 1 À1 À1 1 0 0 0 0 0 Chapter 19 1X h÷ø. Using Proposition 19. 1 G l ˆ k÷.13). 4. ö S ˆ ø1 ‡ ø3 . öl Tˆ 0. Then wg ˆ w for all w P V  .5 there is an irreducible character ø of G such that ø(g) Tˆ ø(1). ÷öl. Note that (1 2 3 4 5)2 is conjugate to (1 3 4 5 2) in A5 . . øl ˆ 0. Since ÷ is not faithful. by Exercise 1. Let n be an integer with n > 0.14 we obtain 1 ÷S ÷A öS öA 15 10 6 3 (1 2 3) 0 1 0 0 (1 2)(3 4) 3 À2 2 À1 (1 2 3 4 5) 0 0 1 p (1 3 4 5 2) 0 0 1 p (1 ‡ 5)a2 (1 À 5)a2 Then ÷ S ˆ ø1 ‡ ø2 ‡ 2ø3 . öi ˆ 1 ˆ 1 ˆ ÷( g)ø( g)ö( g) ˆ ÷( g)ø( g)ö( g) ˆ h÷. øl. 2. ö A ˆ ø4 X . ÷ A ˆ ø2 ‡ ø4 ‡ ø5 . øöiX jGj gPG jGj gPG Similarly. 3. there exists 1 Tˆ g P G with vg ˆ v for all v P V. .  V (n factors). Let V be a CG-module with character ÷.

bÀ1 ab ˆ aÀ1 l.1(3). ÷3 ˆ ÷2 . ÷6 ˆ ÷5 and ÷7 ˆ ÷2 ÷5 . ÷ S as ÷4 and ÷ A as ÷2 . Since G has seven conjugacy classes. these characters are irreducible. b) 36 18 12 18 9 6 12 6 4 1 1 2 1 1 2 2 2 4 1 1 À1 1 1 À1 2 2 À2 1 À1 0 1 À1 0 2 À2 0 1 1 2 1 1 2 À1 À1 À2 1 1 À1 1 1 À1 À1 À1 1 1 À1 0 1 À1 0 À1 1 0 1 1 2 À1 À1 À2 0 0 0 1 1 À1 À1 À1 1 0 0 0 1 À1 0 À1 1 0 0 0 0 g1 24 1 1 1 3 2 2 2 g2 24 1 1 1 3 À2 À2 À2 g3 4 1 1 1 À1 0 0 0 g4 6 1 ù ù2 0 Àù2 Àù À1 g5 6 1 ù2 ù 0 Àù Àù2 À1 g6 6 1 ù2 ù 0 ù ù2 1 g7 6 1 ù ù2 0 ù2 ù 1 Chapter 20 1. Character table of D6 3 D6 ( gi .15 and 17. 1) (b. 4. hj )| ÷1 ÷2 ÷3 ÷1 ÷2 ÷3 ÷1 ÷2 ÷3 3 ÷1 3 ÷1 3 ÷1 3 ÷2 3 ÷2 3 ÷2 3 ÷3 3 ÷3 3 ÷3 (1. a) (1. b: a3 ˆ b2 ˆ 1. b) (b.2) gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 Note: ù ˆ e2ðia3 6.14.Chapter 20 423 5. Exercise 27. Take b to be the re¯ection in the axis shown: . the character table is complete. b) (a. 1) (a. The table also records the trivial character ÷1 . the character table of D6 3 D6 is as shown. these are irreducible by Propositions 13. hj ) |CG ( gi . a) (a. 5. (a) Regard D8 as the subgroup of S4 which permutes the four corners of a square. below. Taking D6 ˆ ka. ÷ i l ˆ 1 for i ˆ 2. We have recorded ÷ as ÷5 . a) (b. as in Example 1. Since k÷ i . Character table of G (cf. 1) (1.

and ÷ an irreducible character of G of degree d. as in Example 19. For an example with d ˆ 3. .13). We obtain ÷1 5 H ˆ ø1 . Since each ø i has degree 1. or using (20. . ‡ dr ø r for some non-negative integers di . ÷3 5 H ˆ ø1 ‡ ø4 . ÷4 5 H ˆ ø3 ‡ ø5 . ø r be the irreducible characters of H. ÷11 5 A6 l ˆ 2. .2). we ®nd that the characters ÷ i 5 A6 (i ˆ 1.5. 7. the inequality (20. .3). take G ˆ A4 . . and take the character table of H to be gi |CH ( gi )| ø1 ø2 ø3 ø4 ø5 1 8 1 1 1 1 2 (1 3)(2 4) 8 1 1 1 1 À2 (1 2 3 4) 4 1 1 À1 À1 0 (1 3) 4 1 À1 1 À1 0 (1 2)(3 4) 4 1 À1 À1 1 0 (see Example 16. . Arguing as in Example 20. 5. Let ø1 . . ⠈ (1 À 1 0 0 À1 0 á â p 5)a2 1 0 0 À1 0 â á 1 À1 1 0 2 À1 À1 1 À1 0 1 À1 0 0 Note: á ˆ (1 ‡ 3. b 3 (1 3) gives the required isomorphism.6) gives ÷(1) ˆ d 1 ‡ X X X ‡ d r < d 2 ‡ X X X ‡ d 2 < nX 1 r 4. ÷ 5 Hl H < 3 follows at once from Proposition 20. 9) are distinct irreducible characters of A6 . .1. Then ÷ 5 H ˆ d1 ø1 ‡ . Let ÷1 . (b) Take the irreducible characters ÷1 . k÷11 5 A6 . . ÷2 5 H ˆ ø4 . Either by direct calculation. Character table of A6 gi 1 (1 2 3) |C A6 ( g i )| 360 9 ø1 ø2 ø3 ø4 ø5 ø6 ø7 1 5 10 9 5 8 8 1 2 1 0 À1 À1 À1 p (1 2)(3 4) (1 2 3 4 5) (1 3 4 5 2) (1 2 3)(4 5 6) (1 2 3 4)(5 6) 8 5 5 9 4 1 1 À2 1 1 0 0 5)a2. . For examples with d ˆ 1 or 2. . . . ÷5 5 H ˆ ø2 ‡ ø5 X 2. Write d ˆ k÷ 5 H. .424 Representations and characters of groups Then a 3 (1 2 3 4). . ÷ 5 Hl H . Also. .17. ÷5 of S4 as in Section 18. ÷11 be the irreducible characters of S6 . H ˆ V4 and ÷ an irreducible character of G of degree 3 (see Section 18. these give the characters ø1 . ø5 in our character table below. The inequality k÷ 5 H. .14.3(3) or Section 18. 3. . . we obtain from ÷11 5 A6 the two irreducible characters which we have called ø6 and ø7 . take G ˆ S3 and H a subgroup of order 2.

Hence sp (u) is a C H-submodule of CH. 2. (a) ÷1 5 H ˆ ÷2 5 H ˆ ø1 . 10. See (20. the elements u and ua form a basis of U 4 G. the induced module U 4 G is irreducible. 35X Chapter 21 1.13). (b) Since every element of G belongs to H or to Ha. upon restriction to A7 we get at least seven irreducible characters of A7 . (b) Using the Frobenius Reciprocity Theorem. ÷3 5 H ˆ ø2 ‡ ø3 . 14. the restriction of the irreducible character of degree 20 to A7 must be the sum of two different irreducible characters of degree 10. 10. Label the characters of H as follows: 1 ø1 ø2 ø3 1 1 1 (1 2 3) 1 ù ù2 (1 3 2) 1 ù2 ù where ù ˆ e2ðia3 . It is suf®cient to prove that if U is a CH-submodule of CH then dim (U 4 G) ˆ jG : Hjdim U . 14. 15. (a) Let u ˆ 1 À a2 ‡ b À a2 b. Since 20 occurs only once in the list of degrees for S7 . we obtain ø1 4 G ˆ ÷1 ‡ ÷2 ‡ ÷4 ‡ ÷5 . From the remaining fourteen irreducible characters of S7 . Hence the irreducible characters of A7 have degrees 1. 21. ø2 4 G ˆ ø3 4 G ˆ ÷3 ‡ ÷4 ‡ ÷5 X 3. Then ua2 ˆ Àu and ub ˆ u. We are told that A7 has exactly nine conjugacy classes. (c) The character ø of U and the character ø 4 G of U 4 G are given by 1 ø 1 a2 À1 b 1 a2 b À1 1 ø4G 2 a2 À2 a 0 b 0 ab 0 Since hø 4 G.Chapter 21 425 5. 6. Let Hgj (1 < j < m) be the distinct right . and we get precisely seven if and only if the restriction of each of the fourteen characters is irreducible. ø 4 Gi ˆ 1. ÷4 5 H ˆ ÷5 5 H ˆ ø1 ‡ ø2 ‡ ø3 .

÷ i 5 H ˆ di ø ‡ â where either â is a character of H or ⠈ 0. .426 Representations and characters of groups cosets of H in G. 5.23. ÷ 5 Hi H Tˆ 0 D ø9 ˆ ø or öX Thus If ø 4 G is irreducible. Let ö be an irreducible character of G. Now suppose that ø9 is an irreducible character of H. 3).17 that (ø(÷ 5 H)) 4 G ˆ (ø 4 G)÷. then for precisely one other irreducible character ö of H we have ø 4 G ˆ ö 4 G. . by the Frobenius Reciprocity Theorem. ÷i G Tˆ 0 D hø9. 7. dim (Ugj ) ˆ dim U (since u 3 ugj (u P U) is a vector space isomorphism). Hence. since ø is irreducible. (Compare Proposition 20. and on all other elements the values are zero. . Also. Thus ÷ i (1) > di ø(1). On elements of cycle-shapes (1). say ø 4 G ˆ ÷. and therefore jG: Hjø(1) > (d 2 ‡ X X X ‡ d 2 )ø(1)X 1 k The required result follows. ‡ dk ÷ k (1). (÷ö) 5 Hi H ˆ hø 4 G. where d i ˆ hø 4 G. Hence dim(U 4 G) ˆ dim(U (CG)) ˆ m dim U . ‡ Ug m is direct (since the elements in Ugj are linear combinations of elements in the right coset Hgj ). ÷ i i G ˆ hø.9. the values are as follows. that either (1) ø 4 G is irreducible. we deduce from Theorem 14. öi G ˆ hø(÷ 5 H). øl H ˆ 1. as in the proof of Proposition 20. by the Frobenius Reciprocity Theorem.11. By using the Frobenius Reciprocity Theorem twice. . ‡ Ugm . . Then U(CG) ˆ Ug1 ‡ . We have hø9 4 G. By applying the result of Exercise 6. The values of ö 4 G and ø 4 G are given by Proposition 21. ÷öi G ˆ h(ø 4 G)÷. (7) and (3. say ÷ 5 H ˆ ø ‡ ö.1 (also twice). . ö 5 Hi H ˆ hø. We have |G: H|ø(1) ˆ d1 ÷1 (1) ‡ . öi G X Since this holds for all irreducible characters ö of G. Then ÷(1) ˆ 2ø(1) and k÷ 5 H. or (2) ø 4 G is the sum of two different irreducible characters of the same degree. Suppose ®rst that ø 4 G is irreducible. ÷ i 5 Hi H . we obtain h(ø(÷ 5 H)) 4 G.) . together with the result of Exercise 19. 4. 3) 12 0 6. where Ugj ˆ {ugj : u P U}. Cycle-shape ö4G ø4G (1) 240 720 (7) 2 À1 (3. hence ÷ 5 H is reducible. we deduce. The sum Ug1 ‡ .

