Lecture Notes in Mathematics

A collection of informal reports and seminars Edited by A. Dold, Heidelberg and B. Eckmann, Z(Jrich

240 Adalbert Kerber

Mathematisches Institut der Justus Liebig-Universit~it Giessen/Deutschland

Representations of Permutation Groups !

$
Springer-Verlag Berlin-Heidelberg- New York 1971

AMS Subject Classifications (1970): 20C30

ISBN 3-540-05693-9 Springer-Vedag Berlin. Heidelberg- New York ISBN 0-387-05693-9 Springer-Verlag New York Heidelberg. Berlin
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically those of translation, reprinting, re-use of illustrations, broadcasting, reproduction by photocopying machine or similar means, and storage in data banks. Under 54 of the German Copyright Law where c0pi~ ace made for other than private use, a fee is payable to the publisher, the amount of the fee to be determined by agreement with the publisher. @ by Sptinser-Verlag Berlin : Heidelberg 1971. Librm7 of Congress Catalog Card Number 72-1839~ Printed in Germany.

R e p r e s e n t a t i o n s o f P e r m u t a t i o n Groups

Part l R e p r e s e n t a t i o n s o f Wreath P r o d u c t s and A p p l i c a t i o n s t o t h e R e p r e s e n t a t i o n ~ h e o r y o f Symmetric and A l t e r n a t i n g Groups

Preface

As a contribution to the theory of representations of permutation groups the theory of representations of wreath products of finite groups is discussed in this first part with subsequent applications to the theory of representations of symmetric and alternating groups. The intention is to give a new description and a further development of the representation theory of symmetric and alternating groups and this will be carried on in the following parts. This seems desirable since following the appearance of the only comprehensive treatment of this theory, namely G.de B.Robinson's book "Representation Theory of the Symmetric Group" (Toronto 1961) a number of papers have been published which continued this work. Moreover some of these papers contain results which allow generalizations which connect this theory more closely with the general representation theory of finite groups. The representation theory of symmetric and alternating groups is s11mmarized as far as is needed here, while a knowledge of the main results of the general representation theory of finite groups over fields is assumed. The results of the representation theory of the symmetric group whose proofs are omitted here will be treated in detail in the following parts. I would express my sincerest th~nks to Prof.H. Boerner, Prof. H.K. Farahat, Dr. M.H. Peel and Prof. G. de B.Robinson to whom I am greatly indebted for very helpful discussions and stimulating encouragement.

Adalbert Kerber

Contents

Introduction

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I @ 5
. . .24 . .

CHAPTER I: Wreath products of groups

I. Permutation groups . . . . . . . . . . . . . . . . . . . . . 2. Wreath products

. . . . . . . . . . . . . . . . .

3, Wreaths with symmetric groups CHAPTER II: Representations

. . . . . . . . . . . . . . . .......... of the symmetric

39 59

of wreath products

@. The ordinary irreducible representations group 5. Representations of wreath products

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . theory of

60 89

CHAPTER III: Application to the representation

symmetric and alternating groups . . . . . . . . . . . . . . 6. Symmetrized outer products of irreducible C-representations of symmetric groups . . . . . . . . . . . . . . . . . . . . 7. Block-structure and decomposition numbers of symmetric and alternating groups . . . . . . . . . . . . . . . . . . . 8. Generalized decomposition numbers of symmetric and alternating groups . . . . . . . . . . . . . . . . . . . . . . . .
References 8ubJect-Index . . ..
.

11@

116 130 168

.
.

.
.

.
. .

.
.

.
.

182

191

Introduction

The derivation of the representation theory of wreath products provides a nice example of the utility of Clifford's theory of representations of groups with normal divisors. Applications of this theory to the representation theory of symmetric and alternating groups arise from the fact, that centralizers o3 elements, normalizers of certain subgroups, Sylow-subgroups as well as defect groups are all direct products of wreath products. Thus for example the theory of the generalized decomposition numbers as well as the theory of symmetrized outer prodmots of irreducible ordinary representations of symmetric groups can be described with the aid of this theory. On the other hand, representations of wreath products with symmetric groups can be described in detail by using the theory of representations of the symmetric group. But nevertheless this is no vicious circle, since the degrees m of the symmetric factors Sm of the applied wreath products G~S m satisfy m<n, if n is the degree of the considered symmetric group S n. Thus on the contrary these applications provide interestimg recursion processes.

Besides this description of representations of wreath products and some of their applications some new results on the modular representation theory of symmetric and alternating groups are cited or proved especially with reference to the theory of decom-

position numbers. It should be mentioned that wreath products are involved in some combinatorial and graph-theoretical devices and the new interest

of physicists shouldn't be forgotten (see the references).

The first examples of wreath products can be found in A. Cauchy's

I . "Exercises d'analyse et de physique mathematlque H

(vol III, 1844)

in E. Netto's "Substitutionentheorie

und ihre Anwendungen auf die

Algebra" and in A. Radzig's dissertation entitled "Die Anwendung des Sylow'schen Satzes auf die symmetrische und die alternirende Gruppe" (1895). Wreath products arise in these cases in connection with the construction of Sylcw-subgroups tric group. The first representation-theoretical consideration of wreath proof the considered symme-

ducts was given by A. Young, who applied his method of deriving the representation hyperoctahedral hyperoctahedral theory of the symmetric group to the so-called

group in 1930 (Young CI~). In modern notation the group is a wreath product $2~S n. W. Specht con-

sidered wreath products of the more general form G~S n (G a finite group) in his dissertation ordinary representation (Specht CI~, 1932) and described their

theory which he generalized to products

of the form G~H (G completely reducible, H a subgroup of Sn) in 1933 (Speoht E2~). Such groups had already appeared in papers of A. Loewy (Loewy EI~, 1927), A. Scholz (Scholz ~I~, 1930) and B. Neumann (Neumann ~I~, 1932). In his paper "Kombinatorische Anzahlbestimmungen pen, Graphen und chemische Verbindungen" fGr Grup-

G. Polya suggested for

G~H the name "G-Kranz um H", of which "G-wreath around H" is a

translation. Wreath products G~S n were also considered by O. Ore

(Ore [1]) and by R. Pz~oht (~zmcht [1]) in 1942.

Since then numerous papers on these groups as well as on the representation theory of special cases have been published (see the papers of Oeima, Puttaswamaiah, Kerber)$ references to some applications to combinatorics, graph theory and physics may be fou~ at the end of these notes.

Chapter I

Wreath products of groups

In the first section, by way of fixing the notation, we give results concerning the symmetric group Sn ( ~ Cl}~S n ~ Sn~Cl]) and the alternating group A n ( ~ [ I ] ~ n ~ An~[1]). In the second section general wreath products G~H are introduced and in the third section attention is restricted to wreath products G~H with H=S n and G a finite group.

I. Permutation groups

A bijective mapping of a set 6 onto itself is called a permutation o_~f 6. If a set of permutations position multiplication of a set 6 together with the com-

is a group, we call this group a permuof 6 is

tation group on G. The group SG of all the permutations called the s,ymmetric group on 6.

The symmetric groups on two finite sets 6' and 6" of the same order n = 16'I = IG"I are obviously isomorphic. Hence we may deI~I = n by

note the symmetric groups of finite sets 6 of order Sn and assume that 6 = ~1,...,n~. symbols. The order of Sn is 1.1 as is well known. Subgroups
ISnl = n!

The elements

of 6 are called

of Sn are called permutation

groups

of de~ree n; their elements are called permutations of degree n. A permutation ~ E Sn (on 6 = ~1,...,n~)is written down in full

by putting the images ~(i) in a row under the symbols i E G, for example
1 n

for short:
=

(il

In accord with the notation ~(i) for the image of the symbol i under the permutation ~, products of permutations have to be read from the right to the left:
:= .

For subsets ~' E Q let

:=
A permutation of the form

( ix i2 "'" ir-X ir ir+1 "ii in) i2 i3 ..- ir iI Jr+ I in

is called cyclic or a cycle. To emphasize the number of symbols which are moved by this cycle, we call it an r-cycle. More briefly we write (il..-i r) , where the l-cycles (Jr+l),... , (in) on the symbols which remain fixed have been omitted. The identity element of Sn, the permutation which consists only of l-cycles will be denoted by I or by ISn. The following is obviously valid: 1.2 This means (il...i r) = (i2...iri I) . . . . . (iril...ir_ I)

that a cycle which arises from a given one by cylical-

ly permuting the symbols describes the same permutation. 2-cycles, i.e. permutations which move exactly two symbols of ~, are called transpositions. The order of a cycle, i.e. the order of the generated cyclic

subgroup <(il...ir)> ~ Sn, iS equal to its length: 1.3 l<(il...ir)> I = r.

The inverse of this cycle is 1.4 (il...ir)-I = (irir_1...i I)

Disjoint cycles (i.e. the sets of really moved symbols are disjoint) describe c o ~ u t i n g written in cycle-notation, permutations. Each permutation can be

i.e. as a product of pairwise disjoint (cf. 1.2)

cycles, which are uniquely determinea - as permutations

-

up to their erder of occurence.

Because of 1.5 (il...i r) = (ilir)(ilir_1)...(ili2) too, can be written

each cycle, and therefore every permutation, as a product of transpositions. transpositions. Since

Hence Sn is generated by the

1.6

(lj,ik+1) = (ik,ik+1)(ijik)(i~,i~+~)
of successive symbols. for

Sn is generated even by the transpositions

Another system of generators of Sn is ~(12),(12...n)S, 1.7 Hence !~ Sn = ((12),(23),...,(n-l,n)) = ((12),(I...n)~ (1...n)r(s2)(1...n)-r = (r+1,r+2),

0_(r(n-2 .

(For further results on systems of generators compare Ooxeter/ Moser KIS, Piccard CIS-C3S.)

We would like now to describe the conjugacy classes of S n. The ordered lengths al, ..., ah ( ~ j ~ j + 1 ) of the cyclic factors of

E Sn (with respect to the cycle-notation of ~ and including the lengths of l-cycles) form a partition

1.9

P~ := ( a l , . . . , e

h) =

of n, i.e. they satisfy 1.10 ~ ~i = n, ~i E ~ I , ajAaj+ I (1~j<n-S) .

P~ is called the partition of ~. If ~ is the partition of ~, then we obtain from 1.3 that the order of the generated subgroup <~> ~ Sn is equal to the least common multiple of the elements a i of ~: 1.!1 P~ = (~l,...,ah) ~ I<~>I = Icm ~i "

Thus we have a criterion whether ~ is p-regular, prsingular or a p-element with respect to a prime number p (i.e. whether p t I<~>I, P I I<~>l or I<~>I is a power of p): If P~ = (~1,...,~n): (i) ~ is p-regular o Vi: p t ~i ' ~i: p I ai , o Vi: ~ is a power of p .

(ii) ~ is p-singular (iii)~ is a p-element

Now we wish to show that each subset of Sn consisting of the permutations with a certain partition ~ forms a conjugacy class of S n. To prove this we notice first that

i
1.13 ~,~-1 = (~(~)) (~'(i)) ( This means, that we obtain ~ , ~ - 1

)
~ )=

(
'~'(i) )

from ~' by an application of

to the symbols in the cycle-notation of ~', e.g. (123)(35)(123) -1 = (15) This application of ~ obviously doesn't disturb the lengths of the cyclic factors of ~' so that we have p~,~-1 = p~, .

Hence a conjugacy class consists of permutations of equal partitions. On the other hand, if we are given ~" E Sn with P~" = P~', there obviously exist permutations ~ which fulfil ~ , - I = ~., namely

all the permutations ~ which map the symbols of ~' onto the symbols of ~" as described above. Therefore denoting by ,N,' that the two permutations are conjugates we have ~i N ~" Because of 1.4 we have P~ = p~-1 o P~' = P~" . so that every permutation in

Sn is a conjugate of its inverse. Groups in which every element is a conjugate of its inverse are called ambivalent. Hence: I_~ Sn is ambivalent.

The use of a second notation facilitates the calculation of the order of a conjugacy class. ~or i = 1,...,n let a i be the number of elements ~j of ~ = P~ with ~j = i, i.e. the number of i-cycles

10

among the cyclic factors of ~. This establishes a one-to-one respondence 1.16

cor-

between the partitions P~ = and the (1n)-matrices = a , O<a i E Z , E ia i = n .

-i

T~ := (al,...,an)

T~ is called the type o f ~. Hence we may describe the conJugacy classes of Sn with the aid of such n-tupels,too: Let C a = C a denote the class of the permu"''' mh

tations of Sn consisting of h cycles of the lengths ~1' respectively Because 1.17 permutations consisting of a i i-cycles for i = 1,...,n.

of 1.13 and 1.2, for ~' and ~" cut of C a we have exactly ~ i ai ai! = ~,,. Hence 1.17 is the order

~ which satisfy ~ , ~ - I

of the centralizer CSn(~' ) of ~' in Sn and we have: ai ai

Ic:=oal :n,l(~i
As

al,) , l%n(:'~Ca)l : ~i

ai, .

an example we consider S 3 (for n~3, S n is not abelian): S 3 = {I,(12),(13),(23),(123),(132) 3 .

This group consists of 3 conjugacy classes corresponding partitions (3), (2,1) and (1,1,1) =: (I 3 ) respectively (1,1,0) and (3,0,0). These classes are c (13) = c (3,o,o) = {i] ,

to the

to the

types (O,O,1),

0(2'1)= 0(1'1'0)
c (3) = o (,,1)

= {(12),(13),(23)3
= {(123),(132)] .

We shall consider the group-theoretical

structure

of the centra-

11

lizers later on. Other subgroups of importance are the alternating groups A nA n ~ Sn consists of th8 permutations the sign of the difference-product ~ E Sn, which don't change

~n '=
i.e.

17
l<i<j <_n

(J-i) ,

1.19

An '= [ = ~

Sn I ~ n

:=

l~i<j<..n

1"I

(~(J)-~(i))

= a n]

(It is easy to see that ~A n = ~ n ) . rence-product A I is I by convention

Recall that the empty diffeso that . satisfying ~A n = -A n,

A I = A 2 = [1] Permutations ~ E Sn\A n, i.e. permutations

are called odd permutations , the elements of A n are called even permutations. It is obvious, that the product of two even permutations is even,

hence A n is a subgroup.

Thus every permutation group P ~ Sn con-

tains a subgroup P+ consisting of its even elements:

1.20

P+ == P Q A n

Depending on ~, left coasts ~P+ consist either of even or of odd permutations. Since the left coset P+ includes all the even per-

mutations we obtain for the index of P+ in P:

1~

IP = P*I ~ 2

Hence P+ is a normal divisor of P in any case: P+ d P.

12

Por m~2 Sn contains the transposition (12), which is odd, thus we obtaim as a special case of 1.21:

122

IAnl = n , / 2 ,

Since 1.5 is valid, a cycle is even if and only if its length r is odd. E. g.
A3 = [ 1 , ( 1 2 3 ) , ( 1 3 2 ) 3 4 S3 .

Hence an investigation of the partition or of the type of ~ allows one to decide whether ~ belongs to A n or to Sn\A n. Hence the conjugaoy classes of Sn - let's call t h e m ~ n - c l a s s e s - belong to

A n or to Sn\A n in full. The question arises, which of the Sn-classes split into ~n-classes. The centralizer of an even permutation ~ in A n is CAn (~)
=

CSn(~)+

OS n ( ~ )

An

Hence we obtain from 1.21: V ~ E An: either CAn(~) = OSn(~) or ICSn(~):CAn(~)I = 2.