3. (b) jG9j ˆ p by Theorem 17. ÷l P Z. The degree of each irreducible character is 1 or 2. Theorems 11. ÷l P Z and k1 G . both k1 G . where r divides pq. (a) By hypothesis. ÷l ˆ 0. öl and k÷reg . ÷l ˆ kb÷reg . hence ÷1 5 H ˆ ø. and the sum of the squares of the degrees of the irreducible characters is 15 (Theorem 11. (b) We have h1 G . Hence every irreducible character has degree 1. and jGj 1 jGj(a ‡ bjGj) ˆ a ‡ bjGjX jGj Since ö is a character. say ø 4 G ˆ ÷1 ‡ ÷2. and so G is abelian by Proposition 9. øl H ˆ 1. Use Theorems 11. ÷l ˆ bjGj÷(1)ajGj ˆ b÷(1). (Compare Proposition 20. Then ö ˆ a1 G ‡ b÷reg . each degree divides 15 (Theorem 22. and r ‡ s ˆ 7 or 10 or 16. . (c) The number of conjugacy classes of G is r ‡ s.12. The number of linear characters divides 15 (Theorem 17. 17.12). and r . öl are integers. see Chapter 25. 22 ˆ 16X Hence r ˆ 4 or 8 or 16. This time. and ø9 is an irreducible character of H such that ø9 4 G has ÷1 or ÷2 as a constituent.12. (a) Since G is non-abelian. ÷1 5 Hi H Tˆ 0 D ø9 ˆ øX Thus If ø 4 G ˆ ÷1 ‡ ÷2 where ÷1 and ÷2 are irreducible characters of G. since both sides of this equation take the same values on all elements of G. there exist a. Hence kö À a1 G .Chapter 22 427 Suppose next that ø 4 G is reducible. Now suppose that ø9 is an irreducible character of H. then kö. öi ˆ 1 (a ‡ bjGj ‡ (jGj À 1)a) ˆ a ‡ b. ÷1 i G Tˆ 0 D hø9. and if there are r characters of degree 1 and s of degree 2.11 again. (c) If ÷ is a non-trivial irreducible character of G. But kö À a1 G .18).18. 1 < s and r ‡ sq 2 ˆ pqX Hence r ˆ q and s ˆ ( p À 1)/q. 17.11 and 22. not every irreducible character has degree 1 (Proposition 9. We have hø9 4 G. Then ÷1 (1) ˆ ø(1) and k÷1 5 H.11 show that there are r irreducible characters of degree 1 and s irreducible characters of degree q. öi ˆ h÷reg .12. 2.) 4.11). b P C such that ö(g) ˆ a for all g Tˆ 1 and ö(1) ˆ a ‡ bjGj. (For more information on groups of order pq. moreover.11. then ø9 ˆ ø. then r divides 16.11 and 22.11).) Chapter 22 1. 12 ‡ s .

6 in some order. Partition Gnf1g into subsets by putting each element with its inverse. This contradicts Proposition 22. The only possibility which is consistent with equation (IV) is that the values of ÷ i (1) for 4 < i < 7 are 1.5. . . (Further information about the number of characters ÷ such that ÷ ˆ ÷ appears in Theorem 23. By Corollary 22. by part (a). 120.) 6.1 and Corollary 23. we have ÷( g) ‡ ÷( g À1 ) ˆ ÷( g) ‡ ÷( g) ˆ 2÷( g).428 Representations and characters of groups (d) Since ÷(1) divides jGj. the second possibility is ruled out by equation (II). (4. 1 G l ˆ 0. or ÷ i (g) ˆ Æ2 for exactly one i and ÷ i (g) ˆ 0 for all other i . 1. with ÷1 ˆ 1 G . Also (III) 2 2 7 ˆ iˆ1 (÷ i (1))2 ˆ 120X Since 5 ‡ 10 . and hence also b. (6.2. (d) We now have the following part of the character table of G: . ÷ i (1)  ÷ i ( g) mod 5 for all i. (b) For all g P G. we deduce that ÷2 (1) ˆ ÷3 (1) ˆ 5. and ÷ i (g) ˆ Æ1 for 4 < i < 7. À1). Let ÷1 . and from equation (III) we have (IV) 7 ˆ (÷ i (1))2 ˆ 69X iˆ4 Hence the possibilities for the pairs of integers (÷ i (1). The stated result follows. . But À÷(1)a2 is a rational number which is not an integer (since ÷(1) divides jGj. say i ˆ 2. part (c) implies that b|G| is an integer. a. (a) By Theorem 22. 1).27. hence is odd). Thus ÷ ˆ 1 G . ÷2 (1)  ÷3 (1)  0 mod 5. by Lagrange's Theorem. and (II) 1 ‡ 7 ˆ iˆ2 ÷ i (1)÷ i ( g) ˆ 0X From equation (I) we deduce that either ÷ i ( g) ˆ 0. 4.16. since 1 ˆ h÷. Each such subset has size 2. . by part (b). (a) If g P G then g has odd order. Hence ˆ ÷( g) ˆ ÷(1) ‡ 2á gPG for some algebraic integer á. By the column orthogonality relations. ÷(g) is an integer for all characters ÷. ÷ i ( g)) with 4 < i < 7 are (1. ÷7 be the irreducible characters of G. then k÷. (b) We deduce from part (a) that ÷ i (g) ˆ 0 for two values of i. and hence á ˆ À÷(1)a2. Therefore. 3. 5. Æ1 for all i. is an integer. 1 G i ˆ ÷( g)X jGj gPG (c) If ÷ Tˆ 1 G in part (b). 1). if g2 ˆ 1 then g ˆ 1. (c) By Corollary 22. Therefore. we have (I) 1‡ 7 ˆ iˆ2 (÷ i ( g))2 ˆ 5.27. 4. and ÷(g) is an algebraic integer.

an (1) First. from which €7 we see that ÷ i (g3 ) ˆ 1 for 1 < i < 4. 1. 1. 1. ÷ i ( g4 )  ÷ i (1) mod 3 and 7 (÷ i ( g4 ))2 ˆ 6. 5 and 7 we obtain ÷2 (g6 ) ˆ À÷3 (g6 ) ˆ ÷4 (g6 ) ˆ À1 and (without loss of generality) ÷5 (g6 ) ˆ À÷6 (g6 ) ˆ 1. € 1. gi Order of gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 g1 1 120 1 5 5 1 4 4 6 g2 2 12 1 À1 1 À1 À2 2 0 g3 2 8 1 1 1 1 0 0 À2 g4 3 6 1 À1 À1 1 1 1 0 g5 4 4 1 1 À1 À1 0 0 0 g6 6 6 1 À1 1 À1 1 À1 0 g7 5 5 1 0 0 1 À1 À1 1 . 0. 0. ÷ i (g5 )  ÷ i (1) mod 2 and 7 (÷ i (g5 ))2 ˆ 4. we deduce that iˆ1 ÷ i (g3 ) ˆ Æ1 for 1 < i < 4 and€ the values of ÷ i ( g3 ) for 5 < i < 7 are that 7 0. 4. (2) Next. 7. We are told that ÷ i (gj ) is€ integer for all i. Since 7 ÷ i (1)÷ i (g5 ) ˆ 0. Æ2 in some order. 0. Also iˆ1 ÷ i (g3 )÷ i ( gr ) ˆ 0 for r ˆ 4. Hence the values of iˆ1 ÷ i (g4 ) for 1 < i < 7 are 1. € Hence ÷ i (g5 ) ˆ Æ1 iˆ1 for 1 < i < 4 and ÷ i (g5 ) ˆ 0 for 5 < i < 7. The character table of G is as shown. Therefore iˆ1 ÷ i (g6 ) ˆ Æ1 for 1 < i < 6 and ÷7 (g6 ) ˆ 0. À1.Chapter 22 gi Order of gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 g1 1 120 1 5 5 1 4 4 6 g2 2 12 1 g3 2 8 1 g4 3 6 1 g5 4 4 1 g6 6 6 1 429 g7 5 5 1 0 0 1 À1 À1 1 We successively calculate the ®ve remaining columns of the character table. À1. € (3) Since ÷ i (g3 )  ÷ i (1) mod 2 and 7 (÷ i (g3 ))2 ˆ 8. (5) Only the entries in column 2 remain to be calculated. These can be obtained from the column orthogonality relations. À2 in order from the top. j. respectively. By applying the column orthogonality relations involving column 6 and columns 3. 0. From the relation iˆ1 ÷ i (1)÷ i ( g 3 ) ˆ 0 we now deduce that the entries in column 3 are 1. 1. iˆ1 we deduce that ÷4 ( g5 ) ˆ À1 and (without loss of generality) ÷2 (g5 ) ˆ À÷3 (g5 ) ˆ 1. € (4) We have ÷ i (g6 )  ÷ i (g4 ) mod 2 and 7 (÷ i (g6 ))2 ˆ 6. 1.

by Lagrange's Theorem. . Let m be the order of g. either i j ˆ 0 or n j is even and i j ˆ n j a2. . which is of the form x n ‡ anÀ1 x nÀ1 ‡ X X X ‡ a1 x ‡ a0 (ar P Z)X Conversely. Hence the number of real irreducible characters is 2 m . Then ÷ A (g) ˆ 1 2 2 2 2((ë1 ‡ ë2 ) À (ë1 ‡ ë2 )) ˆ ë1 ë2 ˆ det (gr) (see Proposition 19. assume that ë is a root of the polynomial p(x) ˆ a0 ‡ a1 x ‡ . The elements g of D2 n for which g2 ˆ 1 are 1. Then det (A À ëIn ) ˆ 0. As p(ë) ˆ 0.13 of é÷ that é÷ ˆ À1 if and only if ÷ A ˆ 1 G . Chapter 23 1. all of whose entries are integers. . it follows that ë is an eigenvalue of A. where m is the number of the integers n1 . . The character ÷ of G ˆ C n1 3 . . 2. . Hence gÀ2 xg2 ˆ x. However. ë is therefore an algebraic integer. 3.8. g irr where for each j. 3 C n r which is given by i i ÷( g 11 X X X g irr ) ˆ ë11 X X X ë irr is real if and only if ë i ˆ Æ1 for all i with 1 < i < r. Since A has integer entries. Since ÷(1) ˆ 2 we have ÷ A (1) ˆ 1. 4. . Suppose ®rst that ë is an eigenvalue of an n 3 n matrix A. and hence ë is a root of the polynomial det (xIn À A). Let ë1 and ë2 be the eigenvalues of gr. it follows that x ˆ 1.14). Therefore x À1 ˆ gÀ1 xg ˆ x. Since x 2 ˆ 1 and x has odd order. € These numbers coincide with ÷(1). It now follows from the De®nition 23. Let I H 0 1 0 XXX 0 f 0 0 1 0 g g f f 0 0 0 0 g f F Aˆf F F gX F g F g f F d 0 0 0 1 e Àa0 Àa1 Àa2 XXX Àa nÀ1 Check that det (xIn À A) ˆ p(x). m ˆ 2n ‡ 1 for some integer n. Assume that x P G and x is real.430 Representations and characters of groups 7. and n ‡ 2 elements if n is even. summing over all the irreducible characters ÷. . Since jGj is odd. . The number of such elements is also 2 m . the elements g of G which satisfy g2 ˆ 1 are i precisely those elements g11 . Adopt the notation of Theorem 9. so g2 P CG (x). The result follows. Since é÷ < 1 for all ÷. nr which are even. Now the ni th root of unity ë i can be À1 if and only if ni is even. it follows from the Frobenius±Schur Count of Involutions that we must have é÷ ˆ 1 for all ÷. a i b (0 < i < n À 1) (and also a na2 if n is even)X This gives n ‡ 1 elements if n is odd. Then g ˆ g2( n‡1) P CG (x). Then gÀ1 xg ˆ x À1 for some g P G. . . ‡ a nÀ1 x nÀ1 ‡ x n (a r P Z).

X X X .16. 2} then â(v i g. å n v2 ) ˆ â(v1 . (b) It is easy to check that if g ˆ a or b and i. v) ˆ 0 for all v P V} is a CGsubmodule of V. . f 9) ˆ ä ij for all i. respectively. The de®nition of â shows that â is symmetric if å n ˆ 1 and â is skew-symmetric if å n ˆ À1. . 7. we have At ˆ ÀA. and de®ne the basis e1 . (c) The elements of T4 n are ai and ai b (0 < i < 2n À 1). v1 ) ˆ å n ˆ â(v1 . Write Q ˆ (q ij ). 6. Clearly é÷1 ˆ é÷3 ˆ 1. there exists a nonzero G-invariant skew-symmetric bilinear form â on V. we may construct a basis f 1 . The result now follows from Exercise 4. . respectively. . â(v1 b. v j g À1 )X For example. f j ). so det A Tˆ 0. Also A is invertible by (à ). Let V be a CG-module with character ÷. . there is an orthogonal matrix Q (i. . b ij ˆ â( f i . Hence ÷ (é÷)÷(1) ˆ 2.e. since V is irreducible it follows that fu P V : â(u. v1 bÀ1 ). f j )X By applying the Gram±Schmidt orthogonalization process. Since é÷ ˆ À1. according to whether n is odd or € even. and é÷2 ˆ é÷4 ˆ 0 or 1. v1 ) ˆ â(v2 . . . (d) Refer the Exercise 18. . j. Then det (ar) ˆ 1 and det (br) ˆ Àå n . Let 9 n i j P ˆ ( pij ) be the n 3 n matrix which is given by ˆ f9ˆ pij f j X i j Then PAP ˆ I n and PBP is symmetric. By part (a) (or part (b)) and the construction of the characters ø j in Exercise 17. as n ˆ ÷(1) the result is proved. . respectively. so é÷ ˆ Æ1. By a well known property of symmetric matrices. (a) First. v j ) ˆ â(v i . v n of V and let A be the n 3 n matrix with ij-entry â(v i . a has order 2n and ai b has order 4. it is easy to check that ÷( g) is real for all g P G. As â is G-invariant.3 for the characters ø j (1 < j < n À 1) and ÷ j (1 < j < 4) of T4 n. so det A ˆ (À1) n det A. v2 of V. en of V by ˆ ei ˆ qij f 9 X j j t t . . Let r be the representation obtained by using the basis v1 . Choose a basis f1 .Chapter 23 431 5. v) ˆ 0 for all v P V g ˆ f0gX (à ) Pick a basis v1 . the subspace {u P V: â(u. we get éø j ˆ À1 or 1. QQt ˆ I) such that Q(PBPt )QÀ1 is diagonal. j P {1. . Hence an is the only element of order 2. Hence â is G-invariant. according to whether j is odd or even. f 9 of V such that â1 ( f 9. The result now follows from Theorem 23. It follows that n is even. according to whether n is odd or even. Since â is skew-symmetric. hence det (gr) ˆ 1 for all g P G if and only if å n ˆ À1. fn of V and de®ne the symmetric n 3 n matrices A ˆ (aij ) and B ˆ (bij ) by aij ˆ â1 ( f i .6. € nÀ1 Therefore jˆ1 (éø j )ø j (1) ˆ 0 or À2. v j ). Therefore det (At ) ˆ (À1) n det A.

since QPAPt Qt ˆ I n . with ÷1 ˆ 1 G ). It is clear that the given set of matrices has size p( p À 1). . and r is a homomorphism as (Hx)(r gh ) ˆ Hxgh ˆ (Hx)(r g )(r h ). i. For the last part. Let c1 . By Schur's Lemma there exists ë P C such that vW ˆ ëv for all v P V9. For closure. Then V9 is a CG-module. By the orthogonality relation for c2 we have € ÷ i (t)2 ˆ |CG (t)| ˆ 2 (the sum over all irreducible characters ÷ i . identity is       1 0 1 y 1 À yx À1 . inverse of is . Now the orthogonality of c1 and c2 gives ÷1 (1) ˆ ÷2 (1) ˆ 1 and ÷2 (t) ˆ À1. .e. we have G9 ˆ 1.11. ej ) ˆ 0 if i Tˆ j. Call it G. which we are assuming to be an irreducible CG-module. (c) Let G ˆ C3 ˆ ka: a3 ˆ 1l. r g is a permutation. . ÷1 and ÷2 are the only linear characters. 2. ej ) ˆ ä ij . G is abelian. Vx P G D g P ’ xPG xÀ1 HxX Finally. 10. 0 1 0 x 0 x À1 Therefore G is a group. Hence |G : G9| ˆ 2 by Theorem 17. 0 x 0 x9 0 xx9 associativity is a property of matrix multiplication. and since QPBPt Qt is diagonalX â(ei . . Let ç ˆ e2ðia5 and å ˆ e2ðia11 . and consider the vector space V9 over C with basis v1 . We have g P ker r D Hxg ˆ Hx. 9. But v1 W ˆ ëv1 P V. r is a homomorphism G 3 Sym(Ù)  S n with kernel which is contained in H. and let V be the RG-submodule of the regular RG-module which is spanned by 1 À a and 1 À a2 . v n be a basis of the RG-module V. xgx À1 Vx P G D „ xPG x À1 P H. c2 be the columns of the character table of G corresponding to the classes {1} and t G . so we may take ÷1 (t) ˆ 1. ÷2 (t) ˆ Æ1 and ÷ i (t) ˆ 0 for i > 3. so ë P R. v n . as Hxg ˆ Hyg A Hx ˆ Hy. . if G is simple then since G9 v G.1. . 8. Further. (a) The proof is similar to that of part (1) of Schur's Lemma 9. ⠈ å 2 ‡ å 6 ‡ å 7 ‡ å 8 ‡ å 10 X . since a linear character must take the value Æ1 on t. . Then V is an irreducible RG-module. Hence G  C2 . Hx. Chapter 25 1.432 Then Representations and characters of groups â1 (ei . note that      1 y 1 y9 1 y9 ‡ yx ˆ . and write á ˆ å ‡ å 3 ‡ å 4 ‡ å 5 ‡ å 9 . De®ne W: V 3 V by vW ˆ av (v P V). . (b) Let v1 .

(a) Note that À1 is the only element of order 2 in ZÃ. Also. 5 gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 1 55 1 1 1 1 1 5 5 a 11 1 1 1 1 1 á â a2 11 1 1 1 1 1 â á b 5 1 ç ç2 ç3 ç4 0 0 b2 5 1 ç2 ç4 ç ç3 0 0 b3 5 1 ç3 ç ç4 ç2 0 0 433 b4 5 1 ç4 ç3 ç2 ç 0 0 3. Then G1 ˆ ha. Also.6 that ö1 (a) and ö2 (a) are not both real.9. This time. m is coprime to q. and it follows from Corollary 15. Hence ö1 (a) and ö2 (a) ÷.Chapter 25 Character table of F11. then a is not conjugate to aÀ1 . since both m and v have u order q modulo p. If p  1 mod 4.6(c). |CG (a)| ˆ p. Hence ö2 (a) ˆ ö1 (a). b9À1 ab9 ˆ av iX Hence G1  G2 . Hence ˆ 0ˆ ÷(1)÷(a) ˆ q ‡ qö1 (a) ‡ qö2 (a)X ÷ irred m Therefore ö1 (a) ‡ ö2 (a) ˆ À1. 2ö1 (a)ö1 (a) ˆ ( p ‡ 1)a2. bÀ m abm ˆ au ˆ av . Also. Let b9 ˆ bm. b9: ap ˆ b9q ˆ 1. and we ®nd that p ö1 (a) and ö2 (a) are (À1 Æ i p)a2. Hence p u m  À1 mod p for some m D the element u of Zà has even order p D q is even D p  1 mod 4X (b) By Proposition 25. Recall that Zà is cyclic. so ˆ pˆ ÷(a)÷(a) ˆ q ‡ ö1 (a)ö1 (a) ‡ ö2 (a)ö2 (a)X ÷ Hence ö1 (a)ö1 (a) ‡ ö2 (a)ö2 (a) ˆ ( p ‡ 1)a2. 4. If p  À1 mod 4. . Hence bm has order q. are (À1 Æ p)a2. there exists an integer m p such that u m  v mod p. then a is conjugate to aÀ1 and so ÷(a) is real for all characters p so (ö1 (a))2 ‡ (ö2 (a))2 ˆ ( p ‡ 1)a2. aG ˆ {au : m P Z}. Therefore aÀ1 P aG D u m  À1 mod p for some m D p  1 mod 4X (c) ÷ i (a) ˆ 1 for all the q linear characters ÷ i of G. so by Exercise 1.

fa r b3 : 0 < r < 8g. . 8. ö1 (a) ˆ mˆ1 å u .10. Let H ˆ ka. .3 (see Theorem 19. ÷6 of G. 6g. Hence we get six linear characters ÷1 . Then ÷ 4 E is the irreducible character ÷3 given in the table which follows. Then H1 v G and Ga H 1  C6 . Then for all h P H. fa r b: 0 < r < 8g.18).3 (see Theorem 19. b2 l. The result now follows from part (c). The characters ÷4 . so E has exactly two linear characters. 5. fa r b4 : 3 B rg. (a) F13. . E9 ˆ H. Also. . fa r : 3 B rg. 7. . . . fa r b5 : 0 < r < 8gX Let H1 ˆ kal. Lift the irreducible character of D6 of degree 2 to obtain ÷7 in the table below. .434 Representations and characters of groups €( pÀ1)a2 m (d) By Theorem 25. as shown. 3. The conjugacy classes of G are f1g. it follows that {u. H 2 bi  D6 . there exist gi P E such that gi Tˆ 1 but ÷ i (gi ) ˆ ÷ i (1) (so gi P Ker ÷ i ).10). Character table of E gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 1 18 1 1 2 2 2 2 a 9 1 1 2 À1 À1 À1 b 9 1 1 À1 2 À1 À1 ab 9 1 1 À1 À1 2 À1 a2 b 9 1 1 À1 À1 À1 2 c 2 1 À1 0 0 0 0 6. u 2 . . bl. ÷5 and ÷6 are obtained similarly. All the elements outside H form a single conjugacy class of E. 6g. a6 g. Then ÷8 ˆ ÷7 ÷2 and ÷9 ˆ ÷7 ÷3 are also irreducible. and for all i with 1 < i < 6. say ÷1 and ÷2 . (c) D6 3 F13. fa r b4 : r ˆ 0. fa r b2 : r ˆ 0.18). the conjugacy class hE consists of h and hÀ1 . u ( pÀ1)a2 } is precisely the set of quadratic residues modulo p. 3. Z(E) ˆ {1}. The ®nal irreducible character ÷10 can be found by using the column orthogonality relations.3 (see Theorem 25. fa3 . fa r b2 : 3 B rg. (b) C2 3 F13. Then H2 v G and Ga H 2 ˆ h H 2 a. A typical non-trivial linear character of H is 1 ÷ 1 a 1 a2 1 b ù b2 ù2 ab ù ab2 ù2 a2 b ù a2 b2 ù2 where ù ˆ e2ðia3 . Let H2 ˆ ka3 . Since Zà is cyclic of order p p À 1 and u has order ( p À 1)a2.