Thus we have for the orders of the conjugacy classes (which are equal to the index of the centralizer of each of their elements): ~ E An: either IOS(~)I = 21cA(~)I or CS(~) = cA(~) .

Hence the Sn-class of ~ E A n splits - and then into two An-classes of the same order - if and only if

CA( ) = Can
We prove a little bit more:

13

For n~l exactly those Sn-classes split (and then into two classes of equal order), the centralizers of whose elements fulfil

CSn( ) = cAn( )
These are exactly the classes the partitions of whose elemerits are of the form m = (~1,...,~h) with pairwise different and odd elements mi"

Proof: It remains to prove the second part of the statement. a) A permutation ~ commutes with each of its cyclic factors. Hence if the class of ~ splits, ~ cannot have an odd cyclic factor (of even length). Analogously we see that ~ cannot have two cyclic factors (il...ir) and (i~...i~) of the same odd length r, for this would imply the existence of an odd permutation in the centralizer:

Sn\A n 9 ( i l l ~ ) . . . ( i r i ~) E CSn(~)
It follows that at most those classes split whose permutation has pairwise different and odd elements a i. b) That these classes really split we derive from the odd order

i of the centralizers of their elements which implies

+

Cgn

= CSn q.e.d.

14

If the Sn-class C a = 0 a splits we shall d e n o t e two An-classes and fix C a+ = C a+ by

by C ~

= Ga~ the

1.24
Though

(1...al)(al+l...al+a2)...(...n)

E C a+ = C a+ .

0 (3)+

= [(123)},

0 (3)-

= {(132)],

it is not true in general that O a- consists of the inverses of the elements of C a+. E.g.
(25)(34)(12345)(25)(34) = (15432) = ( 1 2 3 4 5 ) -1 ,

so that (12345) as well as (12345) -1 belong to C (5)+ c A 5. We prove that (Berggren [ 1 ] ) : I~ A 1 =_A 2 = [1], AT, A6, AIO and A14 are the only ambivalent alternating groups.

Proof: Because of 1.15 and 1.23 we need only consider An-Classes C a of partitions ~ = (al,...,Ch) with pairwise different and odd S1~mmELnds a i. Each of the other An-classes contains with an element its inverse. Let = (il...ir)...(jl...J s) be a product of disjoint cycles of odd lengths. We can form P := (i2ir)(i3ir_S)...(J2Js)(J3Js_1)... the standard-~onju~ator of ~ which satisfies
p~p -1 = ~ -1

(1 )

We notice, that p is an odd permutation if and only if the hum-

15

her of cyclic factors of ~ whose length is congruent 3 modulo @ is odd. a) If the standard-conjugator p for such an element ~ out of a splitting class is odd, then the considered alternating group A n cannot be ambivalent. Por if a E An and a ~ -I = - I equation

(I) would imply (a-lp)~(a-lp) -1 = ~, which is in contradiction to CSn(~) = CAn(~) since a-lp E Sn\A n. b) It remains to show that exactly for the natural numbers n ~ [1,2,5,6,10,14] there are partitions with pairwise different and odd summands ui and so that the number of ui satisfying ui ~ 3 (4) is odd. (i)The only partitions u of the n E [1,2,5,6,10,14] with pairwise different and odd elements =i are as follows:
n = 1: (1); n = 2: ~ ; n = ~,: ( 5 ) ; n~,6,: (5,1); n = 10: (9,1),

(7,3); n = 14: (13,1), (11,3), (9,5), in each of which the number of elements congruent 3 modulo 4 is 0 or 2 and therfore even. Hence the standard-conjugator is even in every case such that these alternating groups are ambivalent. (ii)Let us now look at the n ~ {1,2,5,6,10,14]. We distinguish the natural numbers n with respect to their residue classes modulo 4. I. n = 4k, k E ~ : For n = 4 we have the partition (3,1) and

16

for k>l the partition

(4k - 3,3) with an odd number of ele-

ments congruent 3 modulo 4. 2. n = 4k + I: Because of the ambivalency k~2. of A I and A 5 we assume

(4k - 3,3,1) fulfils the condition. (4(k-I) - 3,

3. n = 4k + 2, kA4 (A2,A6, ASO , A14 are ambivalent): 5,3,1). 4. n = 4k + 3: (n) = (4k + 3) has one element, congruent 3 modulo 4.

and this one is

Thus each of the alternating groups A n with n ~ (1,2,5,6,10,14S possesses a conjugacy class which does not contain the inverse of groups are not

each of its elements and hence these alternating ambivalent.

q.e.d. ~rom the proof of 1.25 we get: 1.26 The two An-classes C~ into which the Sn-class to the partiwith pairwise different and odd elements n>1) are ambivalent if and only if the

tion a = (~l,...,~h) ~i splits (suppose

number of elements

~i of ~ with ~i ~ 3 (4) is even.

We shall come back to the centralizers said something about wreath products, ducts of certain wreath products.

of elements when we have since they are direct pro-

To conclude this section let us look at the double cosets of der certain subgroups which will be of use later on.

17

A. J. Coleman has pointed to these double cosets (Coleman [IS)as elucidating the group-theoretical background of the apparently purely combinatorial introduction to the representation theory of the symmetric group (cf. section 4). For a partition ~=(~1,...,~h) of n, let gV, ..., ~ be pairwise

disjoint subsets of orders l~I = ~i of the set ~ = [1,...,n} of symbols on which Sn acts. Subgroups of the form 1.27 S~ := S~I ... S~h = i S~i ~ Sn

(Sai the subgroup of the ai! permutations fixing the symbols out of ~\G~, 1~i~h) are called Young-subgroups in honour of

A. Young (1873-1940) to whom we are indebted for the theory of representations of the symmetric group in which such subgroups play an important role (cf. section 4). If we are given two partitions, say ~ = (~l,...,~h) and ~ = (~1,...,~k),of n and two Young-subgroups S~ = Sai and S~ = S~j, ~ and ~* out of Sn, we want to show the following (Coleman [1 ]) :

Proof: (i) If ~ = ~'~*~'~, ~' E Sa, ~" E S~, we have ~ ~i N ~Q = (Qi N ~*a ), V i,j. = ~'~*~, Vj.

18

(ii) I ~

n ~1

= I~i n ~*~ I, v i , j ,
as well as the subsets ~

implies that the subsets

N ~*~ form for fixed j

~i N ~Q

two complete dissections of g~ into pairwise disjoint subsets which can be collected into pairs ( ~ subsets of equal order. Hence for each i there is a ~ E Sai satisfying
, J. V

N ~'~i

N ~*g~) of

, ~i(Dia G ~g~) = g~ N ~*g~

Multiplying these ~ ~' ~= ~ . . . ~

together we have for the resulting

~ s~

Thus there is a ~" S~ such that ~ = ~,-I~.~,, as stated. q.e.d. Hence we have a l-I-correspondence between the set of double cosets Su~S~ of S~ and S~ and the system

of rational integers satisfying h k 1.30 ~ zij = -JS~' j=1 zij = ~i' O<zi~j E Z . ~ i=I -We notice that 1.31 S~ N ~ S ~ -I = x i,j Sz < Sn , ij -

if Szi j is the subgroup of the zij! permutations of Sn fixing the elements of G\(g~ G ~Q~), SO := {I] ~ Sn. The number of such systems [zij } which fulfil 1.30 is obviously

19

equal to the coefficient of

:= Xl "'"~t ~1

in h,k i,~=I (1-xiYj)-1 oo

(xi,Y j independent indeterminates); from which it follows that: The number of distinct double cosets..S~S~ of Sa and S~ in Sn is equal to the coefficient of x~y ~ in ~ (1-xiY~) -1.
L

We notice that during these group-theoretical considerations the polynomials ~ ( 1 - x i Y j ) -I make their appearance, which play an important role in the character theory of the symmetric group and in the theory of the so-called S-functions (cf. Littlewood [2], 5.2, 6.4). Special double cosets S ~ S ~ are those corresponding to solutions [zij ] of 1.30 which satisfy O~zij~1, i.e. (cf. 1.31) Sa fl ~ S ~ -1 Por their number we get from 1.32: The number of double cosets Sa~S ~ with the property Sa N ~ S ~ -1 = [1} is equal to the coefficient of xay ~ in
=

[1].

7T(1+xiy )Ooncluding this section we consider certain pairs (u,~) of partitions with respect to this number of certain double cosets.

20

A partition 5 = (51,...,~h)

of n can be illustrated by a Young-

diagram [5] consisting of n nodes in h rows, with 5 i nodes in the i-th row and all the h rows starting in the same column. E.g. the partition (3,2,1,1) =: (3,2,12 ) can be illustrated by

[3,2,12]:

On account of the condition ~j~aj+ I (l~j<_n-1) it makes sense to speak of columns of the diagram [a],and of the length ~ of the

i-th column. Hence to the partition ~ there corresponds the partition

1.34

'

:=

a' := ( ~ , . . . , ~ , = a l )

, ~i

~) j:~>i

~' is called the associated partition of ~, [~'] the Youngdiagram associated with [~] which arises by interchanging the rows and columns, e.g.

[3,2,12']

= [4,2,1]:
@

Partitions ~ and Young-diagrams resp.

[5] with the property ~ = ~' and it will appear

[5] = [5'] are called selfassociated,

in section 4, that the following lemma is crucial: 1.35 Young-subgroups S~ and Sa, possess exactly one double coset

$5~S~, with the property that S~ 0 ~S~,~ -1 = {1].

21

Proof: We want to use 1.33 and hence we prove by induction with respect to h, that coeff, of x~y ~' in i,~=I (i) h=1: Obviously [coeff. of x I ~ Yi in j~l= (1+xlYj)~ = I . (l+xiY j)] = I .

(ii)If we assume that ~h = r, the induction hypothesis is h-1 ~i ~ - I ~'-I ~' ~, h-~h' r+l .TT. [coeff. of (i~_ xi )Yl =1 "''Yrr Yr+l "''Yh' in i,j (1 +xiy j) ]=1 r Multiplying both sides with J~1= xhyj gives ~coeff. of x~y ~' in h ~ h ' (1+xiYj ) j~ixhY J) = I . i,j-1 = ~w Since ~h=r the coefficient of x~y is the same in h~h' r i,~=I (l+xiYj) j~=lXhYj
as i~

(1 +xiY j) i,J-1 = as well as (since a~<_h-1 for j>r) in

(I+xhY j)

h~'
i,'J~l

h'=g I
~ ~ j=l ~ ~ i~J~l

This together with 1.33 yields the statement. q.e.d. The double coset 8~8, with S n ~S,~ -I = [1] may be illustra-

ted by a so-called Young-tableauwhich arises from the Young-diagram [a] by replacing the nodes by the symbols 1,...,n of G. E.g.

22

3 4 2 7

5 6

is a tableau with diagram [3,2,12].

a a t

If gi' gJ

are the sets of symbols in the i-th row, j-th col,lmn

of the tableau T a, the elements of the Young subgroups S a = x Sai resp. S, : X S ~ resp. ~\Q (S~i resp. Sa~ fixing the elements of . \ ~

) are called horizontal permuta}ions resp. vertical

permut@tions of T a. The groups o_~fall the horizontal resp. all the vertical permutatlions of T a are indicated by 1.36 They satisfy 1.37 ~ n v ~ = {I} . H~ resp. Va .

Therefore the tableau T ~ illustrates the double coset SalSa, with S a O ~Sa,~-1 = [I] in the following sense: If S a := H u and if we are given Sa, ~ Sn, then T m illustrates permutations p (namely the p with pSa,p -I = V a) which satisfy S~ N pS~,p-1=[S}.

These are the results on symmetric and alternating groups we wished to summarize first. Prom now on we shall use the cited results on the symmetric group and its conjugacy classes to define and examine wreath products. We shall return in section 4 to double cosets to construct idempotent elements of the group algebra of Sn which generate minimal

23

left ideals affording

irreducible representations

of Sn over the

field of complex numbers.

24

2. Wreath products

We define the wreath product of two groups as follows (see e.g. Huppert [I], I 15):

If G is a group, H a permutation group on the set of symbols ~ = {S,...,n], the set {(f;~) I f mapping ~ into G, ~ 6 H}

together with the composition law

(f;~)(f,;~,) := c~f,,~,)
is called the wreath product G~H of G with H (sometimes G-wreath with H). Included that to f:G ~ G and ~ E H the mapping f~:g - G is defined by

f (~(i))

:= f C i ) ,

V i ~ ~,

and for two mappings f,f':G ~ G their product ff':~ - G by ff'(i) := f(i)f'(i), V i ~ ~ ,

G~H is a group as can be seen easily. While checking this one notices, that 2.2 (f)~, = ~,~, V f,~,~', right

since products of permutations have to be read here from to left.

25

If we denote by e:~ ~ G the mapping with the values e(i) = 1G, v i E ~, and if we define to f:Q ~ G the mapping f-l:~ ~ G by f-1(i) := f(i) -I, V i E Q, and for the inverse of

we have for the identity element of G ~

(f;~) : 2.3 1G~ = (e;1H) , (f;~)-I = (f-11;~-1)_

((f~_1)-1=(f-1)~_1=:

f-1_1)"

The order of this group is 2.4 Eor the derivation

IQ~I

= lainl~l

.
theory of these groups normal divisor is very

of the representation in, the following

w h i c h we a r e i n t e r e s t e d important= 2.5

G* := [ ( f ; 1 H)

] f : Q ~ G] = G1 X . . . x

Gn ~ G~H .

G* is called the basis group of G ~ . n copies G i of G: 2.6 The subgroup 2.7 H' := [ ( e ; ~ ) G i := ((f;1 H)

It is the direct product of

f(J) = 1G, V j + i] ~ G .

I ~ E HI ~H

is the complement of G* and isomorphic to H, i.e. 2.8 G~ = G*H', G* ~ G ~ , G* N H' = I G ~ = (e;1 H)

Let us now look for examples. We shall recognize some subgroups of the symmetric group as permutation representations of cer-

26

tain wreath products. Looking for these examples we may follow the historical development of these ideas. Such permutation groups arose in the process of constructing a p-Sylow-subgroup of the symmetric group, even before Sylow

proved his famous theorems in 1872. The first description of this construction was probably given by A. 0auchy in the third volume of his "Exercises d'analyse et de physique math~matique" which appeared in 1844 (cf. also the sections 39 and 40 of E. Netto's "Substitutionentheorie Anwendungen auf die Algebra" umd ihre

(1882) as well as A. Radzig's dis-

sertation "Die Anwendung des Sylow'schen Satzes auf die symmetrische und die alternirende Gruppe" (1895)). Let us denote by ep(m) the exponent of the maximal power of a prime number p dividing m. If we assume that pS~n, pS+S)n and

aspS<n (O<as<P) , but (as+1)pS>n , then since n! = 1.2'....pS(pS+1)...2pS...aspS(aspS+1)'''(aspS+(n-as we can conclude that ep(n!) = asep(pS!) + ep((n-aspS)!) ps,_ If now a s_Ips-1<n-as . ps))

(as_1+1)pS-1>n-as ps, then in the same way

ep(n~ ) = asep(p s,) + as_sep (ps-1 !) and so on. Hence the problem of constructing a p-Sylow-subgroup be reduced to the construction of a p-Sylow-subgroup of Sn can of sym-

27

metric groups Spr of p-power-degrees pr. The maximal power of p in pr, is ep(pr :)
p ----

p1+p+p2+...+pr-1
_--

p(1+p+...+pr-2)p P P,

2.9 =p

ep(p r-1 !)P

hence every p-Sylow-subgroup pr of S r has the order P 2.10 I rl = Ipr-ll p p ,

if pr-1 is a p-Sylow-subgroup of 8 pr-1" Therefore it would be desirable to construct with the aid of pr-1 _ which is a permutation group of degree pr-1 _ a permutation group of degree pr and of order Ipr-11 p P. On account of 2.10 the result would be a p-Sylow-subgroup of Spr. That this might be done by constructing a faithful permutation representation of pr-1~Cp (Cp := <(1...p)> _(Sp) is suggested by a comparison of 2.4 and 2.10. More generally: from two permutation groups G and H of the degrees m and n construct a permutation group of degree mn and of the order IGInlHI. We prove first:
2.11 If
G

is a permutation representation of G on the set of symbols

r = {1,...,m}

then

(f;~)(i,j) := (f(~(j))(i),~(j)), V (i,j) E 29,

28

yields a permutation representation This permutation representation permutation representation

of G~H on

Fx~.

of G~H is faithful

if the

of G is faithful,

and it is tran-

sitive if both G (on F) and H (on G) are transitive.