2. since H is abelian.11. and so ÷(1) ˆ 1 or p by Theorem 22. b: a9 ˆ b6 ˆ 1. and r ‡ sp2 ˆ pn . Character table of H (a non-abelian group of order 27) gi |C H ( g i )| ÷00 ÷01 ÷02 ÷10 ÷11 ÷12 ÷20 ÷21 ÷22 ö1 ö2 1 27 1 1 1 1 1 1 1 1 1 3 3 z 27 1 1 1 1 1 1 1 1 1 3ù 3ù2 z2 27 1 1 1 1 1 1 1 1 1 3ù2 3ù a 9 1 1 1 ù ù ù ù2 ù2 ù2 0 0 a2 9 1 1 1 ù2 ù2 ù2 ù ù ù 0 0 b 9 1 ù ù2 1 ù ù2 1 ù ù2 0 0 ab 9 1 ù ù2 ù ù2 1 ù2 1 ù 0 0 a2 b 9 1 ù ù2 ù2 1 ù ù ù2 1 0 0 b2 9 1 ù2 ù 1 ù2 ù 1 ù2 ù 0 0 ab2 9 1 ù2 ù ù 1 ù2 ù2 ù 1 0 0 a2 b2 9 1 ù2 ù ù2 ù 1 ù 1 ù2 0 0 Note: ù ˆ e2ðia3 . bÀ1 ab ˆ a2 l gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 ÷8 ÷9 ÷10 Note: ù ˆ e2ðia3 1 54 1 1 1 1 1 1 2 2 2 6 a3 27 1 1 1 1 1 1 2 2 2 À3 a 9 ab2 9 ab4 9 b 6 b2 18 b3 6 1 À1 1 À1 1 À1 0 0 0 0 b4 18 435 b5 6 1 1 1 1 1 1 ù2 ù Àù ù2 1 ù ù2 ù2 ù 1 1 1 À1 1 1 ù2 ù ù ù2 1 ù ù2 Àù2 ù À1 À1 À1 0 2 À1 Àù2 Àù 0 2ù2 À1 Àù Àù2 0 2ù 0 0 0 0 0 1 1 ù Àù2 ù2 ù 1 À1 ù ù2 ù2 Àù 2 0 2ù 0 2ù2 0 0 0 Chapter 26 1. by Corollary 21. Hence ÷(1) < p.Chapter 26 Character table of G ˆ ka. For all other elements h of H. m is at least 2. Let ÷ be an irreducible character of G. by Theorem 11X12X Since s ˆ p nÀ2 À p mÀ2 and s is an integer. by Theorem 17X11. hz 2 }. Assume that G has r linear characters and s irreducible characters of degree p. ø 4 Gi G Tˆ 0 by the Frobenius Reciprocity Theorem. {z} and {z 2 } are conjugacy classes of H. the conjugacy class hH ˆ {h. But ø(1) ˆ 1. hz. {1}. øl H Tˆ 0 for some irreducible character ø of H.20. Then k÷ 5 H. and (ø 4 G)(1) ˆ p. Then r ˆ pm for some m. Therefore h÷.

Here. Therefore G9 ˆ h Zi (see Proposition 17. since g4 ˆ 1 for all g P G. 1}. B. and Gah Zi is abelian while G is non-abelian. Hence every element of G has the form A i B j C k D l Z m for some i. b2 ˆ a8 . The character table of D16 is given in Section 26. AC ˆ ÀCA. BC ˆ CB. (c) A routine calculation shows that every matrix which commutes with each of A. k. the group K which appears in Theorem 26. 1 32 a8 32 a 16 a2 16 a3 16 a4 16 a5 16 a6 16 a7 16 b 4 ab 4 1 À1 À1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 À1 1 1 À1 1 À1 1 À1 1 À1 1 1 1 À1 1 À1 1 À1 1 À1 À1 2 2 p0 À2 0 2 0 0 p p À2 p0 2 2 2 0 Àp2 À2 Àp2 0 2 0 p p 2 2 À 2 0 2 À2 2 0 À 2 0 2 À2 cj c2 j c3 j c4 j c5 j c6 j c7 j 0 5. we get a representation A i B j C k D l Z m 3 (À1) ir‡ js‡ kt‡ lu X Together with the irreducible representation of degree 4. . {ar b: r odd}. 3.12.10). These are as follows: for each (r. C and D has the form ëI for some ë P C. a8 }. Combined with part (b). t. s. . we obtain the characters ÷1 . AD ˆ DA. G has 11 conjugacy classes: {1}. u P {0. 4. By lifting the irreducible characters of D16.3. Then the four characters ø j ( j ˆ 1. (a) Check that AB ˆ ÀBA. 3. m P f0. {a8 }. . jGj > 12 ‡ 42 ˆ 17. j. and GaK  D16 . l.8 (D16 ˆ G1 ) and in Section 18. s. {ar b: r even}.4 is {1. 5. b: a16 ˆ 1. (d) Since G has irreducible representations of degrees 1 and 4. by Theorem 11. so jGj < 32. also G is a 2-group. bÀ1 ab ˆ aÀ1 l gi |CG ( g i )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 øj ( j ˆ 1. G has precisely 16 representations of degree 1. aÀ r } (1 < r < 7). Hence by Corollary 9. these are all the irreducible representations of G. (b) A2 ˆ ÀB2 ˆ ÀC2 ˆ D2 ˆ I. 7) Note: cm ˆ e2ði ma16 ‡ eÀ2ði ma16 ˆ 2 cos (mð/8) 3. Since Gah Zi is abelian. t. We obtain representations as follows: . Since G9 ˆ h Zi. this shows that jGj ˆ 32. Hence Z P G. 5. ÷7 of G as shown in the table at the top of this page. 1g. 7) come from inducing to G those linear characters ÷ of kal for which ÷(a8 ) ˆ À1. it follows that g 2 P h Zi for all g P G. u) with r. (a) Let å ˆ e2ðia8 . the given representation is irreducible.436 Representations and characters of groups Character table of G ˆ ka. BD ˆ ÀDB. CD ˆ ÀDC. . {ar .3.

z 3 d 0 1 0 e. . 1 0 å3 3 2 3 0 0 1 . a2 z}  C2 3 C2 . so the representations are faithful.1(2). .b3 .z3 X À1 1 0 0 i (b) If j ˆ 5. . so G5 T G8 . 6.b3 .b3 2 2 Chapter 26 0 1 1 0 0 1 3 . (d) The following give faithful representations: H I H I H I i 0 0 0 1 0 1 0 0 f g f g f g G7 : a 3 d 0 Ài 0 e. b 3 d 1 0 0 e.16. G9 are isomorphic. . 4 and 8 in G1 . Also jGa Z(G)j Tˆ p by Lemma 26. G9 : G1 Order Order Order Order 1 2 4 8 1 9 2 4 G2 1 1 10 4 G3 1 5 6 4 G4 1 3 4 8 G5 1 3 12 0 G6 1 7 8 0 G7 1 11 4 0 G8 1 3 12 0 G9 1 5 10 0 Therefore no two of G1 . . À1 0 2 3 0 1 . while G8 aG8 9  C2 3 C2 3 C2 . i f G8 : a 3 d 0 0 H 0 0 0 À1 H I 0 0 0 1 0 1 0 0 g f g f g Ài 0 e.2 G1 : a 3 2 G2 : a 3 G3 : a 3 G4 : a 3 2 G5 : a 3 å å 0 À1 3 . 2. Since Z(Gj ) is not cyclic. å5 1 0 3 2 3 2 3 0 0 1 i 0 . Hence a . 7 or 8 then Z(Gj ) ˆ {1. (a) By Lemma 26. and g P CG ( g). by Proposition 9. z 3 d 0 1 0 eX I H I 0 1 0 0 1 0 0 À1 0 1 0 0 1 6. b 3 d À1 0 0 e. so give representations. Therefore j Z(G)j ˆ p or p2 . a2 . (c) Check that the matrices satisfy the required relations. .b3 3 . 7. except possibly G5 and G8 . Gj has no faithful irreducible representation. z. . .1(1) we have {1} Tˆ Z(G) Tˆ G. The following table records the numbers of elements of orders 1. If g P Z(G) then Z(G) < CG (g) Tˆ G. It is easy to see that the matrices generate groups with more than eight elements. But G5 aG5 9  C2 3 C4 . Assume that j Z(G)j ˆ p2 . 3 437 0 å 0 0 å À1 2 3 å 0 2 0 å 0 1 0 .b3 .

Assume that zˆ Then   a c b d  P Z(SL (2. Since a2 commutes with a. so a ˆ Æ1. 2. and assume that Ga Z ˆ haZ. in which case Ga(G9 ’ Z(G)) T Q8 by part (a). Check that   and   1 0 1 1 1 À1 0 1 are elements of G ˆ SL (2. (b) Assume that G has r irreducible characters of degree 1 and s irreducible € characters of degree p.1(1). p))X  1 z A c ˆ 0.12). (a) Let Z ˆ Z(G). and hence ba2 ˆ b3 z ˆ b2 zb ˆ a2 b. if |Z(G)| ˆ p2 then |G9| ˆ p. Therefore Ga Z T Q8 . Chapter 27 1. 3) of order 3 which are not conjugate to each other. bZi. or G9 ’ Z(G) ˆ Z(G). Therefore jGaG9j ˆ r ˆ p2 or p3 . a ˆ dX 0   1 1 1 ˆ 0 0 1    0 1 0 z ˆ À1 0 À1 z Therefore z ˆ aI. Since ÷(1)2 ˆ p4 (Theorem 11. so r ‡ sp2 ˆ p4 . Part (b) follows. then by Exercise 7. G has p2 ‡ ( p4 À p2 )a p conjugacy classes. we have a2 P Z. (b) If G is a non-abelian group of order 16. By parts (a) and (b). (c) Note that G9 ’ Z(G) Tˆ {1} by Lemma 26. 8. and since z P SL (2. as r ‡ s is equal to the number of conjugacy classes of G. a2 Z ˆ b2 Z. and if r ˆ p2 then r ‡ s ˆ 2 p2 À 1. we have a2 ˆ 1. and . Hence |G9 ’ Z(G)| ˆ p. Therefore. b and all elements in Z. in which case Ga(G9 ’ Z(G)) is abelian. The element   À1 0 0 À1 lies in Z(G). bÀ1 abZ ˆ aÀ1 ZX Then a2 ˆ b2 z for some z P Z. either G9 ’ Z(G) ˆ G9. and 1  1 z A c ˆ Àb. with a4 P Z.438 Representations and characters of groups jC G ( g)j ˆ p3 and j gG j ˆ p. p). and if |G9| ˆ p2 then |Z(G)| ˆ p. there are no irreducible characters of degree greater than p.

6. Exercise 13. Assume. The € equation j ÷ j (g4 )÷ j ( g 4 ) ˆ 6 gives áá ˆ 1. g3 can be deduced from the column orthogonality relations.1. they must be ÷6 and ÷7 in some order. 0). First observe that the vector space (Z3 )2 has exactly four 1-dimensional subspaces. a subgroup of S4 of order 12. The characters ÷1 . ÷4 of G are obtained by lifting to G the irreducible characters of A4 (which are given in Section 18. (1. á ˆ Æ1. say ÷5 ÷2 ˆ ÷6 and ÷5 ÷3 ˆ ÷7 . ÷6 . Since ÷5 ÷2 and ÷5 ÷3 are irreducible characters of G of degree 2. ÷(g5 ) ˆ ÷( g4 ) and ÷(g6 ) ˆ ÷( g7 ) for all ÷. ÷3 . one of ÷5 . The values of ÷5 . . without loss of generality. therefore GafÆIg  A4 . ÷7 on the elements g1 . Then ÷5 ( g4 ) ˆ á. namely the spans of the vectors (0. Then á ˆ À1 since ÷5 ( g4 )  ÷5 (1) mod 3.27. (2.5 implies that ÷ j ( g7 ) ˆ À÷ j ( g4 ). that ÷5 is real. Since á is real. whose values on g4 are áù and áù2 . so we obtain a homomorphism ö: G 3 S4 . ÷6 and ÷7 must be real. so by Theorem 23. Check that Ker ö ˆ {ÆI}.2). 1). Hence the following are conjugacy class representatives: g1 1 0 1 24 0 1 g2 À1 0 0 À1 2 24 g3 0 À1 4 4 1 0 g4         1 1 0 1 Order of gi |CG ( gi )| 3 6  g5 1 0 À1 1 3 6   g6 À1 1 0 À1 6 6   g7 À1 À1 0 À1 6 6  Order of gi |CG ( gi )| We now describe how to construct the character table of G. Finally. The group G permutes these subspaces among themselves. Hence GafÆIg  Im ö. Now note that for j ˆ 5. 1) and (1. by Corollary 22. 1). Also á Tˆ 0. ÷2 .Chapter 27  0 À1 1 0  439 has order 4. which is given below. Note that G has three real conjugacy classes. 7. where á is real. g2 .