Proof:

a) It is easy to verify,

that

Cf;~)CCf';~')Ci,j))
And from

= Cff~;~')Ci,j),

v f,f',~,~',i,j

Cf;~)Ci,j) we get
(fC~Cj))Ci),~CJ))

= Cf;~)Ci',j')

= CfC~Cj'))Ci'),~Cj'))

is a permutation,

hence this implies fC~(j))(i)

j = j',

so that we have

= fC~Cj))(i')

fC~(j))

is a permutation,

too, hence i = i'. of G~H as

Thus we have obtained a permutation representation claimed. b) If (f;~) is in the kernel we have Ci,j) = (f;~)Ci,J) = (fC~Cj))Ci),~(j)),

v i,j .

If the given permutation representation

of G is faithful this

implies f(j) = IG, V j, and hence in this case f = e, so that the representation of G~H is faithful, too. on ~, and (i,j)

c) Finally if G is transitive

on F, H transitive

and (i',j') are two symbols out of FxG, there is a ~ E H and a g E G such that ~(j) = j' and g(i) = i'.

29

If we choose an f:g - G so that f(j') (f;~)(i,j) = (f(~(j))(i),~(j)) representation

= g, then = (i',j'), as well.

= (f(j')(i),j')

and the permutation This completes

of G%H is transitive

the proof. q.e.d.

The permutation

representation

of G~H given in

2.11

is of degree

Irx~J = mn. Two permutation called

groups G I and G 2 on G 1 and ~2 are mapping ~ of G I onto ~2

similar , if there is a biJective ~ of G I onto G 2

and an isomorphism 2.12

so that V i E gf, g E G 1

E(g(i))

= ~(g)(e(i)), a subgroup

We would like to describe tation representation

of Smn similar to the permu-

of G ~

given by 2.11. we choose the bijection de-

For such an E:Fx~ ~ A = {1,...,mn] fined by 2.13 To describe E(i,j)

:= (j-1)m + i, 1~i~m, of the p e r m u t a t i o n the set A

l~j~n

. of

an isomorphism

representation

G~H into Smn we first divide Smn acts into n pairwise 2.14

of the symbols

on which

disjoint

subsets

A i of order m, say

A = {1,...,m,m+1,...,2m,...,~n-1)m+l,...,n~} A1 A2 ... An

Now let e be a permutation 2.15

which permutes

the A i cyclically,

say

a := (1,m+1,...,(n-1)m+1)(2,m+2,...)...(m,2m,...,nm). ~ of the permu-

With the aid of a we now define an isomorphism

3O

ration representation

of G~H into Smn as follows. We define first

the images of the values f(i) of f by 2.16 ~(f(i)) := ai-lf(i)a I-i =: ~i '

so that ~i acts on Ai, and 2.17 Now let 2.18 be defined as follows: 2.19 so that 2.20 ~*Ai = A~(i) " ~*((j-1)m+i) := (~(j)-1)m + i, 1~i<_m, 1~j~n , ~(e;~) =: ~* ~(f;1 H) :=~i ~i

This means, that the image of (e;~) under ~ is the permutation of the A i corresponding to the permutation ~ of the symbols i of ~ = [1,...,n). As can easily be seen, this mapping @ defined by 2.21 ~(f;~) := ~l...~n~* of G~H

is an isomorphism of the permutation representation

described in 2.11 into Smn. The image of the permutation representation is obviously 2.22 (G I x...x Gn)H' ~ Smn , of G and

with G I equal to the permutation representation 2.23 G i = ~i-lGlal-i

H' is the permutation group acting on the subsets A i of A as H acts on the symbols i of ~; H' is a subgroup of the subgroup

31

S'n - Smn which consists of the n'. permutations of the A i. < To prove the similarity we have to verify, that 2.12 is valid. Starting with the left hand side of 2.12 we obtain ((f|~)(i,j)) = (f(~(j))(i),~(j)) =(~(j)-1)m + f(~(j))(i) And on the right hand side we have @(f|~)(~(i,j)) -- ~l...~n~*((J-1)m + i) = ~l...~nC(~(j)-l)m + i) = (~(J)-l)m+f(~(j))(i) Thus we have proved the following: . .

2.25 The mappings E and ~ defined by 2.13-2.21 map the permutation representation of G~H described by 2.11 onto a similar subgroup of Smn.

It is interesting to see, that these groups are the permutation groups used by Cauchy and Netto to construct Sylow-subgroups of the symmetric group and which arose at the beginning of the development of the concept of the wreath product of groups. A sketch of this construction reads as follows:

2~A

faithful permutation representation of degree mu of the wreath product of two permutation groups G (on F=~1,...,m~) and H (on G = (1,...,n)) can be obtained as follows: Divide the set A = ~1,...,mn~ on which Stun acts into dis_.. joint subsets AI=F , A 2, ..., An of order m. ~orm the direct product of G = G I (on At=F) with the subgroups G i (on the

32

subsets Ai) which correspond to G 1 (2_(i_(n) and multiply GI x . . . x Gn with the subgroup H' corresponding to H and

acting on the subsets Ai as H acts on the symbols i of 9. The result (G I x...XGn)g' is the desired faithful permutation representation of G~H

and it is transitive if both G (on F) and H (on 9) are transitive.

A simple example is the representation

of S2%S 2 = C2%C 2 (if we

denote by Cn the cyclic group of order n):

C2%02 = { ( 1 , 1 ; 1 ) , ( 1 , ( 1 2 ) ; 1 ) , ( ( 1 2 ) , 1 ; 1 ) , ( ( 1 2 ) , ( 1 2 ) ; 1 ) , ( 1 , 1 ; ( 1 2 ) ) , (1,(12);(12)),((12),1;(12)),((12),(12);(12))3
((f;~) written in the explicit form (f(1),f(2);~)) is similar to

([1,(12)] X {1,(34)])[1,(13)(24)]
= {1,(12),(3#),(12)(3),(13)(24),(1)(23),(1324),(1423)3. This image of C2%02 is a 2-Sylow-subgroup Coming back to the p-Sylow-subgroups tions: of S pr _I on F={1 , .. ., pr-1 3, of S 4.

we get from these considera-

2.26 If pr-1 is a p-Sylow-subgroup

and if Cp = <(1...p)> < Sp, then the subgroup of S r similar to pr-1%Cp as described in 2.25 is a p-Sylow-subgroup.

33

Using this result we can construct a p-Sylow-subgroup If

of Sn.

t ~ aipi , O_(ai<p, i=0 then as can be seen from considerations above the exponent of n = the maximal power of p dividing n! is ep(n!) = a I + a2(1+p) +...+ at(1+p+...+p t-l) 2.27 = alep(p!) + a2ep(p2!) +...+ atep(pt!)

Let us now divide the set G = [1,...,n] on which Sn acts into pairwise disjoint subsets as follows. We divide ~ into the subset Go := [1'''''ao] of order ao, into the a I subsets ~11 := := [ao+(a1-1)p+1,...,ao+alP3 of order

[ao+1,...,ao+P ), ..., ~1aI

p, ..., a t subsets ~t1''''' Qta t of order pt. On each of these ". subsets ~iJ form a p-Sylow-subgroup pl I~C p of Sp I (regarded as subgroup of Sn). And now take their direct product t ai x ( x (pi-l~Op)) ~ Sn i=I := pi-1~Op ...x pi-1~Op (a i factors)). e~(n!) The resulting subgroup is of order p ~ and hence a p,Sylowsubgroup of Sn. Thus we have derived the well-known construction of p-Sylow-subgroups first given by Cauchy (for this and other of symmetric groups cf. ( a i (pi-1~Op) x

results concerning p-Sylow-subgroups

also the papers of L. Kaloujnine and the paper of Weir (see the references)):

34

2,28 If n = Z aipi (O~ai<P) , then every p-Sylow-subgroup ai (pi-1%Cp)) is of the form ( x i

of Sn

This as well as 2.26 gives a recursion formula for the construction of a p-Sylow-subgroup. Because of the associativity of the

wreath product multiplication

this can be written explicitly:

2.~

Let G, H and I be permutation groups on F, G and A. Then (G%H)%I a n d G%(H%I) are corresponding permutation groups,

if we identify (FG)xA and F(~A) according to

((i,j),~)

= Ci,CJ,~))

Proof: Let (f;=) be an element of (GxH)%I so that f(i) = (fi;~i) E G%H, i E A. And let (f*;~*) be an element of G%(H%I) with ~* = (f';~) E H~I so that f'(i) = ~i and f*(i,j) = fj(i). Then on the one hand

(f~)(Ci,J),k)

= ((fck);~C~))(i,j),~(~)) = (Cf~c~)C~c~)CJ))Ci),~(~)(j)),~(k))

And on the other hand

Cf*;~*)Ci,Cj,~))

= Cf*CCf';~)CJ,~))Ci),Cf';~)Cj,k))

= (f*Cf'C~Ck))CJ),~C~))Ci),Cf'(~(k))(J),~(k)))

= Cf~c~)C~c~)C~))Ci),(~C~)Cj),~(~))

Thus (fI~) acts as (f*;~*) if we identify the symbols as described. q.e.d.

35

A corollary is

2.30 If Op := <(1...p)> ~ Sp, then every p-Sylow-subgroup of S r is similar to the wreath product
r

0p := Op~...~0p (r factors)

Every p-Sylow-subgroup of Sn (n = Z alP i, O~ai<P) is similar

to a x(xi( r

Op))

As we have seen, Sylow-subgrcups of symmetric groups are direct products of wreath products. In the introduction we claimed,that this is also valid for the centralizers of elements in Sn. To show this, we start with a special case, namely the centralizer of the permutation = ~1...~n := (1...m)(m+1,...,2m)...(...nm) The centralizer E Stun

CSIIm (~)

of this permutation is

2.51
That means

Smn(~) = (<~I >'"<~n>)S~ "

CSIIm (x)

is the product of the subgroup generated by

the cyclic factors ~i of ~ with the subgroup S~ of the n! permutations of S m n W h i c h permute the sets of symbols in the cyclic factors of ~ as they stand. Following the considerations preceeding 2.24 and 2.25 we see, that this subgroup is a permutation representation of Cm~S nAnd applying this to the subsets of cyclic factors of the same

36

length in a general permutation we have:

2.~

If ~ E Sn is of type T~ = (al,...,an), then the centralizer ...... of ~ in Sn is a faithful permu%ation representation of

x (ci~Sai) i
(ci~s o := [I], ci:= <(i...i)> _( si) .

(Remark: Prom this we can derive at once the result 1.23 about the splitting of Sn-classes into An-Classes.) Analogously, we have the following result which will be of use later on:

If Smn is the symmetric group on G = {1,...,mn}, then the n normalizer NSm n( Sm) of the subgroup n X Sm := Sm X... Sm (n factors) (i-th factor Sm on gi := {(i-1)m+1,...,im]) is the faith........ ........ ful permutation representation of Sm~S n described in 2.24 and 2.25.

These three examples, the p-Sylow-subgroups, the centralizers n of elements and the normalizers of subgroups of the form x Sm in symmetric groups show, how useful this concept of the wreath product is. Moreover they give a hint as to how the representation theory of wreath products may be applied to the representation theory of the symmetric group. The centralizers of p-elements as well as the p-Sylow-subgroups

37

of the centralizers of p-regular elements play an important role in the p-modular representation theory of finite groups. On the other hand it is known, that the so-called symmetrized outer products of two irreducible representationsof Sm and Sn are representations of Smn induced by certain irreducible representan tions of NSm n( x Sm). The other way round we may ask how the representation theory of the symmetric group can be applied to derive the representation theory of certain wreath products. About 1930 A. Young applied his methods to the so-called hyperoctahedral groups. In our notation these are groups of the form S2~Sn, and they arise by representing the elements of Sn by permutation matrices and allowing not only +I as nonvanishing entries but also -I (cf. Young [I]). Pollowing a suggestion of I. Schur, W. Specht then considered in his dissertation such groups where not only ~1 are allowed as nonvaniehing entries but even the elements of a group G if the matrix multiplication is defined appropriately (Specht [11). These groups are obviously of the form G~S n. In a following paper (Specht [2~) he considered the general case G~H, H a finite permutation group. He showed that especially the ordinary representation theory of wreath products of the form G~S n can be largely derived with the aid of the representation theory of Sn.

38

Nevertheless the application of the theory of the representations of wreath products to the representation theory of the symmetric group is not a vicious circle. For those wreath products G~S m whose representations we shall apply to Sn satisfy m<n , so that on the contrary this application provides an interesting recursion process. In order to derive the representation theory of wreath products of the form G ~ n it is necessary to examine such groups more closely.

This we shall do in the following section.

39

3. Wreaths with symmetric groups

We shall now consider wreath products of the form G~Sn, - wreaths with s~,-etric ~ . group. Since @~S n may arise as described at the end of the last section by inserting elements of G for the nonvanishing entries of permutation matrices representing the elements of Sn, the group G~S n is sometimes called the complete monomial group (of degree n) of G. O. Ore called it symmetry (of degree n) of G (Ore [I]). Wreaths Om~S n of cyclic groups with symmetric groups have been called ~eneralized s~,~,-etric groups by M. Osima (Osima [I]), and analogously the groups Cm~A n were called generalized alternating groups (Puttaswamaiah [I]). As has been mentioned above, the special case 02~S n is called a hyperoctahedral ~ . Let us first consider the conJugacy classes of G~Sn, which have been characterized by W. Specht (Specht [1]). Since [ I ~ S n ~ Sn this characterization provides a generalization of the characterization of the conJugacy classes of symmetric groups by their cycle decomposition. We choose the notation so that we shall abtain a generalization of the characterization by the type of the permutation. For the time being let G denote a finite

4O

Let (f;~) be an element of G~S n. The cycle-notation of ~ is uniquely determined if we agree to begin each cyclic factor with the least symbol included in it. With this convention we can associate with each cyclic factor (j ~(j)... r(j)) of ~ with respect to f:g - G the uniquely determined element 3.1 ff ...f r(j) = f(j)f( -1(j))...f(-r(j)) E G

We call this element the cycleproduct associated with (j... r(j)) with respect t_~of. This done we can give the following definition which generalizes the concept of the type of a permutation to the concept of the type of (f;~). As we shall see soon afterwards, this type characterizes a conjugacy class of G~S n.