. where 2 3 2 3 2 3 1 0 2 0 4 0 h1 ˆ Z. notice that T is isomorphic to the group of order 21 whose character table is found in Exercise 17. Apply Proposition 17. where ÷ is an irreducible character of G. Representatives of the conjugacy classes of T are h1 . (a) For the character table of T.25.440 Representations and characters of groups Character table of SL (2. 1 G l ˆ 1. 3) gi |CG ( gi )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 Note: ù ˆ e2ðia3 3.6. 4. Hence 1 T : G ˆ 1 G ‡ ÷. 1 T : Gl ˆ 2 and k1 T : G. h2 ˆ Z. . . . .2 and Example 21. h3 ˆ Z. ë : Gl ˆ 1. The values of 1 T : G and ë : G are as follows (see Proposition 21. kë : G. h5 . write ö ˆ ë : G. so ë : G is irreducible. 0 1 0 4 0 2 2 3 2 3 1 1 1 À1 h4 ˆ Z. h5 ˆ ZX 0 1 0 1 Two of the linear characters of T are hi |CT (h i )| 1T ë h1 21 1 1 h2 3 1 ù h3 3 1 ù2 h4 7 1 1 h5 7 1 1 g1 24 1 1 1 3 2 2 2 g2 24 1 1 1 3 À2 À2 À2 g3 4 1 1 1 À1 0 0 0 g4 6 1 ù ù2 0 À1 Àù Àù2 g5 6 1 ù2 ù 0 À1 Àù2 Àù g6 6 1 ù2 ù 0 1 ù2 ù g7 6 1 ù ù2 0 1 ù ù2 where ù ˆ e2ðia3 .23): gi |CG ( g i )| 1T 4 G ë4G g1 168 8 8 g2 8 0 0 g3 4 0 0 g4 3 2 À1 g5 7 1 1 g6 7 1 1 We ®nd that k1 T : G. Also.

. are equal to 4. . öl ˆ k÷ S . We calculate that kæ. ÷l ˆ 1. (b) By lifting. 0. 5. 168. The remaining irreducible characters ÷4 . Note that because g2 lies in Z(G). Æ1. . and ÷11 (1) ˆ 8. 1 G l ˆ k÷ S . ÷2 and ÷6 . ö and ÷ are three of the six irreducible characters of G. by Corollary 22. apply part (c). . ÷5 can readily be calculated using the column orthogonality relations (noting that gi is real if and only if 1 < i < 4). ÷11 (1). Next.) Also. Æ1. ÷7 (1)2 ‡ ÷8 (1)2 ‡ ÷11 (1)2 ˆ 96. so ÷9 (1) ˆ ÷10 (1) ˆ 6. Further.14): gi |CG ( g i )| ÷ ÷S æ ø g1 168 7 28 12 6 g2 8 À1 4 4 2 g3 4 À1 0 0 0 g4 3 1 1 0 0 g5 7 0 0 À2 À1 g6 7 0 0 À2 À1 441 We ®nd that k÷ S . ÷ and ø are the characters ÷1 . . 2 2 jˆ1 (÷ j (1)) ˆ 168. since ÀI is not in € kernel of these characters). ÷(g6 ) P Z for all characters ÷. ÷11 . €11 say ÷9 and ÷10 . Exercise 14. in the character table of G given at the end of Chapter 27. we obtain the characters ÷1 . (a) Compare the proof of Lemma 27. and so æ ˆ 2ø with ø irreducible. Since there are only six irreducible characters in all. respectively. g3 and g6 : g1 øA 6 g2 6 g3 2 g6 0 . The congruences ÷(1)  ÷(g6 ) mod 3 now give the remaining values on g6 . Use part (c) to ®ll in the values on g2 and g7 .1. we deduce that ÷ j (g3 ) ˆ 0 for 7 < j < 11.7). (e) € Theorem 22. . or æ is the sum of four distinct irreducible characters (cf. æ cannot therefore be the sum of four irreducible characters. (d) The characters 1 G . ÷ j (1) is even. (f ) By Proposition 19. the values of ÷ j (g6 ) for 7 < j < 11 must be Æ1.27 again. ø A has the following values on g1 .5 (noting that ÷ j (ÀI) Tˆ ÷ j (I) for 7 < j < 11. have degrees 2divisible by 6. say ÷7 (1) and ÷8 (1). Hence there is a character æ of G such that ÷ S ˆ 1 G ‡ ö ‡ ÷ ‡ æX The values of æ are as shown above. The only possibility is that two of ÷7 (1). ÷6 in the character table shown below. gi and gi g2 have the same centralizer for all i. and 12 ‡ 6 . æl ˆ 4.16.Chapter 27 (c) The values of ÷ and ÷ S are as shown below (see Proposition 19. g2 .14. Since By 11 2 jˆ1 (÷ j (g6 )) ˆ 6. . the (d) Since 11 ÷ j (g3 )÷ j ( g3 ) ˆ 8. By Corollary 22. ÷3 . jˆ1 (Alternatively. . The values of ø are as shown above. (c) Use Exercise 13. Now 1 G . 0. two of ÷7 . and none is a constituent of æ. so either æ ˆ 2ø for some irreducible character ø. ÷8 (1). in some order. ö.27.

thereby completing the character table of G. we ®ll in the values of ÷7 and ÷8 . we have 11 ÷ j ( g i )÷ j ( g 6 ) ˆ 0. ø(g5 ) ˆ 0. Then the value of ø A on g6 shows that ÷1 is not a constituent of ø A . ÷(g4 ) is real for all ÷. The column orthogonality relations now let us ®nd the remaining values of ÷9 and ÷10 . we get 8 p (x 2 À x)a2 ˆ øp( g8 ) ˆ À1. ÷4 . Then ÷8 ˆ ÷7 . and therefore. Let x ˆ ø(g8 ). by &  ' a b Ã. € (g) For i Tˆ 6. Similarly. This allows us to ®ll in the jˆ1 values of ÷11 . For all ÷. Say A ÷7 ( g 8 ) ˆ (1 À i 7)a2.bPZ Tˆ Z: a P Z 11 11 X 0 aÀ1 Then T is generated by     1 1 2 0 xˆ Z and y ˆ Z. using this fact and part (c). Since g2 is conjugate to g8 . Therefore x ˆ (1 Æ i 7)a2. Let Z ˆ {ÆI} and de®ne the subgroup T of G. ®nally. Then 4 €11 € and 11 ÷ j ( g 4 )2 ˆ 8 imply that jˆ1 ÷ j ( g 1 )÷ j ( g 4 ) ˆ 0 p jˆ1 p ÷9 (g4 ) ˆ À÷10 (g4 ) ˆ Æ 2. of order 55. the value on g3 forces ø A ˆ ÷6 . Hence 4 (ø( g4 )2 À ø( g 3 ))a2 ˆ ø A ( g4 ) ˆ ÷6 ( g 4 ) ˆ 0. 6. 7) gi g1 Order of g i 1 |CG ( g i )| 336 ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 ÷8 ÷9 ÷10 ÷11 1 7 8 3 3 6 4 4 6 6 8 g2 2 336 1 7 8 3 3 6 À4 À4 À6 À6 À8 g3 4 8 g4 8 8 g5 8 8 g6 3 6 g7 6 6 1 1 À1 0 0 0 À1 À1 0 0 1 g8 7 14 1 0 1 á á À1 Àá Àá À1 À1 1 g9 14 14 1 0 1 á á À1 á á 1 1 À1 g10 7 14 1 0 1 á á À1 Àá Àá À1 À1 1 g11 14 14 1 0 1 á á À1 á á 1 1 À1 1 1 1 1 À1 À1 À1 1 0 0 0 À1 À1 1 1 0 À1 1 1 0 2 0 0 0 0 0 0 1 0 0 0 1 p p 0 0 p 2 Àp 2 0 À 2 2 0 0 0 0 À1 p Note: á ˆ (À1 ‡ i 7)a2 The values of ø A on g1 and g2 show that ø A is a linear combination of ÷1 . Say ÷9 (g4 ) ˆ 2. ø(g4 ) ˆ 0. Now g2 is conjugate to g3 . Since g4 is conjugate to gÀ1 .442 Representations and characters of groups Character table of SL (2. 0 1 0 6 and T has ®ve linear characters æ j (0 < j < 4). ÷( g 10 ) ˆ ÷(g8 ). ÷5 and ÷6 . where æ j : x u y v 3 e2ði jva5 X .

10.16. But 8 (÷ j (1))2 ˆ 250. By Corollary 22. By considering 8 ÷ j (1)÷ j (g2 ) ˆ 0. We have hæ0 4 G. ÷8 (1) are 10. Next. 3 for all irreducible ÷.Chapter 27 443 The characters æ1 4 G and æ2 4 G are irreducible. and 8 ÷ j (g3 )2 ˆ 6. € Since ÷(1)  ÷(g3 ) mod 3 for all characters ÷. Note that ÷(gj ) is an integer for 1 < j < 4 and all characters ÷. note that e2ðia5 ‡ eÀ2ðia5 ˆ (À1 ‡ 5)a2X) Let ÷1 ˆ 1 G . without loss jˆ5 of generality. ÷6 (g2 ) have opposite signs. 2. and 8 ÷ j (g4 )2 ˆ 6. € 6 and ÷ j (g2 ) ˆ Æ1 for j ˆ 7. € Since ÷(g4 )  ÷(g2 ) mod 3 for all characters ÷. € 8 2 Now ÷(g4 )  ÷(g3 ) mod 2 for all ÷. ÷6 . ÷6 (1). Hence æ0 4 G ˆ ÷1 ‡ ÷2 for an irreducible character ÷2 of G. We may now conclude from the facts €8 2 that ÷(g2 )  ÷(g1 ) mod 2 and that ÷ j (g2 ) ˆ Æ2 for j ˆ 5. 8. namely ÷1 . by Theorem 22. ÷7 . ⠈ (À1 À p 5)a2 and 㠈 (À1 ‡ i 11)a2 € Since 8 ÷ j (g5 )÷ j ( g 5 ) ˆ 5. 5. ÷(g2 )  ÷(g4 ) mod 3 for all ÷. Character table of PSL (2. hence. æ0 4 Gi ˆ 2. and 8 ÷ j ( g2 )2 ˆ 12. ÷3 . we jˆ1 see that ÷7 (g2 ) ˆ ÷8 (g2 ) ˆ 1. ÷2 . the jˆ1 values of ÷ j (g3 ) (5 < j < 8) are as shown. 3 and 5 of the character table. € ÷ j (1)  0 mod 5 for 5 < j < 8.27. jˆ1 ÷ j (g2 ) ˆ 12. ÷5 (g2 ) ˆ 2 ˆ À÷6 ( g2 ). without loss of generality. so ÷ j (g4 ) ˆ Æ1 for 5 < j < 8. The column orthogonality relations now enable us to ®nish the character table. ÷8 take the value 0 on g5 . and ÷5 (g2 ). jˆ1 hence |÷( g2 )| . we deduce that the remaining irreducible jˆ1 characters ÷5 . 11) gi Order of g i |CG ( g i )| ÷1 ÷2 ÷3 ÷4 ÷5 ÷6 ÷7 ÷8 Note: á ˆ (À1 ‡ p g1 1 660 1 11 12 12 10 10 5 5 g2 2 12 1 À1 0 0 2 À2 1 1 g3 3 6 1 À1 0 0 1 1 À1 À1 p g4 6 6 1 À1 0 0 À1 1 1 1 g5 5 5 1 1 á â 0 0 0 0 g6 5 5 1 1 â á 0 0 0 0 g7 11 11 1 0 1 1 À1 À1 ã ã g8 11 11 1 0 1 1 À1 À1 ã ã 5)a2. . 5. we jˆ1 can complete column 4. they are ÷3 and ÷4 in the p table. ÷5 (1). ÷4 . and ˆ jˆ1 ÷ j (g4 ) € 6. We have now found four of the eight irreducible characters of G. ÷1 i ˆ 1 and hæ0 4 G. We have now completed columns 1. (In calculating ÷3 (g5 ). respectively. ÷7 (1).