~.2 Def.: Let the permutation ~ E Sn be of type T~=(al,...,an), and let f be a mapping of ~ into G such that (f;~) E G~S n There are ak cycleproducts associated with the ak cycles of length k of ~ with respect to f as described in 3.1. Let CI,...,C s be an ordering of the conjugacy classes of the finite group G. If now exactly aik of these ak cycleproducts belong to C i we call the (sxn)-matrix i row index, 1~i~s, T(f;~) "= (aik)

k colnmn index, I <_k<_n,

the type o f (f;~).

41

In the special case s=1, i.e. in case ~I]~S n = Sn, we have the row (al,...an) and hence 3.2 is a generalization of the definition 1.16 of the type of a permutation. We would like to show that as in 1.14/1.16, two elements of

G~S n belong to the same conjugacy class if and only if they are of the same type. Before we can prove this we need some preliminary considerations. At first we notice, that the entries aik of the type (aik) of an element (f;~) E G~S n, with ~ of type (al,...,an), satisfy the following equations: 3.3 0(aik E ~,
--

E aik = ak,
i

Z kaik = n . i,k

To each (sxn)-matrix (aik) whose entries satisfy 3.3, there are elements in G~Sn, which are of this type, since f ranges over all the mappings of g into G. Since we are merely interested in the type, we need only determine a cycleproduct up to conjugation in G. Therefore we show first, that the convention that the symbol j in 3.1 is the least symbol of the considered cycle is umnecessary, i.e. that

f...frCJ) ~
Proof: We can assume O~s(_r. Then

f...f

rC

SCj)),

f...f r(~S(j)) = f(s(j))...f(~(j))f(j)...f(~s+1(j)). If a := f ( s ( j ) ) . . . f ( ~ ( j ) ) , b := f(j)f( - 1 ( j ) ) . . . f ( s + 1 ( j ) ) ,

(I)

42

then ab is the right hand side of the equation (1), and ba is the cycleproduct to (j... r(~)) with respect to f (cf. 3.1). But ab - ha: a-laba = ha. q.e.d. Another useful remark is

3.5

T(f~) = T(e;~')(f;~)(e;~') -I
= ~(f,;1)Cf~)Cf,~l) -1 ' Y f,f',~,~'.

Proof: We have

(e;~,)(f;~)(e;~,)-I

= (f

,;~,~,-1),

(fVll)(f;~)(f,;1)-1 = (fvff~-l|~). Hence it suffices to prove, that the right hand sides of these two equations are elements of type T(f;~). To decide this, it suffices to show, that the cycleproduct g to the cyclic factor (j...~r(j)) of ~ with respect to f is conjugate to the following two cycleproducts: the cycleproduct g' belonging to the cyclic factor (~,(j)...~,~r(j)) of ~,~,-I with respect to f~, and the cycleproduct g" associated with (j... r(j)) with respect to f,ff~-l. Using 3.4 we obtain

g' ~ f~,(f~,) ~,~,-l"''(f~ ')~, r ,-1(~'(J))

=

f~,f~,~...f

rC~'CJ))

And for g" we have g, = f,ff,-lf,f f,T1...f, f f,-I (j)

43

z , t ~ j g f ,'~ - ' - - " - ~

q.e.d. A last l e m m a b e f o r e we prove, that the type characterizes a conJugacy class: 3~6 If T(f;~) = T(f';~'), then there is a ~" E Sn satisfying = ~,~,~.-I and with the property, that for each cyclic factor of ~ the two cycleproducts with respect to f and with respect to f' are conjugates. ~N

Proof: If T(f;~) = T(f';~') we obtain from 3.3: ~ ~ ~'. Hence there is a ~ E Sn, which satisfies = ~,~-1. The set of all these ~ forms a right coset of the centralizer of ~. Since
(e;~)(f';~,)(e;~) -1 = (f';~) ,

3.5 implies

The cycleproduct to the factor (j... r(j)) of ~ with respect to the mapping f is

f...fr(J) ,
and with respect to f': f, ...f, r~ (j)

Since (I) is valid, there is a ~* from the centralizer of ~,

44

which yields, if it is applied to f': f' .... f'r -(J) ~ f'''f r (j) "

Thus ~" := ~*~ fulfils the statement. q.e.d. Now we are ready to characterize the conjugacy classes (Specht ~I~):

3.7
Proof: a) If (f;~) N (f';~'), then there are f" and ~" so that Cf.;~,,)Cf;~)Cf,,;n,,)-I = (f,,;1)Ce;~,)Cf;~)Ce;~")-lcf";1) -I
= (f';~') ,

and using 3.5 we obtain T(f;~) = T(f';~') as claimed. b) If the other way round T(f;~) = T(f';~'), then by 3.6 there is a ~" E Sn satisfying ~ = ~,~,~,,-I. Now (e;~,,)(f,;~,)(e;~,)-1 = (f~,,;~,~,~,-1) = (f~,;~) Because of 3.5, this implies T(f';~') = T(f~,,','~). And since these two elements are conjugates, it suffices to show, that

C23

We assume now that ~" has been chosen in such a way, that for each cyclic factor (j... r(j)) of

f'--ZrCJ)

(cf. 3.6).
Let gj be an element of G such that

45

f .f r(J) -- gj(f~n...f r (j)lg~1

oe ! !

Having chosen such a gj and starting with u = 0 we obtain from the equations f(-u(j)) elements g-u-1(j)' gj = f' (~-u(j))g-lu I g~-u(j) ~" - - (j) (2)

which are uniquely determined

(1_(u_(r), if

g~-o(j)

is fixed.

Having done this for every cyclic factor of ~ we define the mapping f*:G ~ G by f*(i) = gi' V i E ~ And this mapping satisfies (f.;ll(f&.;~)(f.;l)-I -- (f.f&.f~-l;~) -- (f;~)

(the last equation follows from (2)), which proves (I) q.e.d. Hence G~S n contains exactly as many conjugacy classes as there are types, i.e. (sn)-matrices (aik) whose entries satisfy 3.3.

For the number of types or conjugacy classes we prove now (Specht [I]): ~8 If p(m) is the number of partitions of m for m E N and p(O) := I, the number of conJugacy classes of G~S n is Z P(nl)...p(n s) ,

(n)

if the sum is taken over all the s-tupels (n)=(nl,..,n s) (s=number of conjugacy classes of G) with Z ni=n, 0<_ni E 7.

46

Proof: For a type (aik) we define n i := ~. kaik. Then (n 1,...,n s) k is such an s-tupel, and all the s-tupels occur in this way. n i

is the sum of the elements of the i-th row of (aik) weighted with their column number. Therefore if the other rows are fixed, there

are exactly P(ni) possibilities

for the i-th row to be the row of

a type. And this proves the assertion. q.e.d. If we wish to derive the order of such a conJugacy class, the conjugacy class of G~S n which is characterized (aik),
we

say of

by the type

set ak := T aik , l<k<n .

Then there are

elements of type (a 1,...,an)

in S n. We choose one of them, say ~.

The ak cyclic factors of ~ which are of length k can be distributed into the s conJugacy classes of G in

alkl ~

"'" \

.... ask

. alk!...ask !

ways which are in accordance with the considered type (aik). T, et f:G ~ G be a mapping which yields such a distribution cycleproducts. of the

It remains to show, what freedom of choice is left

for choosing the values of f. To assure that

47

f...f k_1(j) E C i we may choose the values f(j), f(~-1(j)), ..., f(-k+2(j)) at

will and can choose an f(~-k+1(j)) E G so that the complete product is an element of O i ~ G. Hence there exist akl alk!...ask !

(lelk-llcil) ~

mappings f:~ ~ G which distribute the ak k-cycles of ~ as the considered type (aik) prescribes. We have to multiply this number of mappings with the number of elements of type (al,...,an) and take the product over all i and k to obtain the order of the conjugacy class of G~S n which is characterized by the considered type (aik). Thus we have (8pecht

[1]),
3~ The conjugacy class of G~S n consisting of the elements of the elements of type (aik) has the order

tQ nl/Tr aik!(klal/Icil)aik i,k

Therefore the order of the centralizer of an element of type (aik) is i~ aik!(klGl/Icil)aik

Before we can describe this centralizer in detail we have to consider the orders of the elements. At first we notice, that for u E N:

48

3.1o
what implies,

(f;~)u = (ff~...fu_1~u) ,
that the order of (f;~) is a multiple of the order

of ~. The order of ~ is the least common multiple of the lengths of the cyclic factors of ~ (of. 1.11). On the other hand we let u E N be a multiple of the length of such a v_>O that

every cyclic factor of ~. If i E ~, then there is

i = ~v(j) and j is the least symbol of the cyclic factor of which includes this symbol i. Let g be the cycleproduct with this cycle and with respect to f. Then we have associated

f...fu_1(i) = fC~V(j))f(~v-ICjl)...fC~v-u+1(j))
3.11 = fQ~v(j))...f(~(j))g.f(~(j))-1...ft~vtj))-1.g W - -. . ---- . . . . . . - . . . if u = wr for the length r of the cycle including i. The order of (f;~) is the minimal u E N so that u f...f u_1(i) = IG, V i E S, i.e. f'''f~u-1 3.11 we obtain = IS n and ,

= e. Hence from

5.12 The order of (f;~)

(of type (aik)) is the least common mul-

tiple of the products km i of the lengths k of the cyclic factors of ~ with the orders w i of the corresponding
L ,,

cycle-

products with respect to f: l<(f;~)>l = 1.c.m. {k.i] i,k:aik>O .

Hence

(f;~)

is p-regular

(i.e. p ~ l<(f;~)>l)

if p does not dicorol-

vide any one of these products kw i. We have the following lary:

49

3~The

p-regular classes of G~S n are exactly the conjugacy clas-

ses belonging to types (aik) where nonvanishing entries occur only in rows to p-regular classes C i of G and columns with p-regular numbers k. The p-classes correspond to those types (aik) where only in columns with p-power-numbers k and only in rows belonging to p-classes C i of G do vanishing entries occur. The remaining Cypes belong to p-singular classes.

3.12 and 3.13 are generalizations of 1.11 and 1.12: for G = { 1 } ,

i.e. for o n e - r o w e d t y p e s we o b t a i n 1.11 and 1 . 1 2 a t once.

As is well-known, the number of representations of a finite group over the field of complex numbers which have real character is equal to the number of ambivalent conjugacy classes (i.e. classes containing the inverse of each of its elements). As we have seen in section 1, each conjugacy class of a symmetric group is ambivalent, while only the alternating groups At, A 2, A 5, A 6, AIO and A14 are ambivalent (cf. 1.15/1.25). This implies that all the ordinary characters of Sn and all the ordinary characters of these six alternating groups are real. We show, that this is the case for G~Sn, too, if it is valid for G (Kerber [7]):

~.15 If G is ambivalent, then G~S n is ambivalent.

Proof: There is obviously a 1-1-correspondence between the cycles

5O

(j... r(j)) of ~ and (j... -r(j)) of - 1 . Let g be the cycle product to (j... r(j)) with respect to f. Then the oycleproduct to (j... -r(j)) with respect to f-~1 is

(recall that (f;~)-1 = (f-11|~-I)) " If now G is an ambivalent group, then g ~ g-l, y g E G, what implies, that in this case

And this proves the assertion since two elements of the same type are conjugates (of. 3.7). q.e.d. A special case is V m,n: Sm~S n is ambivalent.

3.14/3.15 generalize a result of Berggren (Berggren [1]). He proved the ambivalency of G~S 2 assuming that G is ambivalent. Using this and the associativity of the wreath product multiplication he showed, that the 2-Sylow-eubgroups of symmetric groups

are ambivalent. And this implies, that every 2-group can be embedded in an ambivalent 2-group. Cyclic groups Cp of order p are not ambivalent in general, therefore that not every p-Sylow-subgroup plied by the following: of Sn is ambivalent is im-

51

The ambivalency ambivalency

of G~H implies the ambivalency

of G and the

of H.

Proof:

If G~H is ambivalent, so that

then for every (f;~) E G~H there is

an (f';~')

~,~,_1;~'~ '-I) = (f-111~-I).(I)~_ (f,;~,)Cf;~)Cf,~,)-1 = (f,f,f,-1 This implies,

that every ~ E H is conjugate to its inverse,

i.e. H is ambivalent. And if we choose a constant f:~ ~ G, say f(i) = g, i E G, then for ~ = I we obtain from (I), that there is an f' f,ff,-1 so that

= f-1 what implies g N g-1. G has to be ambivalent as well. q.e.d.

Using 1.25 we have the corollary:

Ai~S m and Sm~A i with i E {1,2,5,6,10,14}

are the only ambi-

valent wreath products of alternating with symmetric groups.

Before we conclude this section with an example let us describe the centralizer of an element (f;~) E G~S n. of the centralizer of a permu-

As in the case of the description

tation in Sn (cf. section 2, 2.31/2.32) deration of a special case:

(fl~) E e~s n ,

we start with the consi-

T~ = (0,...,0,I)

i.e. ~ is a cycle of length n.

52

In this case the type (aik) of (f;~) has exactly one nonvanishing entry, a I in the last col,~mn. Since the centralizers of conjugates are conjugate subgroups, can assume, that (f;1) E G 1, i.e. f(i) = 1G, V i ~ I, f(1) E C i, we

if aim is this only nonvanishing entry of (aik). A subgroup of the centralizer of (f;~) is the cyclic subgroup <(f;~)> ~ G%Sn, generated by (f;~) itself. But (f;~) commutes also with the elements (f';1) E G* whose mappings f' are constant on g and so that their value is an element of f(1) in G: f': f'(i) = g E CG(f(1)), V i E ~ This follows from (f';1)(f;~)(f';1) -1 = (f,ff~-1;~) = ( f , f f , - 1 ) = (f;~)
-

of the centralizer

The subgroup of these (f';1) is the diagonal of the basis group of the subgroup CG(f(1))~S n ~ G~Sn: {(f';1)} = diag(GG(f(1))*) ~ GG(f(1))* ~ G*
.

Let us multipl~ these two subgroups of the centralizer and we obtain 3.18 diag(GG(f(1))*)<(f;~)>.

We would like to show, that this is a subgroup and of the same order as CG~Sn(f;~) and hence equal to this centralizer. we have for r,s E Z:

If (f';1), (f";1) E diag(CG(f(1))*),

53

(f,;1)(f;~)r((f.;1)(f~)s) -I
=

(f,;1)(f;~)r-s(f.-1;1)

(f,f,,-1;1)(f;~)r-s .

Hence this subset 3.18 is a subgroup. If we now assume, that

(f,;1)(f;~) r = (f.~1)Cf~) s

we have
(f.-If,;1) = (f;~)s-r .

This is fulfilled only if s-r m 0 (n). On the other hand we have (f;~)n = (ff ...f n_1;~n) = (f(1),...,f(1);1)Ediag(CG(f(1)).) and hence if r m t (n) there is an (f';1) E diag(CG(f(1))*) satisfying

(f;~)r = (f,;llCf;~)t.
Thus the order of this subgroup of CG~Sn(f;~) is
lalagC%(fC1ll*l<Cf~l>l =

CIQJ/lcilln,

and this is the order of the centralizer of (f;~), as can be seen from 3.9. Therefore we have proved the following first step towards a construction of the centralizer of a general element of G~S n (Ore [I ]) : If ~ is an n-cycle and (f;1) E G I, then the centralizer of (f;~) in G~S n is the product of the diagonal subgroup of the basis group CG(f(1)) ~ of CG(f(1))~S n _( G~S n and the

54

cyclic subgroup generated by (f;~), i.e. CG~SnCf;u) = diag(CGCfC1))*)<Cf;~)> in this special case.