3) is then as follows. since r ˆ r q and q is even. . q). It now follows easily that GL(2. q). q)  Z 3 SL(2.444 Representations and characters of groups Chapter 28 1. Then s in F q      a b s 0 aas bas ˆ X 0 s cas das c d The ®rst matrix in the product is sI and the second belongs to SL(2. indexed by unordered pairs fs. Every element r of F q can be expressed as a square. 0 1 (c) There are (q À 2)a2 conjugacy classes with representatives   s 0 d s.   a b Suppose that P GL(2. 3 2 3 2 3 2 0 2 0 1 1 2 g2 ˆ g3 ˆ g4 ˆ 1 0 2 0 1 0 3 2 3 2 3 2 0 0 1 0 1 0 g6 ˆ g7 ˆ g8 ˆ 2 2 0 1 2 1 where 3 1 2 1 1 3 X The character table of GL(2. q) where Z ˆ fsI : s P Fà g.s À1 ˆ . We take g 1 . X 2 1 g1 ˆ 0 2 1 g5 ˆ 0 X X .   1 1 (b) The matrix u1 ˆ has centralizer of order q. s À1 g of elements 0 s À1 from F q nF2 . Each such element has centralizer of order q À 1. gi |CG ( g i )| ë0 ë1 ø0 ø1 ø0. We may write ad À bc as s 2 for some c d à . g8 as representatives of the conjugacy classes. q) have representatives as follows.1 ÷1 ÷2 ÷4 g1 48 1 1 3 3 4 2 2 2 g2 48 1 1 3 3 À4 À2 2 À2 g3 6 1 1 0 0 1 À1 À1 À1 g4 6 1 1 0 0 À1 1 À1 1 g5 4 1 À1 1 À1 0 0 0 0 g6 8 1 1 À1 À1 0 0 2 0 g7 8 1 À1 À1 1 p0  i 2 0 p Ài 2 g8 8 1 À1 À1 1 0 p Ài 2 p0  i 2 2. (a) The identity I has centralizer of order q 3 À q. q You should have no dif®culty in proving that the conjugacy classes of SL(2.

b. 2 3 2 3 1 0 1 1 g1 ˆ g2 ˆ 0 1 0 1 2 3 2 3 2 3 ç 0 1‡ç 0 0 ç2 g3 ˆ g4 ˆ g5 ˆ 0 1 ‡ ç2 0 ç ‡ ç2 0 1 ‡ ç ‡ ç2 2 3 2 3 2 3 2 3 0 1 0 1 0 1 0 1 g6 ˆ g9 ˆ X g7 ˆ g8 ˆ 1 1 1 ç 1 ç2 1 ç ‡ ç2 We may choose a generator å of Fà so that å 7 ‡ å À7 ˆ ç.Chapter 28 445 (d) There are qa2 conjugacy classes with representatives   0 1 vr ˆ . By restricting characters from GL(2. The subscripts for ø0. x 2 ‡ ç2 x ‡ 1. g 7 . Hence we may write F8 ˆ fa ‡ bç ‡ cç2 : a. q) you will quickly be able to prove that the character table of SL(2. Note ®rst that PSL(2.s À1 1 1 s i ‡ s Ài 0 vr 1 À1 0 À(r i ‡ r Ài ) Here. g4 . x 2 ‡ (ç ‡ ç2 )x ‡ 1X The companion matrices for these quadratics give use the conjugacy class representatives g 6 . If q Tˆ 2 then the kernel of every non-trivial character is the identity subgroup. s À1 g of elements from F8 nF2 are fç. I ë0 ø0 ø0. Each such element has centralizer of order q ‡ 1. 8). g 5 below.3). fç2 . The irreducible monic quadratics over F8 with constant term 1 are x 2 ‡ x ‡ 1. ç ‡ ç2 gX These give us the conjugacy class representatives g3 . as follows. g8 . indexed by unordered pairs fr. g 9 of the conjugacy classes of SL(2. x 2 ‡ çx ‡ 1. 1 ‡ ç ‡ ç2 g. q) to SL(2. and therefore SL(2. 8). and the subscripts for ÷ i satisfy 1 < i < qa2. q) is as follows. Then 64 . The polynomial x 3 ‡ x ‡ 1 is irreducible over F2. 1 ‡ ç2 g. 8)  SL(2. 3. q) is simple. f1 ‡ ç. X X X .i ÷i 1 q q‡1 qÀ1 u1 1 0 1 À1 d s. r À1 g of elements 1 r ‡ r À1 from F q 2 nF q such that r 1‡q ˆ 1. g9 below We can now list representatives g 1 . we have used the function r 3 r de®ned in (28.i satisfy 1 < i < (q À 2)a2. c P F2 and ç3 ˆ 1 ‡ çgX The pairs fs.

446

Representations and characters of groups

å 14 ‡ å À14 ˆ ç2 , å 21 ‡ å À21 ˆ 1 and å 28 ‡ å À28 ˆ ç4 ˆ ç ‡ ç2 . The character table of SL(2, 8) is then as follows.
gi |CG ( g i )| ë0 ø0 ø0,1 ø0,2 ø0,3 ÷3 ÷1 ÷2 ÷4 g1 504 1 8 9 9 9 7 7 7 7 g2 8 1 0 1 1 1 À1 À1 À1 À1 g3 7 1 1 g4 7 1 1 A 0 0 0 0 0 0 0 0 0 0 0 0 g5 7 1 1 g6 9 1 À1 0 0 0 À2 1 1 1 g7 9 1 À1 0 0 0 1 g8 9 1 À1 0 0 0 1 B g9 9 1 À1 0 0 0 1

Here, the 3 3 3 submatrices A and B are given by H I 2 cos(2ða7) 2 cos(4ða7) 2 cos(6ða7) f g A ˆ d 2 cos(4ða7) 2 cos(6ða7) 2 cos(2ða7) e 2 cos(6ða7) 2 cos(2ða7) 2 cos(4ða7) H À2 cos(2ða9) À2 cos(4ða9) À2 cos(8ða9) I

f g B ˆ d À2 cos(4ða9) À2 cos(8ða9) À2 cos(2ða9) eX À2 cos(8ða9) À2 cos(2ða9) À2 cos(4ða9)

Chapter 29
1. (a) It is straightforward to check that ö is a homomorphism. For x, y P G, the element (x, y) P G 3 G sends x to y, so that action is transitive. (b) (G 3 G)1 ˆ f( g, g) : g P Gg, and ker ö ˆ f(z, z) : z P Z(G)g. (c) Every orbit of G 3 G on G 3 G contains an ordered pair of the form (1, x), and if ( g, h) sends (1, x) to (1, y) then g ˆ h and y ˆ g À1 xg. Hence if C1 , X X X , C k are the conjugacy classes of G, and xi P C i, then (1, xi ) (1 < i < k) are orbit representatives for the action of G 3 G on G 3 G, and so the rank is equal to k. Since x(( g, h)ö) ˆ x if and only if xhx À1 ˆ g, we see that ð( g, h) ˆ jfix G ( g, h)j is equal to 0 if g is not conjugate to h, and is equal to jC G (x)j if g is conjugate to h (since in the latter case, if xhx À1 ˆ g then an arbitrary element y P G such that yhy À1 ˆ g is of the form y ˆ xc with c P € (x)). Hence using the column orthogonality relations we see CG that ð ˆ ÷ 3 ÷. 2. There are q 2 À 1 non-zero vectors in V, and each 1-dimensional subspace

Chapter 29

447

contains q À 1 of them; also two 1-dimensional subspaces have no non-zero vectors in common. Hence jÙj ˆ (q 2 À 1)a(q À 1) ˆ q ‡ 1. 3. Use the notation for the conjugacy classes and irreducible characters of GL(2, q) given in Proposition 28.4 and Theorem 28.5. It is easy to check that ð takes the values q 2 À 1, q À 1, q À 1 on the classes with representatives I, u1 , d 1, t respectively, and takes the value 0 on all other classes. Taking inner products we ®nd hð, 1 G i ˆ hð, ø0 i ˆ hð, ø0, j i ˆ 1 (1 < j < q À 2)X € qÀ2 As 1 G ‡ ø0 ‡ 1 ø0, j has degree q 2 À 1 ˆ ð(1), we conclude that € qÀ2 ð ˆ 1 G ‡ ø0 ‡ 1 ø0, j . 4. Observe that the coset H 1 x is ®xed by g if and only if xgx À1 P H 1 . If G is abelian this amounts to g P H 1 , and hence we see that ð1 ( g) ˆ 0 if g P H 1 a and ð1 ( g) ˆ jG : H 1 j if g P H 1 . Thus H 1 ˆ f g P G : ð1 ( g) Tˆ 0g. Likewise for H 2 ; since ð1 ˆ ð2 we deduce that H 1 ˆ H 2 . As a counterexample for G non-abelian, take G ˆ D8 ˆ ha, b : a4 ˆ b2 ˆ 1, bÀ1 ab ˆ aÀ1 i with H 1 ˆ hbi, H 2 ˆ ha2 bi. Then ð1 ˆ ð2 but H 1 Tˆ H 2 . € € 1 5. By Proposition 29.4 we have 1 ˆ jGj gPG jfixÙ ( g)j, hence jfixÙ ( g)j ˆ jGj. Since jfixÙ ( g)j is a non-negative integer for each g, and jfixÙ (1)j ˆ jÙj . 1, we must have jfixÙ ( g)j ˆ 0 for some g. 6. Write ð ˆ ð( nÀ2,1,1) . Calculating inner products using Proposition 29.6, as in the proof of Theorem 29.13, we ®nd hð, ði ˆ 7, hð, 1i ˆ 1, hð, ð( nÀ1,1) i ˆ 3, hð, ð( nÀ2,2) i ˆ 4X Using Theorem 29.13 it follows that ð( nÀ2,1,1) ˆ 1 ‡ 2÷ ( nÀ1,1) ‡ ÷ ( nÀ2,2) ‡ ÷ with ÷ irreducible. Hence ÷ ˆ 1 ‡ ð À ð( nÀ1,1) À ð( nÀ2,2) , from which it is easy to calculate that ÷(1) ˆ 1(n À 1)(n À 2), 2 ÷(12) ˆ 1(n À 2)(n À 5), ÷(123) ˆ 1(n À 4)(n À 5). For n ˆ 6, ÷ (4,1,1) is the 2 2 character ÷5 in Example 19.17.

Chapter 30
1. By Theorem 30.4, a245 ˆ 168/(8´3) ˆ 7. Hence, by (30.3), PSL (2, 7) contains elements a and b such that a has order 2, b has order 3 and ab has order 7. p p 2. No: a225 ˆ (1 ‡ (À1 ‡ i 7)a6 ‡ (À1 À i 7)a6 À 4a6)168a(8X8) ˆ 0, and similarly a226 ˆ 0. Hence PSL (2,7) does not contain two involutions whose product has order 7. 3. Yes. Number the conjugacy classes of PSL (2, 11) as in the solution to Exercise 27.6. Then   660 1 a235 ˆ X 1‡ ˆ 10X 12 6 11 Therefore PSL (2, 11) contains elements x and y such that x, y and xy have orders 2, 3 and 5, respectively. Let H be the subgroup kx, yl of PSL (2, 11).