Using this we would like to describe the centralizer of a general element (f;~) E G~S n. Let us denote by ~Jk (1--<J--<aik)the aik cyclic factors of a whose oycleproducte with respect to f are elements of C i _~ G (if aik>O). T,et ~ik be the cycle 3.20 ~Jk = (rJ k ~(rJk)...~k-X(rJk))

so that riJ is the least symbol of this cycle. k We notice that 3.21 ~ = i,~,k ~Jik "

Without loss of generality we can assume, that f is of the following form:

3.22 f(r~k) E Ci , f(~-l(r~k)) =. ..=

f(-k+1(rjk)) = IQ.
rJik

If we now define a mapping fJik:~ ~ G by

==

f(rJ k)
1G

if

s=

elsewhere ,

then 3.24 and these factors (fJ .~J ~ are commutative since they have dis ik' ik" joint cyclic factors if we look at the permutation representation

55

of G~S n. Regarded as elements of subgroups of the form

(Gr G=(r) x . . . x

G k _ l ( r ) ) S ~ ~ G~Sk

the n factors ffJ -~J S are elements of the special form whose ik' ik ~ centralizers were described in 3.19. Using the same argument as in section 2 to get the centralizer of a general permutation we obtain from 3.19 and 3.24 (Ore [1]):

3~

If T(f;~) = (aik) then the centralizer of (f;~) in G~S n is a subgroup conjugate to the centralizer

of the special element 3.24 of this type (see 3.20-3.24).

It is easy to check, that this agrees with 3.9, and that for G = [1} we get 2.32. Example. To conclude this section, we consider a subgroup of S 6, which is a faithful permutation representation of S3%S2:

([1,(12),(13),(23),(123),(132)]

x [1,(45),(46),(56),(456),(465)]) ~ S6

[1,(14)(25)(36)]

We would like to describe its conJugacy classes to illustrate 3.7 and give a complete system of representatives of these classes. This we shall do according to the ordering

01 := [13,

C2 := [ ( 1 2 ) , ( 1 3 ) , ( 2 3 ) ] ,

C3 := [ ( 1 2 3 ) , ( 1 3 2 ) ]

of the conjugacy classes of S 3.

56

The order of this group is 3!22! = 72. (i) The image of the element (f|~) = (f(1),f(2);~) := (1,1;I)

under the permutation representation described in 2.24/2.25 is of course the identity of S 6 and hence it constitutes a conjugacy class by itself. The cycleproducts belonging to the two 1-cycles of the permutation ~ = I = (I)(2) E S 2 with respect to this mapping f = e are (1) ~ (2) ~ Therefore 1E $3 ,

I E S3

(f;~) = (e;1) = (1,1|I) is of the type

[i l.
(ii)The cycleproducts belonging to the factors of the permutation of (fl~) := ((12),1;I) are (1) ~ (12)
(2) 1 .

Hence this element is of type

The image of this element under the permutation representation is f(1)af(2)G-11 * Thus (12) E
S6 =

(12)-1-1

(12) E

S6

is a representing element for this conjugacy

5?

class of S3%S 2. Because of 3.9 the order of this class is

72/(3!/1)1(3:/3) 1 = 6.

(iii)((123),I;I)

is of type

. A representative

element of

this conJugacy class is (123) E S

as can be seen analogously 6 to (ii). The order of this class is 4. . A permutation representing

(iv) ((12), (12) ;1 ) is of type

this conJugacy class is (12)(45), the class is of order 9. (v) T((123),(123);1) = .This class is of order and re-

presented by (123) (456).

(vi) ~(1,1~(12)) = . (1,1~(12))
i s mapped o n t o ( 1 4 ) ( 2 5 ) ( 3 6 ) ,

hence this permutation is a representative conjugacy class of order 6. (vii)T((12),(123);1) =

element of this

.The image of this class of $3~S 2

contains (12)(456), and its order is 12. (viii)The cycleproduct belonging to the permutation of ((123),
(123);(12)) is (12) ~ f ( 1 ) f ( 2 ) = (132) ,

hence T((123),(123);(12)) = li !] " is

= (153426)

The image under the permutation representation

f(1)af(2)o-1(12) * = (123)(456)(14)(25)(36)

The order of this class

is 12.

58

(ix) T((12),(123);(12)) = der of the class is 18.

. The image i s (15)6263), the or-

These calculations will be of use later on for the evaluation of the character table of $3~S 2. It will be useful for these calculation to observe, that the groups Sm~S n possess two normal divisors of index 2: Sm%Sn+:= Sm~S n G Amm and Sm~=~ n. Having noticed this we only need to evaluate a small part of the character table and then use the symmetries arising from this fact. Thus it is helpful to have the representatives of the conjugacy classes in permutational form as well as in the form (f;~) as in our example. We see at once that the conjugacy classes described in (i), (iii), (iv), (v) and (ix) are the classes consisting of even permutations and hence these classes form $3~$2 +. ~oreover the classes (i), (ii), (iii), (iv), (v) and (vii) are the classes consisting of elements of the form (f;1) and hence they build up the normal divisor S3~A 2 = $ 3 ~ I ] = S~ ~ S3xS 3.

We have now finished summarizing the group-theoretical results which we need to describe the representation theory of the symmetric and alternating groups as well as of wreath products.

Chapter II

Representations of wreath products

We would like to describe the representations of wreath products, especially of wreath products G~S n with symmetric groups. A first section contains a preparatory s~mmary of the ordinary representation theory of the symmetric group. The following section contains the representation theory of wreath products of finite groups over an algebraically closed field.

6O

4. The ordinary irreducible representations

of the symmetric group

The groundfield is the field C of complex numbers. Since C is algebraically closed and of characteristic zero, the number of irreducible C-representations of Sn is equal to the number of conjugacy classes of S nAs we have seen in the first section, there is a l-l-correspondence between the conjug~cy classes of Sn and the partitions a = (~1,...,~h), a i E N, ~jAaj+1 (1~j<_h-1), Zai=n of n. Hence if we can associate with each of these partitions an

irreducible C-representation of Sn so that representations associated with different partitions are inequivalent, we have a complete system of irreducible C-representations of S nHow this can be done we shall describe now. The procedure using a consideration of primitive idempotents of the group algebra is well known, but a very useful hint to clarify this process by using Meckey's intertwining number theorem we owe to A. J. Coleman (Coleman [1], cf. also Bayar [I], Burrow [1 ],[2], GthndGzalp [1], Malzan [I], Munkholm [I], and the hint following 2.29 in Robinson [5]).

61

Togelher with a partition ~ we consider the Young-diagram

.

...... ... ... .

a I nodes a2 nodes
e e e

[aS:

e e e e e e e e e

..... (cf. section I) and its first Young-tableau

I 2

~hnodes

............

a1

4.1

T 1 .=

~1 +1 1 +2 " ' " . ..............

eeeeeen

al+a2

It follows from the results of section I, that the groups H I and a V1 of the horizontal and vertical permutations subgroups with the property
4.2 H1 n V; = [1} .

of T~ are Young

Let us denote by IH~ the identity representation AV~ the alternating representation
== (1) ,

of H a and by I

a of V1' defined by
v

4.3
:=

(I)
(-1),

, V ~ E -I

Va+.

= VI N A n " of Sn induced

V ~ e -1"1

If IH~ t Sn and AV~ t Sn are the representations

by IH~ and AVe, the following is the crucial theorem:

4,4 IH~ ~ Sn and AV~ t Sn have exactly one irreducible constituent in common, in each case with multiplicity I.

62

Proof: The assertion is fulfilled if and only if the inner product (IH~ t Sn, AV~ ~ Sn) of the corresponding characters IHV t Sn AV~ t Sn X and X satisfies 4.5 (IH~

Sn,AV ~

Sn) = 1

(see Curtis/Reiner [I], Ex. 31.1). ~or the inner product is defined by

(IH t

~ Sn, A V 1

Z xIH1tSn(~)AV11Sn(~ -1 ) Sn) :=

and hence 4.5 implies the assertion because of the orthogonality relations of the irreducible characters. To prove 4.5 we notice, that this inner product is the intertwining number of the two induced representations. Hence u s i n g Mackey's intertwining number theorem (Curtis~einer we obtain [I],(44.5))

(IH~ t Sn,AV ~ t S = i(IE~ n)

r Sn,AV V r

Sn)

~
HI~V 1

i(IH~ ~ Eln~Vl~-1,(AV~)~ ~ Eln~Vl ~ ~ ~-I),

if the sum is taken over a complete system of pairwise different a ~ ~ double cosets HI~V I of H I ~and V I in Sn, if "~" means restriction and if we denote by (AVe) ~ the representation (of ~VV~ -I) conjugate to AVV and defined by ( A V ) ~ ( ~ ' x -I) := AV (~'), V ~' E V I Now the intersection H~ ~ ~ symmetric subgroups, say -I is obviously a direct product of

63

HI

G ~V

-1

i,j 8zij

(cf. 1.31). And therefore the restriction IH~ ~ ~-~1.~Vl -I = I(H~N~V~-I) ~ is equal to the restriction (AV~)~ $ HIN~VI~ -I = A(H~G~V~-I) ~ if and only if zij~1 , V i,j, i.e. if and only if H~ R ~ V ~ -1 = [1] .

Hence

~ -1 = { I, if H I ~ V I ~ ={I] 0 otherwise .

From 1.35 we know, that there is exactly one double coset

~ ~ HI~V m with the property HIN~VI~ -1 = [I] (namely the double coset 1

HI~V~, see 4.2). This implies 4.5 and the theorem is proved. q.e.d. Let us denote by [~] the equivalence class of this uniquely determined common constituent: [~] := IH~ t Sn n AV~ t Sn Since with the notation (I n) := (I,...,I) we have IH~ ln) = I{1] = AV~ n) , the induced representation IH~ ln) ~ Sn as well as AV~ n) t 8n is the regular representation RS n of 8n. Thus IH~ ln) ~ Sn O (AV~ ln) t Sn = AS n) = RS n G AS n = AS n, (IH~n) ~ Sn = ISn) ~ AV~ n) ~ Sn = ISn O RS n = ISn .

64

Using notation 4.6 we get therefrom:

~-7 In] = Is n ,

[1 n] = AS n

Let us denote by "@" the outer tensor product multiplication (following the notation of Ourtis/Reiner [1], cf. % 43) and substla tute 8a and 8a, for the isomorphic subgroups H I and V~ from 4.7: We obtain

4.8

zs~ = [~1] ~ ' " ~

[%] = ~ [~i]
, i

a'

ASa, = [1 a'~] ,...@

[1 cr'~'] = @ [1 3.] i

The induced representations will be denoted as follows: V[ai]

a'

= [al ]''" [@h] := (~i[~i]) f Sn '

a' a:

"[~[1 3.] = [1 1]...[1 a]~''] := (@[li 1]) f Sn

Therefore a second formulation of 4.6 is (cf. Robinson [5],2.29):
~I I

4,10

[a] = [a1]...[al:l]

13

[1 1]...[1~h']

Using again the notation of Curtis/Reiner [I] we denote by "@" the inner tensor product multiplication. By definition of the alternating representation we have

[1 n] [1 n] = [n] .
Hence [a'] = IS~, t Sn n AS a ~ Sn = (ASa,~S n n ISa~S n) @ [In], thus

65

4.11

[a']

[~] @ [1 n] [~]
and

This means, that the representations

[~']

differ only on

the odd permutations and there only in the sign. 4.6 characterizes certain irreducible C-representations of Sn, and we would like to show that the system of these representations [m] is a complete system of pairwise inequivalent irreducible C-representations (or ordinary representations as they are also called) of S n~o prove this it suffices - as has been said at the beginning of this section - to show, that for ~ + ~ [~] and [6] are inequivalent. To prove this we consider minimal left ideals of the group algebra CS n of Sn over C, which afford the representations

We know, that the simple two-sided ideal of the group algebra CG of a finite group G consisting of minimal left ideals affording the irreducible C-representation of G with character C is generated by the central and up to a numerical factor idempotent element 4.12 D C(g-1)g gEG If this irreducible representation with character { is onedimensional, then the simple two-sided ideal is a minimal left ideal itself, since the number of the minimal s1~mmands is equal to the dimension of the afforded irreducible representation as is well

66

known. In this case, the element 4 . 1 2 is even a primitive idempotent (up to a numerical factor). Applying this to IH~ and AV~ and ! | setting 4.13
~(I "= I~ ~ .~

ep := ~I if p is an even/odd permutation,

we obtain:

a a 4,14 IH~ is afforded by CHI](I, AV~ is afforded by CV1?rI.

These elements ~I and ~1 are up to a numerical factor primitive idempotents of the subalgebras CH~ and CV~ of CSn. Regarded as elements of CS n they are still idempotent up to a numerical factor but of course in general no longer central elements or primitive. The left ideals generated by afford the induced representations:

4,~ ~t

~ ~ ~o~o~

~
by

c~

~
--

~s~

~
CH 1

~ ,
.

AV~ ? Sn is a f f o r d e d

CSn~ ~ ~
--

CSn @ a C V ~ CV I

And we would like to show, that the product

e ~ -= ~( ~ I

generates a minimal left ideal of CSn affording []. At first we notice, that because of HV G V I = {I} the coefficient of 1Sn in e~ is I and hence

4.17

e~ + o

67

4.5 implies,
4.18

that the C-dimension of ~CSn~r ~ is I:

z ( iCSn 1= = c) =

Therefore we obtain with 4.17 John von Neumann's lemma (see Boerher [2], IV, theorem 2.9): 4.19
c[ ~ ~1 CSnlrl = Ce~

"

Applying this to the special element ~1~I E CS n we get 4.20 (e)2 = ~ I =~e , ~ E C.

It remains to prove, that this complex number ~ is unequal to zero and that e I is primitive. can be evaluated by calculating in two ways the trace of the linear transformation of CS n afforded by the multiplication a CS n with e 1 from the right hand side. At first we assume the basis of CS n to be adapted to the submodule CSneV, i.e. that the first (Sne ~ : C) =: f~ basis vectors span CSne V. With respect to such a basis the multiplication with e~ is described by the matrix of

on account of 4.20. If on the other hand we choose the elements ~ of Sn to be the basis of CSn, the trace of the multiplication with e~ is obviously n!-times the coefficient of ISn in e~, which is I as we have

68

mentioned above. 0omparing this with 4.21 we obtain 4.22 n! ~ = (CSne~,c) n!

= ~

.
f~/n', an idempotent

Hence e~ is up to the numerical factor - 1 = 1 element of CSn, e I is essentially idempotent

To show the primitivity of e~ we notice, that 4.23 s C nel~ ~ = ~ : CSn~1~ I _ (C S n ~ ) ( C S n ~ ~)

.

The two factors of the right hand side of 4.23 are direct sums of minimal left ideals. Since nonisomorphic minimal left ideals annihilate each other, we obtain from 4.4, that this right hand side of 4.23 is either the ideal {0} or a minimal left ideal which affords [~]. Because of eV $ 0 the left ideal CSne V is not the zero ideal and hence 4.10 implies:

(f~/n~)e I is a primitive idempotent of CS n and the minimal

left ideal CSne V generated by e I affords the irreducible C-representation [~] of Sn.

To show the completeness of this system of irreducible ordinary representations C~S of Sn it therefore suffices to prove the following : 4.25 ~ ~ ~ ~ CSne ~ ~ CSne1~ -

69

We shall prove this with the aid of a lemma. Before stating this lemma we order the partitions, diagrams and representations according to the row lengths mi" We say that m respectively Ca] precedes ~ respectively C~] (a and ~ partitions of n), for short: C~] ~ ~ ] , if the first nonvanishing difference ai-~ i is positive.

Now the lemma reads as follows:

4,26 If [~] > [~] and T ~, T ~ are Young-tableaux with diagrams [a] and CP~, then there are at least two symbols which appear in T ~ in the aame row and in T 6 in the same col~m~.

Proof: If this were not the ease, the a I symbols of the first row of T ~ would appear in different columns of T 6 61~a 1. Since a > 6 this implies ~I = 61" A vertical permutation - which doesn't disturb the distribution of the symbols in the columns of T 6 - transfers these ~I symbols to the places of the first row of T 6. Disregarding this new first row and using the same argument as above we get a2 = ~2 and so on, arriving finally at ~ = 6 which is a contradiction to the assumption ~ > ~. q.e.d. Pro9f of 4.2~: If ~ ~ 6 we can assume without restriction that so that

If ~ E Sn, 4.26 implies that in

7O

I)
~T~ - m o o o I ~ I o o m a o l o

there are two symbols, say s and t , appearing i n the same row of ~T~ and i n the same column of TI~. T h e i r t r a n s p o s i t i o n
to ~j~-1,

( s t ) belongs

the group of the horizontal permutations of ~T~ as well

as to VI~. Thus

(st)~

-1 = ~ - I ,

,1~(st) = _ , i~,

from what follows, that e 1 ~ e ~ -I = -e1~(st)~eV~ -I = - e 1 ~ e ~ -I . e 1 ~ e ~ -I = O, V ~ES n, ~ e1~e~ = 0, V ~ESn, ~ e1~xe~ = O,V xECS n. And it is well known, that this implies the statement (cf. Boerner [2], III, theorem 3.8). q.e.d. We s!imm~rize, what we have proved: The representations [a] defined by 4.6 form a complete system [[~] := IS~$S n N AS~,~S n I a partition of n] of pairwise inequivalent and irreducible C-representations of S n. Thus we have completed the first step towards an explicit description of the ordinary irreducible representations of Sn. The representing matrices themselves were given by A. Young (Young [2]), we shall describe the derivation in detail in a

71

following part. Here we shall only sketch the next steps briefly. We return to the definition 4.16 of the generating primitive idem5 potents. To define e I we have used only the first Young-tableau TIm with Youmg-diagram [~] . The element e I = ~i~1 constructed with the aid of the groups H I and V~ of the horizontal a and vertical permutations of T I generates a minimal left ideal out of the simple two-sided ideal in CSn to which the irreducible ordinary representation [51 corresponds. This simple two-sided ideal is a direct sum of fa = (CSne~:C) minimal left ideals which are isomorphic to Sne ~. Thus we

ask for elements generating the remaining minimal left ideals out of this simple two-sided ideal. Presumably some of the elements e i := ~i~i 5 constructed analogously to e I but with the aid of other tableaux a T i generate these left ideals. This is actually the case (that CSne ~ ~ CSne ~ is trivial by definition of [5]). Obviously there are n! different tahleaux with Young-diagram [5], namely the tableaux ~T I , ~ E Sn. We pick out some of them, the so-called standard-tableau~, characterized by

the property, that in such a tableau the symbols in each row and

72

in each colnmn are in increasing order, i.e. that the symbol in the position (i,j) (i-th row, j-th colnmn) precedes that in the (k,1)-position if i<_k and j~l. E.g. 123 45 124 35 125 34 134 25 135 24 [3,2].

are all the standard-tableaux with Young-diagram

Let us denote by f~ the number of standard-tableaux with Youngdiagram [a]. It will appear, that fa = fa = (CSne?:). We arrange the standard-tableaux T i in dictionary order, i.e. i<j if the first nonvanishing difference of symbols located at ~ the same place in Ti and Tj (comparing the symbols in a row from the left to the right and the rows downwards) is negative. The

above arrangement of the standard tableaux with diagram [3,2] provides an example Let H ~ a n d of this ordering.

V i be the groups of the horizontal permutations and of of T~, let ' ~i

~ .=

the vertical permutations 4.28 and 4.29

:= ~ ~

Ei :=

~ '~ ~EH i

pEV i

a ~

~ ~p~p,

1~i<_f~.

ei

~i~i = ~EHi,PE~ i

We would like to show, that 4.30 ~ CSne ~ ~,i

is a direct sum and equal to CSn. This would imply f~ = f~ and that $ CSne; is the simple two-sided ideal of CSn, whose s1~mmands i

73

CSne ~ afford [a]. The first step towards a proof of 4.30 is the proof of
a

4.~I

i < j

eje i = O.

And this will be shown with the aid of the following lemma: If i<j, then there are at least two symbols appearing in the same row of T i and in the same column of T .

Proof: Let (k,1) be the place of the first node of [a] replaced a by different symbols, say by s resp. t in T i resp. Tj. i<j implies s<t. We ask for the place (m,o), where s appears in T~. On account of the standardness of T~, t precedes all the symbols located at places (p,q) with p_~k and q~l. Since (k,1) is the first place occupied by different symbols, this implies m>k, o<l. Thus the place (k,o) is occupied by the same symbol, say by r, in T. a l as well as in T ~. Hence r and s fulfil the statement. q.e.d. Denoting by r and s two symbols appearing in the same row of T i and in the same column of T~., we have for their transposition (rs) : ~ a ~ a (rs) E HinV a ~ eje i = -e~(rs) e i = -eje i~ ~ j This proves 4.31. To proceed with the proof of that 4.30 is a direct sum and equal to CSn, we consider an equation ~ eje i ~= 0

?4

xle I + ... + xf

ae

(x i E CSn). If we multiply by e I from the right hand side, since 4.31 is valid, we obtain ~xle ~ = O, hence xle ~ = O. Then we mula tiply by e2 and get x2e 2 = 0 and so on. This proves, that the sum 4.30 is a direct sum. A combinatorial 4.33 consideration (see Boerner [2], IV, ~ 7) shows that Z (f~)2 = n! . a

Since 4.30 is a direct sum we have fu_>f~, together with 4.33 we conclude, that f~ = fa and this completes the proof, so that the

following is valid: f~ .34 CS n =

Sne ~ a i=I .

Hence 4.35 f~ ~ CSne ~ i=1

is the two-sided ideal to which [~] belongs. The theorem of Wedderburn says: ~ CSne ~ is isomorphic to the ring i of (faxfa)-matrices over C. Hence all what remains toconstruct representing matrices themselves eik~ of ~ CSne ~ satisfying i 4.36 (Gjk the Kronecker-symbol: Thus the matrices 4.37 is to find a basis of elements

ea ~ = Gjkeil ijekl Gjk=O if jSk, =I, if j=k). of

(d~k(~)) built from the coefficients f~ ~ = ~ d~ " "e ~ i,k=1 ik t~; ik

75

are the elements of ~aS. Por this the reader is referred to Boernor CIS, C2S. Corresponding to three ways of choosing the basis elements eik there are three forms of the representing matrices: Youn~'s natural form (d~k E Z), Young's seminormal form (dik E Q) and Young's crthogonal form (d~k E R, the matrices are orthogonal). The natural form is derived in Boerner ~IS, the orthogonal and seminormal form in B o o m e r form now. To get the matrices representing elements of the subgroup Sn_ I Sn (consisting of the permutations fixing the symbol n E ~) in reduced form, i.e. with the matrices of the irreducible constituents along the main diagonal and zeros elsewhere, we choose an ordering of the standard-tableaux with respect to this symbol n: f~ I to obtain the last letter sequence T~, ..., T~ of the standardtableaux with Young-diagram ~ S , we take at first the standardE2S. we shall describe the seminormal

tableaux containing n in the last row, then the standard-tableaux containing n in the last but one row and so on. Then we order the standard-tableaux containing n in the same row with respect to n-S and so on. E.g. 4.38 I 2 3 45 I 2 4 35 133 4 25 I 2 5 34 I 3 5 24

is the last letter sequence of the standard-tableaux with Youngdiagram E3,2~. To distinguish between the last letter sequence i and the dictionary ordering we have exchanged the indices: T~ may

76

be different from T iBefore we can formulate the theorem we have to introduce the axial i i distance d~(r,s) of two symbols r and s in T~: if r is located at (ir,Jr) and s at (is,Js), we define: 4.39 i d~(r,s) := (jr-Js) + (is-i r) .

Thus d~(r,s) is the number of steps we need to come from r to s, if steps to the left and downwards are counted positively and steps to the right and upwards are counted negatively. Now the theorem describing Young's seminormal form of [m] reads as follows: I fm 4.40 If Ta, ..., T~ is the last letter sequence of the standardtableaux with Young-diagram

[~],

then from the matrices (t,t+1)

(d~k(t,t+1)) representing the transpositions (l~t<_n-1) can be built a representation if we set

equivalent to [~]

(i) d ~i(t,t+l ) := !1, if T~ contains t and t+1 in the same i row/c ol~mn, (ii)and for the submatrix

d~i(t,t+1 )
d~i(t,t+1 )

dij(t,t+l djj(t,t+1

da(,~+I) -1 1-di(t, t+l )-2~

I
I

k
if T a = (t,t+1)T (iii)d~j(t,t+1)

~
and i<j,

:=

:= 0 for all the other entries.

77

Later

on

we shall use this theorem for the evaluation of wreath products.

of

examples

of representations

Por this we shall characters

also use some results about the ordinary irreducible of the symmetric group which we snmmarize now.

A tool of great utility is the theorem, [a] can be written in the determinantal

that the character form

Ca of

as a linear combination characters

(with rational

integral coefficients)

of

induced by identity representations as follows:

of Young subgroups.

4.41 has to be understood

~[~i+~(i)-i] 4.42 ~[~i ] ~ = ~ e~X ~ES h ,

if X

denotes the character of -- [~i ] (see 4.9),

if ~i~O, V i,

and if we set X minantal

~[~i]

= O, if one ~i<O, and if we set in the deter4.41:

expression

[0]
For example

:= 1,

[m]

.-'- O, i f

m<O

[3]
[3,1 2 ] = 0
what means, that

[4] [1]

[5] [2]

= [3][~][,]

- [3][2]

- [4][~]

+ [5]

c(3,12)

= x[3][1][~]

_ x[3][2]

_ x[4311]

+ x[ 5]

Above all it is remarkable,

that 4.41 like 4.4 can be proved using in section I and Mackey's

the results about double cosets mentioned

78

intertwining number theorem (see Coleman [1], of. also ettudGzalp [I]) so that we can get a homogeneous and lucid derivation of these fundamental and classical results following Goleman's hint to use Mackey's theorem. A corollary of 4.41 is C~(~) = { (-1)r'o elsewhereif [~]=[n-r,lr],o<r<_n-1.

~or the only Young subgroup containing n-cycles is Sn itself. Hence an n-cycle ~ has a nonvanishing character value C~(~) at most when [n] occurs in the determinantal expression 4.41, and this is obviously the case only if [~] is of the form [n-r,lr]. In this case we have fin-r] [n-r+1] ... In-l] In] [n-r'sr] = I..i..... !]!...i~i.!~]!.!~!
I
(

0
+ ....

0
.,

...
............

[1] I
,

1)r[n]

eaoh s~mm~na ~ In]

hence 4.42 is valid. To prove 4.41 we introduce the concept of a hook, whose importance was first noted by T. Nakayama (Nakayama [1]). v~ith the aid of this concept a very simple recursion formula for ~a can be formulated and a very simple equation for f~ can be

79

given. A hook we call each F-shaped arrangement H~j of nodes out of a Young-diagram [~] which consists of the (i,j)-node, the corner of the hook, as well as of the (i,k)-nodes, k>J, which form the arm of the hook, and the (l,j)-nodes, l>i, which form the leg of the hook H~j.~ The (i,ai)-node is called the hand of the hook, the .. .. (a~,j)-node is called the foot of H ~ . ij"

/
Hij

corner

hand

.J .....

I ,

leg

arm

foot

The number 4.43 hij := ~i - j + ~j - i + 1 of the

of nodes the hook H~j consists of is called the ~ hook. The number
4.44

I i j "= e~ - i

is called the leg-length of Hij. a To Hij corresponds the part of length h a of the rim of [u] ij consisting of the nodes between the hand and the arm of H~j~ inclusive. E.g.
@ @ @

80

where the encircled nodes indicate the part of the rim which corresponds to the hook H(3'2'12)11 ~ [3'2'12]" This associated Rart of the rim will be denoted by Rij And it is important, that the result [~]\R~j of removing R~j from [a] is a Young-diagram again or equal to [0] (what is sometimes called the zero-dia6ram). E.g.
iJ ~J

[3,2,12]\R13'2'12)1

f "

[1] =

"

/
Using this notation we can formulat~the following two very important formulae:

~.45 ("Murnaghan-Nakayama-formula") If ~ Sn is of type T~ = (al,...,a n) so that ak ~ 0, and ~* E Sn_ k is of type T~* = (al,...,ak_1,ak-S,ak+1,...,an), then

C~(~) if we set C [0] "= I~ .

i,j:hij= k

This is the recursion formula for the ordinary irreducible characters of Sn. The formula for the dimension of [~] we owe to Frame, Robinson and Thrall (Frame/Robinson/Thrall it is of fascinating simplicity: [I]), and

81

$.$6 The dimension of [ ~

is the quotient of n! and the product

of all the hook-lengths:

! n.l.T

i,j

hij

4.45 implies, that C~(~) = O if ~ contains a k-cycle but [a~ contains no k-hook, and it implies 4.42. It should be observed, that the order of removing two parts of the resulting dia-

the rim associated with hooks is immaterial, gram will be the same in both cases.

Now we have mentioned the most important results about the ordinary irreducible representations. ducible representations later on. We consider the representations Certain analogues concerning re-

can be formulated, which will be of use

[~

an i r r e d u c i b l e of Sn-

representation

of Sm, [ ~

an i r r e d u c i b l e

re-

presentation

There is an analogue to 4.40 which describes the representing matrices of C ~ . To formulate this we have to generalize the notation of the Youngdiagram. To do this we consider a diagram ~ and a diagram ~

which can be superimposed upon CmS, upper left hand corner upon upper left hand corner such that ~ is contained entirely within

82

The residuum of [ a ] and denoted by

not covered by liB] is called a skew-diagram

[a]

[p]

E.g.
[v,6,3,1] ['z,2,1] =

[6,3,1]

[2,1]

([a]-[~] is by no means uniquely determined in terms of [a] and

[p]).