448

Representations and characters of groups

There is a homomorphism W from A5 onto H (W sends a 3 x, b 3 y). Since Ker W 3 A5 and A5 is simple, we deduce that H  A5 . 4. Suppose that G is a group whose p character table is p where á ˆ (1 ‡ 5)a2, ⠈ (1 À 5)a2.
g1 ÷1 ÷2 ÷3 ÷4 ÷5 1 4 5 3 3 g2 1 1 À1 0 0 g3 1 0 1 À1 À1 g4 1 À1 0 á â g5 1 À1 0 â á

€ For 1 < i < 5 we have jC G ( g i )j ˆ 5 ÷ j ( g i )÷ j ( g i ). Therefore the jˆ1 centralizers of g1 , g2 , g3 , g4 , g5 have orders 60, 3, 4, 5, 5, respectively. Hence the orders of g2 , g4 and g5 are 3, 5 and 5; also the order of g3 must be 2, since for no other i (except i ˆ 1) is |CG (g i )| even. Now a324 ˆ 60a(4 . 3). Therefore G contains elements x and y such that x has order 2, y has order 3 and xy has order 5. As in the solution to Exercise 3, G has a subgroup H with H  A5 . Since jGj ˆ 60, we have G  A5 . € 5. (a) Using the fact that 7 ÷ j (g i )÷ j ( g i ) ˆ |CG (g i )|, we ®nd that the jˆ1 centralizer orders and class sizes are as follows:
g1 |CG ( g i )| | gG | i 360 1 g2 8 45 g3 4 90 g4 9 40 g5 9 40 g6 5 72 g7 5 72

Hence jGj ˆ 360. Also G is simple, by Proposition 17.6. (b) By the Frobenius±Schur Count of Involutions (Corollary 23.17), the number of involutions t in G is bounded by 1‡ t <
7 ˆ jˆ1

÷ j (1) ˆ 46X

By considering jC G ( g i )j, we see that g i has even order only for i ˆ 2 and 3. Since t < 45, only g2 can be an involution. Hence g3 has order 4. (This information about the orders of g2 and g3 can also be deduced using Sylow's Theorem.) (c) Clearly g6 and g7 have order 5, and at least one of g4 and g5 has order 3. If j ˆ 4 or 5 and k ˆ 6 or 7 then

Chapter 30
a2 jk ˆ ˆ ˆ ÷( g2 )÷( g j )÷ ( g k ) jGj jCG ( g 2 )j jCG ( g j )j ÷ ÷(1)

449

360 ˆ 5X 8.9 Therefore G contains elements x and y such that x has order 2, y has order 3 and xy has order 5. As in the solution to Exercise 3, the subgroup H ˆ kx, yl of G is isomorphic to A5 . (d) If g, h P G then ( gh)r: Hx 3 Hxgh, and ( gr)(hr): Hx 3 Hxg 3 HxghX Hence r is a homomorphism. (e) Since G is simple, Ker r ˆ {1}. Hence G is isomorphic to a subgroup K of S6 . Since jS6 : Kj ˆ 63a360 ˆ 2, K must be A6 . 6. Consider the ®gure in Example 30.6(3). We shall explain how to label the vertices by elements of G. Choose a vertex and label it 1. Label the vertices according to the following inductive rule. Assume that v is a vertex and an adjacent vertex u is labelled by g. Then label v by ga gb gb
À1

if the edge uv has no arrow, if the edge uv has an arrow from u to v, if the edge uv has an arrow from v to uX

For example, if you decided to label the bottom left-hand vertex by 1, then part of the labelling would be

The relation a2 ˆ 1 ensures that the labelling is consistent along unmarked edges; since b3 ˆ 1, the labelling is consistent around triangles; and the relation abababab ˆ 1 deals with the octagons. Every element in G has the form given by the label of one of the 24 vertices, so jGj < 24. 7. The conjugacy classes of PSL(2, 7) are given in Lemma 27.1 The element g 2 is an involution with centralizer of order 8 given in the proof of     2 2 2 4 Lemma 27.1. Letting a ˆ , bˆ , we see that the À2 2 4 À2 centralizer is generated by a and b, and a4 ˆ b2 ˆ 1, bÀ1 ab ˆ aÀ1 À, hence the centralizer is isomorphic to D8 . As in Exercises 12.3 and 12.4, we see that C A6 ((12)(34)) has order 8 and is generated by (1324) and (13)(24), hence as above is isomorphic to D8 .

450

Representations and characters of groups

8. Let G be the simple group PSL(2, 17). In the ®eld Z17 the element 4 is a   4 0 fourth root of unity, so t ˆ Z is an involution. Calculate that 0 À4 C G (t) is generated by the group of diagonal matrices together with     0 1 3 0 bˆ Z, hence is generated by b and a ˆ Z. As À1 0 0 6 a 8 ˆ b2 ˆ 1 and bÀ1 ab ˆ aÀ1, we have C G (t)  D16 .

Chapter 31
1. Assume that G has an abelian subgroup H of index p r ( p prime), and that |G| . p. If H ˆ {1} then |G| ˆ p r and so G is not simple (see Lemma 26.1(1)). So assume that H Tˆ {1}; pick 1 Tˆ h P H. Then H < CG (h) as H is abelian, so |G:CG (h)| is a power of p. If |G:CG (h)| ˆ 1 then khl v G and G is not simple. And if |G:CG (h)| . 1, then G is not simple by Theorem 31.3. 2. By Burnside's Theorem, jGj is divisible by at least three distinct primes. Since 3 . 5 . 7 . 80, jGj is even. Then by Exercise 13.8, |G| is divisible by 4. Since 4 . 3 . 7 . 80, the only possibility is that jGj ˆ 4 . 3 . 5 ˆ 60.

Chapter 32
1. (a) The fact that BB ˆ I follows from the observation that for all i, j, d(ei b, ej b) ˆ d(ei , ej ) ˆ ä ij X Since 1 ˆ det I ˆ (det B)(det Bt ) ˆ (det B)2 , we have det B ˆ Æ1. (b) (i) The eigenvalues of C are the roots of det (C À xI), which is a cubic polynomial over R. Therefore, C has one or three real eigenvalues. (ii) Moreover, the product of the eigenvalues of C is det C ˆ 1. If C has three real eigenvalues, then they cannot all be negative; and if C has one real eigenvalue ë and a pair of conjugate non-real eigenvalues ì, ì, then ë ìì ˆ 1, and hence ë . 0. Therefore C has a real positive eigenvalue, say ë. (iii) Let v be an eigenvector for ë. Then d(v, v) ˆ d(vC, vC) ˆ d(ëv, ëv) ˆ ë2 d(v, v), and so ë ˆ 1. (c) Let c be the isometry v 3 vC. By (b), c ®xes a vector v; it is now easy to convince yourself that c must be a rotation about the axis through v. The required result for b follows from the de®nition of c. (d) Take three orthogonal axes, one of which is the axis of the rotation b. With respect to these axes, the matrix of b is H I 1 0 0 d 0 cos ö sin ö eX 0 Àsin ö cos ö Hence tr B ˆ 1 ‡ 2 cos ö.
t

Chapter 32

451

2. We regard G as a subgroup of O(R3 ). It is easy to see that the translation submodule T (which consists of all the translation modes) is isomorphic to the RG-module given by the natural action of G on R3 . Hence by part (d) of Exercise 1, V if g is a rotation through ö, about some b 1 ‡ 2 cos ö, b b b b axis, b ` ÷ T ( g) ˆ b À(1 ‡ 2 cos ö), if the element À g of O(R3 ) is a rotation b b b b b X through öX Now consider the rotation submodule R, which consists of all the rotation modes. A rotation mode is speci®ed by a 3-dimensional vector öv, where v is a unit vector along the axis of the rotation and ö denotes the angle of rotation, taken positive in the right-hand screw sense. Let g P G, and consider g acting on öv. It sends v to vg, and if g is a rotation, it preserves the sense of the rotation; however, if g is a re¯ection then it transforms a right-hand screw to a left-hand screw, and hence sends öv to (Àö)(vg). Therefore @ if g is a rotation, ÷ T ( g), ÷ R ( g) ˆ À÷T ( g), if g is not a rotation, and so ÷ T ‡ ÷ R has the required form. 3. The matrix A is I H p p À3a2 0 3a4 À 3a4 p3a4 3a4 p f 0 À1a2 À 3a4 p1a4 3a4 1a4 g g f p k f 3a4 À 3a4 À3a4 3a4 0 0 g gX f p p 3a4 À5a4 0 1 g m f À 3a4 p1a4 g f p d 3a4 3a4 0 0 À3a4 À 3a4 e p p 3a4 1a4 0 1 À 3a4 À5a4 4. A simpler basis is given by
1 2(r1 1 2(r1 1 2(r1

‡ r2 ) ˆ (v12 ‡ v21 ) À (v34 ‡ v43 ), ‡ r3 ) ˆ (v13 ‡ v31 ) À (v24 ‡ v42 ), ‡ r4 ) ˆ (v14 ‡ v41 ) À (v23 ‡ v32 )X

We chose r1, r2, r3, r4 to be the images of w1, w2, w3, w4 under an RGisomorphism. (Compare the construction of the matrix B, towards the end of Example 32.20.) 5. (a) This is a routine geometrical exercise. (b) Let the displaced positions of the atoms be 09, 19, 29, 39, 49. The distance of 19 from the plane through 1 perpendicular to 12 is x12 ‡ 1(x13 ‡ x14 ). Similarly, the distance of 29 from the plane through 2 2 perpendicular to 12 is x21 ‡ 1(x23 ‡ x24 ). Therefore 12 has decreased by 2 x12 ‡ x21 ‡ 1(x13 ‡ x14 ‡ x23 ‡ x24 ). The other calculations can be done in 2 the same way. (c) We express the force at each atom as a vector, and then write this vector

452

Representations and characters of groups
as a linear combination of our three chosen unit vectors at the initial position of the atom. Let d ij denote the decrease in the length ij, as calculated in part (b). Then, for example, at vertex 1, the contributions to the component of the force vector in the direction 12 are as follows: Àk1 d12 from the force between atoms 1 and 2; zero from the force between atoms 1 and 3 and from the force between atoms 1 and 4; and Àk2 d10 sec (/012) from the force between atoms 1 and 0. Hence p m1 12 ˆ Àk 1 d 12 À 1 (3a2)k 2 d 10 X x 3

Upon substituting for d12 and d10 from part (b), we obtain the given expression for 12. x The other accelerations are calculated in the same way. (d) The entries in the 15 3 15 matrix A are the coef®cients which appear in the equations of motion  ˆ xA. If you write down the matrix A, then x you will easily verify that the given vectors are eigenvectors of A (with eigenvalues À(4k 1 ‡ k 2 )am1 , 0, 0, 0, Àk 1 am1 , Àk 1 am1 , p p I À2k 2 a3m1 À4k 2 p 2a(m2 p3) À2k 2 a3m1 p À4k 2 2a(m2 3) eX p À2k 2 3a(9m1 2) À4k 2 a3m2 p (f ) You will ®nd that (1, À2, 6) is an eigenvector of B, with eigenvalue 0. This agrees with the statement in Example 32.20 that the translation vector r1 À 2s1 ‡ 3 cos Ww1 is an eigenvector of A. 6. (a) Assign coordinate axes along the edges of the square, as shown below. respectively). (e) The matrix B is H À(6k 1 ‡ k 2 )a3m1 d À(3k 1 ‡ k 2 )a3m1 p p Àk 2 3a(9m1 2)

The symmetry group is G ˆ D8. Let a denote the rotation sending P 3 Q 3 R 3 S 3 P, and let b denote the re¯ection in the axis PR. The character ÷ of the RG-module R8 is

1 ÷ 8

a2 0

a 0

b 0

ab 0

Chapter 32

453

We refer to the character table of D8 which is given in Example 16.3(3), and see that ÷ ˆ ÷1 ‡ ÷2 ‡ ÷3 ‡ ÷4 ‡ 2÷5 X The rotation mode is (t ‡ â)v, where v ˆ (1, À1, 1, À1, 1, À1, 1, À1) P V÷2 X The translation modes are (t ‡ â)v, where v is in the span of v1 , v2 and v1 ˆ (1, 0, 0, 1, À1, 0, 0, À1), v2 ˆ (0, 1, À1, 0, 0, À1, 1, 0)X The homogeneous components V÷1, V÷3 and V÷4 are spanned by the vectors (1, 1, 1, 1, 1, 1, 1, 1), (1, 1, À1, À1, 1, 1, À1, À1) and (1, À1, À1, 1, 1, À1, À1, 1), respectively. The ®nal set of eigen-vectors is given by V÷5 ’ R8 which is spanned by vib (1, 0, 0, À1, À1, 0, 0, 1) and (0, 1, 1, 0, 0, À1, À1, 0)X (b) The matrix A is 1 f0 f f0 f k f0 À f mf0 f f0 f d0 1 H 0 1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 I 1 0g g 0g g 0g gX 0g g 0g g 0e 1

7. (a, b) Let å i (1 < i < m) be the projection which is given by å i : u1 ‡ X X X ‡ um 3 ui (where u k P Uk for all k). Then w 3 wAå j WÀ1 W i j (w P Ui ) gives an RG-homomorphism from Ui to Ui. By Exercise 23.8, there exist ë ij P R such that for all w P Ui, Since €m wAå j ˆ ë ij wWÀ1 W j X i is the identity endomorphism of U1 È . . . È Um, we have m ˆ ë ij wWÀ1 W j for all w P Ui X wA ˆ i
jˆ1 jˆ1 å j

Now take in turn w ˆ uW i and w ˆ vW i to obtain the results of parts (a) and (b) of the question. (c) Take a basis u1 , . . . , u n of U1. Assume that the eigenvectors of A u are known. For all k with 1 < k < n, the eigenvectors of A in sp (u k W1 , . . . , u k W m ) are given by the eigenvectors of A u (see part (b)). Hence we know all the eigenvectors of A in U1 È . . . È Um.

Bibliography

Books mentioned in the text
H. S. M. Coxeter and W. J. O. Moser, Generators and Relations for Discrete Groups (Fourth Edition), Springer-Verlag, 1980. J. B. Fraleigh, A First Course in Abstract Algebra (Third Edition), AddisonWesley, 1982. D. Gorenstein, Finite Simple Groups: An Introduction to their Classi®cation, Plenum Press, New York, 1982. G. D. James, The Representation Theory of the Symmetric Groups, Lecture Notes in Mathematics No. 682, Spring-Verlag, 1978. D. S. Passman, Permutation Groups, Benjamin, 1968. H. Pollard and H. G. Diamond, The Theory of Algebraic Numbers (Second Edition), Carus Mathematical Monographs No. 9, Mathematical Association of America, 1975. J. J. Rotman, An Introduction to the Theory of Groups (Third Edition), Allyn and Bacon, 1984. D. S. Schonland, Molecular Symmetry ± an Introduction to Group Theory and its uses in Chemistry, Van Nostrand, 1965.

Suggestions for further reading
M. J. Collins, Representations and Characters of Finite Groups, Cambridge University Press, 1990. C. W. Curtis and I. Reiner, Methods of Representation Theory with Applications to Finite Groups and Orders, Volume I, Wiley-Interscience, 1981. W. Feit, Characters of Finite Groups, Benjamin, 1967. I. M. Isaacs, Character Theory of Finite Groups, Academic Press, 1976. W. Ledermann, Introduction to Group Characters (Second Edition), Cambridge University Press, 1987. J. P. Serre, Linear Representations of Finite Groups, Springer-Verlag, 1977.

454

116. 172. 11. 354. 24 character. 443 Q8 . 240. 195 generalized. 55. 2 Cn . 122. 106 centre of group. 111 antisymmetric part.Index A4 . 160 D8. 122 character table. 112. 416 S4 . 5. 244. 417 F11. 269 symmetric. 128. 230. 85. 116. 445 T12. 116. 263 realized over R. 54 bijection. 136. 440 SL(2. 122. 88 centralizer.3). 129 product. 45. 363. 114. 150 trivial. 6 bilinear form. 160 C3 . 11. of order 18. 412 C2 3 C2 . 182 D2n ( n even). 269 Brauer±Fowler Theroem. 107. 207. 196. 192 real. 153 change of basis. 82. 160 C4 . 82 D6. 5. 327 PSL(2. 359 A6 . 247 faithful. 278 Burnside's Lemma. 253 irreducible. 308 A5 . 2 basis. 83. 2. 442 SL(2. 364 C.5 . 318 PSL(2. 161 D10.q .11). 269 skew-symmetric. 205 SL(2.7).7). 119 regular. 118 degree. 312. 298 of group algebra. 236 integer-valued. 423 E. 340 Burnside's Theorem. 127. 222. 10.8). 81. 419 D2n ( n odd). 181 A5 . 415 D12  S3 3 C2 . 234. 445 PSL(2. 433 Fp. 82 action. 130.3. 186 T4 n.q). 111. 15 natural. 180 S5 . 337 algebra. 220. 424 C2 . 265 reducible. 343 abelian group. 176. 9. 125. 159 A4 . 420 455 . 415 Cn . 273 associative. 11. 359 A6 . 355 induced. 112. 434 F7. 3. 291 GL(2. 174 permutation. 362 algebraic number. 181. 223 An .q). 9. 360 A7 . 125 linear. 262 S6 . 221. 201. 183 D6 3 D6. 361 alternating group. 359. 56 algebraic integer. 119 kernel of.

302. 82 alternating. 3 GL(2. 12. 8 cycle-shape. 143. 24 eigenvector. 18 F ˆ R or C. 206 direct sum. 234. 107 class function. 50. 91. 206 order 16. 376 homomorphism. 311 symmetric. 2 composition factor. 15 direct product. 2 general linear. 290 FG. 2. 181 factor. 312 quaternion. 61 FG-isomorphism. homogeneous component. 55 H < G. 324. 361 constituent. 226. 343 general linear group. 20.456 Representations and characters of groups FG-submodule. 306. 213 coset. 216 complete set. 66 external. 308 order p3 . 3 F n. 195 faithful module. 91 irreducible representation. 63 U6n. 10. 85 faithful representation. 187. 125. 8 cycle notation. 74 composition. 173 diagonalization. 6 GL(n.q). 249 derived subgroup. 4. 1 abelian. 95. 104. 96 congruences. 82. 18 eigenvalue. 134 involution. 116. 104 conjugate. 88 dicyclic. 109. 3 H v G. 353. 254 symmetry. 318. 368 simple. 24 endomorphism. 9 indicator function. 5. 90 common. 9 HomCG (V W). 228 inner product. 46 even permutation. 44. 365 special linear. 300 class algebra constants. 2. 20 equivalent. 256 maximal. 6 invertible. 304 orthogonal. 79 isomorphism. 12. 30. 236 induced module. 364 soluble. 107. 181 dimension. 101 completely reducible. 61 ideal. 2 order p3 . 88 D2n. 6 bijective. 367 projective special linear. 3. 12. 311. 56. 32. 122. 82. 119 irreducible module. 368 group algebra. 307 order 27. 12. 12. 232 Frobenius±Schur Count of Involutions. 305. 109 cyclic group. 421 direct product. 3 group. 290 Frobenius Reciprocity Theorem. 11. 4. 273 induced character. 278. 2. 301 order pq. 114 Clifford's Theorem. 230. 6 injective. 250. 81. 63 Frobenius group. 363. 257 index of subgroup. 6 surjective. 3. 9 ®nite. 15 Fp. 381 external direct sum. 277 function. 420 dihedral group. 50. 259 conjugacy class. 9 faithful character. 291 p-group. see module . 34 FG-module. 49 FG-homomorphism. 152 class sum. 17. 107. 2. 53 factor group.q . 79. 435 order . 6. 178.F). 421 V24 . 178 dihedral. 32. 175. 5 rotation. 7. 74. 3 order. 5. 257 proper. 83 dicyclic group T4 n. 111 cyclic. 349 class equation. 107. 277. 10. 281. 2. 422 V8 n . 5 expansion±contraction mode. 96 . 11. 9. 181 degree. 353 irreducible character.

56. 127. 319. 50 regular character. 150 regular module. 122. 43 natural normal normal normal 217 basis. 5 orbit. 113. 344 SL(2. 210 rotation group. 5 permutation module. 79 kernel of.3). 18 linearly dependent. 319. 278. 3 rank. 45 powers of characters. 201. 311 stabilizer. 416 R. 340 permutation character. 440 SL(2. 21 change of basis. 321. 311. 129. 180. 62. 336. 3. 312 457 Q8 . 311 SL(2. projection. 19. 9 lift. 54 modes of vibration. 177. 161 PSL(2. 171. 275 S5 . 56 representation. p). 360 PSL(2. 171.7). 50 reducible representation. 3 primitive root. 363. 278. 343. 85 irreducible. 45. 124. 269 special linear group. 354. 258 permutation. 15 Maschke's Theorem. 256. 262 S6 . 34 representatives. 263 reducible character. 26 identity. 175. 173. 27. 67 projective special linear group. 338 order of G. 70. p). 176. 5. 193 presentation. 368 rotation mode. 10. 320. 251 subgroup. 45. 394 S4 . 19 real character. 50 regular. 30. 9. 263 real conjugacy class. 4 normal. 128. 76 matrix. 39 completely reducible.q). 3. 109. 116. 116. 249 equivalent. 50 regular. 250. 50. 110. 367 orthogonality relations. 113. 34 irreducible. 56 trivial. 74 faithful. 365 symmetric bilinear form. 45 methane. 74 Sylow's Theorem. 364 skew-symmetric bilinear form. 217 submodule. 379 rotation submodule. 21 invertible. 263 real element. 342 Rank±Nullity Theorem. 174 linear transformation. 5 odd. 113. 4 cyclic. 10. 318. 384 minimal polynomial. 205 S7 . 373 p-complement. 46 faithful. 105 restriction. 24 diagonal. 177. 4 orthogonal group. 45. 85 permutation. 15 linearly independent. 116. 192 . 169 linear character. 215. 339 subgroup. 116. 312 p-group. 3.11). 2 order of g. 223 Sn . 56 regular representation. 78 simple group. 4 derived. 23 permutation. 79. 44. 111. 318. 3. 5 even. 361 module. 216. 124 reducible. 442 SL(2. 34. 119 reducible module. 354. 5. 173 generated. 44. 56 trivial. 30 degree. 50. 9. 416 quaternion group. 34. 380. 269 odd permutation. 216. 254. 278. 445 Schur's Lemma.7). 49 irreducible. 32. 298 p9-part. 359 PSL(2. 215. 372. 359. 125 Lagrange's Theorem.Index kernel. 312. 284 product of characters. 340 permutation matrix. 62 reducible.

178. 3. 34 U6n. 187. 175. 117 transitive. 374 Z. 368 T4 n. 187. 338. 194 vibratory modes. 381 water. 196. 178. 190 tensor product space. 188 trace. 369. 254 symmetric part. 380. 420 tensor product module. 379 translation submodule. 394 transposition. 5 . 421 V8n. 281. 2 symmetric group.458 Representations and characters of groups trivial character. 229 translation mode. 116. 109. 122 module. 43 representation. 421 Vandermonde matrix. 273 symmetry group. 178. 341 transitivity of induction. 187.

Sign up to vote on this title
UsefulNot useful