Special cases o f skew diagrams consist o f two disjoint diagrams:

C~ P]

"=

E.g.
[4,3,2,1] [2 2 ] = [ 2 , 1 1 2 , 1 ] = "

We shall consider only these special cases of skew diagrams. i P Tableaux and standard-tableaux Ta;~ are defined as before

83

The number of standard-tableaux with Young-diagram [a;~] is obviously

448
: ,

i.e. equal to the dimension of [a][~]. The theorem analogous to 4.40 reads as follows (Robinson [5], 3.1): 1 fa;~ ^ 4.49 If Tail,. ..,T a;p (f~;~:=(m+n)!faf~/m!n!) . . . . . is the last letter .

sequence of the standard-tableaux with skew diagram [~;~], then we can build a representation equivalent to [~][~],

if we take for the entries of the matrices (d-'~(t,t+1)) representing the transpositions (i) (t,t+1):

d~l~(t,t+1 ) = ~I, if t and t+1 occur in the same rbw/ i column of T~;~,

(ii) [d~ ~(t't+1) ~ijaa; ~(t't+11~. Id i~;~(t,t+1)-I 1-d~; ~(t,t+1)-~i -I [d~(t,t+1) or d~(t,t+1 := 1 d~;~(t,t+1) J

[~

2] T~;~ in the same diagram constii

if t and t+1 occur in tuent or not. (iii)d~(t,t+1)

!

= 0 for all the other entries.

The formula analogous to the Murnaghan-Nakayama-formula reads as follows (Osima [I ]) : 4.50 If ~ E Sm+ n is of type T~ = (al,...,am+ n) with ak ~ 0 and if

84

~*

E Sm+n_ k

is of type T~*=(al,...,ak_1,ak-l,ak+1,...,am+n),

then we have for the value X~;~(~) of the character of [~][~]

on ~

i,j:hi~=k
if we set X [O] := I.

( -1)

l~ X

The following theorem (stated by Littlewood and Richardson and proved first by Robinson) describes, how we can get the irreducible constituents of [~][~]:

4.51 ("Littlewood-Richardson-rule") The irreducible constituents of [~][~] are exactly the representations [y] of Sm+n, whose diagrams arise by adding the nodes of [~] to [~] according to the following rules: (i) To [~] add the nodes of the first row of [~]. These may be added to one row or divided into subsets preserving their order and added to different rows, the first subset to one row of [~], the second to a subsequent row, the third to a subsequent to this and so on. After the additions the resulting diagram may not contain two added nodes in the same column. (ii)Next add the second row of [~] according to the same rules followed by the remaining rows in succession and such that each node of [~] appears in a later row of

85

the compound diagram above it in IriS.

[]

than that node immediately

A special and very important case is the branching theorem for the irreducible o rdinar 2 representations of Sn:

4.52

[~] ~ Sn_ I

i'j:hiJ=1
theorem

It is easily to prove this using 4.51 and the reciprocity

of ~robenius. Por 4.51 says, that the constituents of [~] $ Sn+ I = [~][I] are exactly the representations [] of Sn+ 1, whose dia-

grams arise by adding a node to [~]. Thus by Frobenius' reciprocity theorem the diagrams of the constituents of [~] ~ Sn_ I arise by subtracting a node in all the possible ways.

Concluding this section we would like to consider the connection between the irreducible C-representations of Sn and An.

A first remark follows immediately from 4.11:

4.53

[~2 ~ A~ ~ [~'2 ~ An

Thus [~] and [a'] are associated representation s in the sense of Clifford's theory of representations of groups with normal divi-

sors (Clifford [1], cf. also Boerner [2], III,~ 13). But from Clifford's theory we get much more:

4.54 (i) a + ~' ~ [~] ~ An = [~'] ~ An is irreducible.

86

(ii) a = ~' ~ ([a]=[a']) ~ A n = [~]+ + [a]- with two irreduoible and conjugate representations [ ~ of A n (i.e.

[~]+(a) is equivalent to [m]-, V a E Sn, if

[~]+(a)(a~a-1)

:= [ ~ ] + ( ~ ) ,

v ~ e An).

Hence for the characters we may use the Murnaghan-Nakayama-formula again in the case ~ ~ ~'. If ~ = ~' difficulties arise also because of the splitting of the Sn-classes. But we have (see Boerner [2]) a result of Frobenius:

("Frobenius ' theorem") If a = a' the class with the partition := (h~1,h22,...,hkk) is a splitting Sn-olass C ~ = O ~+ U C ~-. On this class C~ has the value

C~ = ( - 1 ) ( n - k ) / 2 , and the values of Ce~ on C~ are

Cp a+

1 ~ ~(C~ _ +

4 r,p hii V

if we denote the irreducible constituents of [~] ~ A n in this way. On all the other classes with partitions V $ ~, we have
5+ _

= ~/2

87

We recall the considerations of the first section concerning the ambivalency of alternating groups (cf. 1.25) and take into account, that the ambivalency of a group is equivalent to the reality of its ordinary irreducible characters. Hence the tables of A I = A 2 = (I], A5, A6, AIO and A14 are the only character tables of alternating groups containing only real entries, while e.g. A3 (3,0,0)
1

(0,0,1) + 1

(0,0,1)1

[3] [2,1] + [2,1]-

1
1 ( -i ~ ) / 2 -1

(-1-qg)/2
(-1 + i ~ ) / 2

is the character table of A 3. The table of Sn is real for each n. It should be mentioned, that Sn, A n and the wreath products of the form Cp~S n (p a prime, n_~3) are Na6ao-groups,i.e. they are the only groups with these character tables (see Nagao [I], Oyama

[1], Yokonuma [ 1 ] ) .
For the construction of the representing matrices see Puttaswamaiah [I] and Puttaswamaiah/Robinson [I].

These theorems contain the results we need from the ordinary representation theory of Sn and A n. Below we shall summarize

the results of the modular theory. The results from the ordinary theory suffice to give a detailed description of the ordinary representation theory of G~S n. But before deriving this theory,

88

we would like to consider the more general case G~H, field will be assumed to be algebraically closed.

the ground-

89

5. Representations of wreath products

==

Let K denot~an algebraically closed field, G a finite group and H a permutation group of degree n, i.e. a subgroup of Sn, the symmetric group on Q = [1,...,n]. We would like to derive, how the irreducible K-representations of G~H can be constructed. For K = ,W. Specht has done this in 1933 (Specht [2]) after having treated the special case G~S n (Specht [I]). His results can be generalized to groundfields K of any characteristic, if K is assumed to be algebraically closed. Using the theory of representations of groups with normal divisors given by A.H. Clifford in 1937 (Clifford [I]) the derivation of the shortened considerably (Kerber [2],[4]). now. We apply Clifford's theory to the normal divisor G* = G I ... Gn _~ G~H , the basis group of G~H, which is a direct product of n subgroups G i isomorphic to G: G = G i := ~(f;1 H) [ f:g ~ G, f(j)=1 G, V j$i} ~ G~H (cf. section 2). Since the groundfleld K is assumed to be algebraically closed, desired results can be We describe this

9O

the irreducible K-representations sor products 5.1

of G* are exactly the outer ten-

~* := F1 @'''@ Pn

of irreducible K-representations F i of G with the representing matrices 5.2 F*(f;1H) := F1(f(1)) x ... x Pn(f(n)) (Kronecker product). G~J~F. of this repre-

At first we have to derive the i n e r t i a ~ sentation F*, which is defined by 5.3 o~F. := C(f;~)

F *(~;~) ~ F*~

("~" indicates equivalency, F*(flg)(f';1 H) := P*(fl~)-1(f'lIg)(f;~) P*Cf-~1 f'~-lf~-l;1H ) = F*(cf-lfvf)~-l;1H ))" Since G* ~ G~HF. this group is obviously a product 5.4 G~HF.= G*H~.

of G* with a subgroup H~. of the complement H' of G*. Hi. will be called the inertia factor of F*: 5-5 We notice, that 5.6 F*(e;~)(f;IH) = F*(e;~)-1(f;IH)(e;~) = F*(f _I;IH) H~. = [(e1~) J F*(e;~) ~ F*] .

To describe the inertia factor explicitly we distinguish the irreducible K-representations 5.1 with respect to their type:

Let PI,...,Fr be a fixed arrangement of the r pairwise inequivalent K-representatlons of G. We call F* = FI~...~P n to be of type (n)=(nl,...,n r)

91

(with respect to the above arrangement),

if nj is the

number of factors Pi of P* equivalent to PJ. Let F* be of type (n) and let Snj be the subgroup of Sn (>_H) consisting of the elements permuting exactly the nj indices of the nj factors Fi of P* which are equivalent to F j. We set 5.8 Sin ) .-= S'nl ... S'nr with S'nj := [(e;~) I ~ E Snj]

We would like to prove, what is suggested by 5.6:

5.9

E'

n Sin )

Proof: 5.2 and 5.6 imply F*Ce;~)Cf;IH)

= Fl(fC~C1)))x...XFnCfC~Cn)))

The question is, for which ~ this representation is equivalent to F*. Since P* as well as p.(e;~) are irreducible representations, we can use a character-theoretical argument.

(i) If ~ E HN (n)' then the trace of Fi(f(~(i))) is equal to the 'S trace of P~(i)(f(~(i))). Thus tr'F*(f;IH) = tr P*(e;~)(f;IH), V f, ~ F* F,(e;~)

H' nsI= >

(ii)If the other way round (e;~) is an element of the inertia factor, we have tr P*(f;1 H) = tr F*(e;~)(f;1 H) = tr P*(f _I;IH), V f. If we choose (f;1 H) E G i in this special case such that f(j) = IG, V j~i, then

92

(j~i fj)tr Fi(f(i)) = (j~ -~-l(i)fJ)tr

F _l(i)(f(i))

(I)

with the dimensions fJ of the factors Fj of F*. And this is valid for each f(i) E G. Let us consider the associated Brauer characters resp. the traces (if charK=0) i . (I) so that we obtain ~i(f(i))f~-1(i) = ~ (if charK=p)

We are allowed to simplify

~-I(i)(f(i))f i, V

f(i) 6

G.

Since the rows of the character tables and the Brauer characters of irreducible modular representations are linearly independent, we obtain i = ~-1(i) ~ F i N F _i(i) ~ ~ E S(n )

'nSln)
q.e.d.

From this we obtain for the inertia group:

5.10

G~H~. = G*(HQS(n))'

= G~(HQS(n ))

~ollowing Clifford's theory we now have to extend F* to a representation of its inertia group. It is not possible to extend a linear representation of a normal divisor to a linear representation of the inertia group in general. But in our

We can often extend it to a projective representation.

case here, where G*, the normal divisor, is a semidirect factor of the inertia group, we are fortunately able to extend ~* to a

linear representation of G~HF.. How this can be done, Specht has

93

shown (Specht [I ]). We can assume without restriction, that equivalent factors Pi of P* are not only equivalent, but even equal: 5.11 If now Fj ~ Pk ~ Pi (f(i)) = Pk(f(i)), V f(i) ~ G .

5.12

F*Cf;1 H) = (f

lplCfC1))...f

npnCfCn)))

is the matrix representing (f;1H) , then we set for ~ E HnS(n): 5.13 ~*(f;~)
:=

(f~l~ _1 (1)

(f(1))...f~n~ _I (n)

(f(n)))

and it is easy ~o verify, that these matrices form a representation ~* of G~H~.. Since
~

5.14

F*

G* = F* ,

this representation is irreducible. Now let F" be an irreducible K-representation of HnS(n ) (if H'nS!n ) is the inertia factor of P*) and F' according to
5.15 F'Cf; ) :=

the corresponding representation of the inertia group. Multiplying these two representations together, the result 5.16 ~* F'

with the representing matrices 5.17 (~* @ P')(f;~) := ~*(f;~)xF'(f;~)

is an irreducible K-representation of G~J~p,, as can be seen with the aid of Clifford's theory. The most important result is, that the representation induced by

94

5.16 is irreducible:

~ P

:= (~* @ F') ~ G~H is irreducible and every irreducible K-representation of G~H is of this form.

It remains to investigate, which representations F* and P' resp. F" have to run through such that F runs exactly through a complete system of pairwise inequivalent and irreducible K-representations of G~H. Using Clifford's notation, we call two irreducible representations of G~H associated (with respec t t o G * ) , if their restrictions to

G* have an irreducible constituent in common. From Clifford's theory we know, that there is a l-l-correspondence between the classes of associated representations of G~H and the classes of representations of G* which are conjugates with respect to G~H. Two representations F* and F** of G* are conjugates with respect t_.o G~H, if there is an (f;~) E G~H so that 5.19 F* (f;~) N F**

And this correspondence is as follows: the restriction to G* of every element out of a class of associated representations is (up to its multiplicity) just the corresponding class of conjugate representations. Hence it suffices, that in ~ the representation F* runs through a complete system of representatives of the classes of conjugate

95

representations of G*. Moreover Clifford's theory yields, that associated representations differ only in the factor, which is an irreducible representation of the inertia factor. Hence it suffices, that - while P* is fixed - P' runs through a complete system of irreducible K-representations of H~.. Thus we have obtained the following theorem:

5.20 The irreducible K-representation F = (~* @ F') t G~H runs exactly through a complete system of pairwise inequivalent and irreducible K-representations of G~H if F* runs through a complete system of pairwise not conjugate but irreducible K-representations of G*, and, while P* is fixed, F" runs through a complete system of pairwise inequivalent K-representations o f HOS(n ) .

In the special case H = Sn two representations of G* are conjugates if and only if they are of the same type. Hence F runs through a complete system of irreducible K-representations of G~S n if P* runs through a complete system of irreducible K-representations with pairwise different types and P" - while P* is fixed - runs through a complete system of pairwise inequivalent and irreducible K-representatlons of S(n ). Thus the number of irreducible ordinary representations of G~S n is

96

5.21
" '

E p ( n . ) . . . p ( n s) ,
(n) '

if s is the number of conJugacy classes of G, p(m) is the number of partitions of m, p(O) := I, and if the sum is taken over all the types (n) = (nl,...,ns). This agrees with 3.8, the number of conjugacy classes of G~S n. Of course 5.21 yields also the number of irreducible K-representations if charK = p does not divide IG~Snl as long as K is algebraically closed. If charK = p I IG~Snl we have for the number of irreducible K-representations of G~Sn: Z pr(nl)...pr(n t) ,

(n)

if t is the number of p-regular classes of G, pr(m) is the number of p-regular partitions of m (i.e. p doesn't divide the elements of the partition), pr(0) types (n) = (nl,...,nt). As an example we derive the irreducible C-representations of the normalizer of $3S 3 in $6, which is a faithful permutation representation of $3~S 2 (cf. section 3): := I, and the sum is taken over all the

(i)The representations of the basis group S~, their types inertia groups and inertia factors: S~ ~ S3xS 3 has the following irreducible C-representations:

[3]I[32

[32~[2,1]

[32~[132

[2,12~[3]

[2,1]~[2,1]

97

[2,1111113 ]

[1311113]

[1311112,1]

[13111113 ] .

With respect to the arrangement [3], [2,1], [13 ] of the irreducible C-representations of $3, the types of these representations are (2,0,0) (0,1,1) (1,1,0) (1,0,1) (1,0,1) (0,1,1) (1,1,0) (0,0,2) (0,2,0)

Hence a complete system of irreducible C-representations of S3 with pairwise different types is

[[311113], [311112,1], [3111113], [2,1 ]1112,1 ], [2,1111113], [1311111311.

The corresponding inertia groups are:

s3,,~ 2, s~, s~, s3,s 2, s-~, s3~ 2,

the inertia factors:

st,, s.i, s~, s~, s~, s~.

(ii)The representations of S3~$2: The irreducible ordinary representations of S2 are [2] and [12], the only one of SI is [I]. Thus we get for the irreducible c-representations of $3,$2:

[311113] [2], = [311113] , [311113] [12] ' ,

([311112,1] @ [1]'I t $3"S 2 = [3]#[2,1] ~ $3'~ 2 ,,

([3]#[i 3] ~ [1],I I' s3-s 2 = [3111[I 3] f s3-s2

[2,111112,1] [2]' = [ 2 , 1 ] # [ 2 , 1 ]
[2,1]~[2,1] [12] , ,

([2,1]#[I 3] [I]')

~' 83-82 = [2,1]#[13 ] t 83"S 2 ,

98

[13],[1 z] [2], = [ l Z ] , [ 1 3 ]
[13]~[13] @ [12] ' .

Their degrees are 1,1,4,2,4,4,4,1,1

in accordance with

12 + 12 + 42 + 22 + 42 + 42 + 42 + 12 + 1 2 = 72 = IS3~$21.
(iii)Representin6 matrices: As a numerical example we shall evaluate the matrix of (14)(25)(36), the image of

[2,1]~[2,1] @ [12] ' representing

(e;(12)) under the permutation representation of $3~S 2. (To get all the elements of a representation it suffices to evaluate

the matrices representing generating elements. For generators of wreath products Sm%S" see Neumann [I].) First, using the results of section 4, we obtain the matrix ef

[2,1]~[2,1]

representing

(e;1):

[2,1]~[2,1](e;1)
12 45 3 6 12 46 3 5

[::][::]
13 45 13 46 2 6 2 5

12 3

13

45 6

46 5

1 2 (1)) = (f~1~1(1)f~2P2

99

=~

[2,1]II[2,1](e;(12))

= (f11~2(1)f22131(1))

0 01 O0
1

(We have to permute the second and third colllmn.) On account of [ 1 2 ] ( 1 2 )

= (-1)

we obtain therefrom:

r2011#r2,11

r 21,

o
1 0

1
0 0

o x
0 1

[-1]

0 -1 -1 0 0 0
-

This is a detailed description of the example Robinson gave (Robinson [5], 3.515).

Having now given a complete and detailed description of the construction of the irreducible representations of G~H over an algebraically closed field we may now turn to special cases.

The procedure becomes much simpler if G is an abelian group. Special cases of this, namely wreath products of the form Cm~S n resp. Cm~A n of cyclic groups with symmetric resp alternating groups have been considered by Young, Robinson, Osima, Puttaswamaiah and Frame (see Young LI], Robinson [1], Osima [I],[3], Puttaswamaiah [1],[2], Frame [I]), whose results can now be generalized.

If G is abelian, E* is onedimensional such that

5.23

~*(f;~) = F*Cf;1 H) = ~ F i C f C i ) ) l

100

If ~1,...,~iH:HflS(n) l is a complete system of representations of the left ccsets of the inertia factor HQS(n ) of F* in H and F" is an irreducible representation of HnS(n ) we denote (see the notation of Ourtis/Reiner [1 ]) : (g* @ P')(f -1;~il~k )' if ~i -1 ~i ~ k 0

(~* F')Cf -1;~11~Zk ) := ~i

elsewhere .

E S(n )

With this notation and 5.23 we obtain, if G is abelian: F(f;~) = ((~* @ F')(f _1;~[1~k ))

= (F*(f _ ~lu)9"(q1~k ))
~i I if
" 1

{ F"C~[1~k ), if ~ [ I ~ k E HGS(n ) 0 , elsewhere

Since

we have obtained ~.24 If G is abelian we have for the representing matrices of

P=

(~*F')

? Q~: Q)
, matrix

F(f;~) = (F*Cf -1;IH)'~"(ql

i.e. t h a t the [H:S(n)[2
the

submatrices
numerical

of which t h i s
F*(f ~ i

consists

a r e up t o

factors

-I;1H)

101

equal to the submatrices of which the matrix (F" t H)(~) consists.

The evaluation of the matrices of F" $ H has been described for the special case H = Sn in 4.49 in case that there are only two factors, what can easily be generalized. This method describing the construction of the matrices of the irreducible representations of G ~ for any abelian G generalizes the

result~ of Puttaswamaiah for Cm~S n (Puttaswamaiah [2]) to wreath products G~Sn, G abelian, and the results of Frame on C2~S n (Frame [I]).

Let us return to 5.18. This theorem describes, how we can get the irreducible K-representations of G~H from the irreducible K-representations of G and the irreducible K-representations of certain subgroups HNS(n ) of H. Thus the representation theory of Sm~S n can he derived to a large extent with the aid of the representation theory of the symmetric group. This will be shown below, where we shall consider especially the repercussion of this fact on the representation theory of the symmetric group, for example on the theory of the symmetrized outer products [~]Q[~] ordinary representations L ~ of irreducible

and E~] of symmetric groups.

To generalize this theory of symmetrized outer products of symmetric groups we point now to certain irreducible representations

102

of G~H with the aid of which we shall define symmetrized outer products for any representations of any permutation groups such that [a]Q[~] is a special case.

At first we indicate certain irreducible K-representations of G ~ by a special notation: If P* is of type (0,...,O,n,O,...,O), i.e.

if all the factors F i of F* = ~Pi are equivalent to a certain irreducible K-representation of G, say to PJ, then we denote this by

5.25

(FS;P ,,) := ~S~...~F j P,

n (The inertia factor of F* = ~ PJ is H' OSn' = H ', hence P" is an irreducible K-representation of H.) The special case of the representations (a;~) of a subgroup Sm~S n of Smn (cf. 2.33) plays an important role in the ordinary representation theory of the symmetric group. The induced representations (~|~) ~ Smu =: [~]G[~] are the so-called symmetrized outer products of irreducible C-representations of symmetric groups (of. section 6). Thus we get in a natural way the following generalization of this concept (Kerber [4]):

.~Let

G_KSm, H<_S n. Then we can identify G ~

with a subgroup of

Smn (cf. 2.24/2.25), and if PG and FH are any two K-representations of G and H (for any groundfield K), then we call

~QeFH := (FO;FH) t Stun = (~ ~a) ~

~ Stun

the s2mmetrized outer product of FG and FH .

1o3

A rule which is obvious from PH

the foregoing considerations is: If

Z ~k

k indicates the decomposition of PH into irreducible composition factors, then obviously PG PH has the same decomposition as

z
k

CFG kl

Hence in case of complete reducibility the symmetrized outer product multiplication is additive on the right hand side:

5.27 If H is completely reducible over K, then

z
k k

This generalizes a well known rule for the symmetrized outeT product of c-representations of symmetric groups. We would like to consider the characters of G~H for a moment to derive some of their properties which will be of use later on. If the factor F* out of P = (9* @ P') ~ G~H is of type (n) = (n I, .o,nr) , then obviously every element (f;~) whose permutation ~ has a partition P~ which is not a subpartition of the type (n) has zero as character value under F:

~.28 If the factor P* out of P = (9* @ F') ~ G ~

is of type (n) =

(nl,...,nr) , then all the elements (f;~) have 0 as character value under P whose conJugacy class has an empty intersection with S(n .

104

Nor under this assumption not any one of the conjugates Ce;~i)-ICf;~lCe;~i) -- (f i I ; ~ I ~ i ) of (f;~) is contained in the inertia group from which P is induced, since for e~ery i: ~ i 1 ~ i ~ HGS(n ) . Thus we have only O-matrices

along the leading diagonal of N(f;~). q.e.d. The difficulty of getting the characters of wreath products explicitly, except for special cases, arises from the fact that to pass from N*(f;IH) to ~*(f;~) we have to permute the columns of N*(f;IH) so that the leading diagonal will be disturbed. Nor a discussion of the evaluation of these characters the reader is referred to Littlewood [3]. A useful remark is:

5.29 If ~ is an n-cycle then the character of (f;~) vanishes under all the irreducible representations of G~H which are not of the form (PJ;P"). Under (PJ;P") the element (e;~) has the character value

C(pJ;F,,)(e;~)= ~N"(~)
if fF~ is the dimension of N j.

Proof: The first statement follows from 5.28. Furthermore we know that (FJ;F")(e;~) = E ( n x P"(~) .

And in this first factor on the right hand side, which arises

105

from the (f~J)n-rowed~ identity matrix by certain collm~ermutations the leading diagonal contains,except O's~as many 1's as is the dimension of PJ, for exactly the elements fii...fii (1~i~f Fj) remain in this leading diagonal (cf. 5.13). q.e.d.

pJ

FJ

We would like now to derive some results concerning the modular representation theory of wreath products (cf. Kerber [21) which will be of use later on in applying this theory to the representation theory of the symmetric group. First we consider the case where G is a p-group. Then G ~ conmodular

tains with its basis group a normal divisor which is a p-group as well as its centralizer if G $ ~I~. For if (f;IH) E G i, f $ e, ~ $ I, we have I $ (e;~)(f;IH)(e;~) -1 = (f~;1 H) E G -I Hence the following is valid:

(i)

$ Gi -

5.3O

G ~ [I~

CQ~(G*)

~ G*.

Thus we can apply a lemma of Brauer (Brauer [11, lemma 2) which says, that a group possesses exactly one p-block if it contains a normal p-group which includes its own centralizer. We have obtained: 5.31 If G $ ~I] is a p-group, then G~H possesses exactly one p-block.

106

Furthermore

in this case the only irreducible p-modular represenso that each p-modu-

tation of G* is the identity representation lar irreducible representation 5.32

of G~H is of the form in H' so

IG* @ F' = P'

Since G~H = G*H', the p-regular elements are contained that an ordinary irreducible representation to H':

of G~H has the same

Brauer character as its restriction

If G is a p-group,

then a p-modular representation

associated has the

with an ordinary irreducible representation P of G ~ same decompositiQn numbers as F $ H'.

In case that G is an abelian p-group, sional so that ~*(f;~) and the associated presentation.
=

the factor F* is onedimen-

F*(f;1H) is the identity re-

p-modular representation

Hence the following is valid:

5~

If G is an abelian p-group,

then a p-modular representation representation numbers as F =

associated with the ordinary irreducible

(~* @ F') ~ G~H has the same decomposition p-modular representation associated

with F" ~ H.

As

an example we would like to evaluate the decomposition matrix

of $2~S 3 with respect to p=2.

~)

O O
O ~ C/~ O ~ O

=-~

~ 13

--~ 0 0 ~ ~1 II "~ d" ~O "4 el" HO

--~

--~

-~

-~

-~

"~

-~

-~

"~

H" O ~ c+

O D (I)

el" I:~
! --~ ~ ~

I-b I-"
cf

~ !
O 0

..-.,. ~ ~ ~ ~ r--, .,,, ID" ~ h) .~ ~ ~ "e~ ~

-" ~ I

0" ,~

O ,d'c-I-

O I1::1

,~~

v ~ ~-~ d0 ~ ~.~ h) ,~ ~ 0 ~ -* Z r~ ,.,,1 II 0 --', I',) c+ H. 0 II I::1 ~:~ 0 c+

6'--"I f--~

I'-'-I

--~

~-'-~ |

Ct~

IX)

h)

el-

-,~

et-

m
~ 0

t't
!

108

If on the other hand the order IGI of G is relatively prime to p, the situation is quite different. For then every p-modular representation associated to ~* is irreducible since ~* ~ G* = F* is irreducible and we obtain with a lemma of 0sima (Osima [3], lemma 5):

5.56 If (IGI,p) = 1 the p-block of GaS to which F=(~* F')$GaS

belongs has the same decomposition matrix as the block of the inertia factor to which F" belongs. (~* P~) t GaS and (~* @ F~) ~ GaS belong to the same block of GaS if and only if F~ and P~ belong to the same block of the inertia factor H'GS(n ) of P*.

Using this and the well known decomposition matrix

E2, ]
[1 3 ]

of S 3 with respect to p = 3 we obtain as decomposition matrix of 82%S 3 for p = 3 and the same arrangement as in 5.35 of the rows:
1 1 0 0 1 1 1

5.37

1
1 1 1 0 1

109

It is more difficult to evaluate the decomposition matrix if neither G is an abelian p-group nor (IGI,p) = I. But since 5.18 is independent of the characteristic we can using this fact. As sometimes proceed

an example we evaluate the decomposition ma-

trix of S3%S 2 for p = 2. [3] and [2,ii are the 2-modular irreducible representations of S 3, [2] is the only one of 8 2 . Hence

[3]~[3] [2]' ,
([3]@[2,1] ( [ 1 ] @ [ 1 1 / ' / t 83%82 = ['3"]~[2,i']" '~ S3"vS2 ,

[2,111112,1]

[2],

are the 2-modular irreducible representations of S3%S 2. The table of Brauer characters of this group is therefore

(1,1;11

((1231,1;11
1

((1231,(1231;11
1

[311113] [2],
V

[2,1]II[2,1] [2]'

4 4

-2 1

1 -2

[3]~[2,1]
Prom this

t $3'~ 2

it follows that

(3;2)

"1
1 1

0
0 0

0
0 0

(3;12 )
(13;1 2 ) (13;2)

(2,1;21
5.38 (2,1;12 )

[311112,1] 1' s3%s2

[3]tt[13] 1' S3%82 [2,11ti[ 13 ] 1" 83%82

1 0 0 0 2 0

0 1 1 0 0 0

0 0 0 1 0 1

110

is the decomposition matrix of $3~S 2 for p = 2 (of. also the character table of $3~S 2 in section 6).

Concluding this section we point once mere to the of the irreducible representations of G~H.

construction

If we know how to construct the irreducible representations of G and of the subgroups HnS(n ) of H we get the representing matrices P(f;~) as we have described above: First we form Kronecker

products of irreducible K-representations of G (see 5.1/5.2), then we have to permute the columns of these matrices (see 5.13). After this we have to form Kronecker products once more (see 5.17) and at last we have to induce (see 5.18) to get P(f;~). Hence the entries of P(f;~) have all the properties which are properties of the entries of the irreducible representations of G and HOS(n ) and which are invariant with respect to col~imn permutation and the Kroneoker product multiplication as well as the inducing process. Prom this consideration we get some results about splitting fields (of. Kerber [7]):

5.39 If S c K (K algebraically closed) is a splitting field for G and the subgroups HNS(nl, then S is a splitting field for G~H, too, i.e. every irreducible K-representation of G~H is realizable in S.

111

E. g. if all the ordinary irreducible representations of G and of the subgroups HOS(n ) are realizable in ~ (resp. Q), then all the ordinary irreduzible representation of G~H are realizable in R (resp. Q) as well. Thus for example all the ordinary irreducible representations of Sm~S n are realizable in Q and hence even in N. A weaker assumption would be that the ordinary characters of G as well as of the subgroups HnS(n ) are real. But it is possible, that this is not carried over to G~H since the leading diagonal is disturbed when passing from F*(f;IH) to ~*(f;E). We know that the reality of the ordinary characters is equivalent to the ambivalency of the group. And as we have proved in section 3 (cf. 3.14) G~S n is ambivalent if G is ambivalent. Thus we have,though this does not follow from the construction of the matrices (Kerber [7]~:

5.40 If the characters of G are real, the characters of G~S n are real as well.

From 3.16 we get:

5.41 If the (ordinary) character table of G or of H is not real, then the character table of G ~ is complex.

A finite group is called an M - _ ~ ,

if every irreducible repre-

sentation over an algebraically closed field whose characteristic doesn't divide IS l is induced by a onedimensional representation

112

of a suitable subgroup. Another corollary to 5.18 concerning this conception is (Seitz [1], Kerber [61):

5.42 If G and the subgroups HGS(n ) are M-groups, then G~H is an M-group.

Proof: If K is algebraically closed and charK ~ IGI, then every irreducible K-representation of G and hence also of G* is equivalent to a representation, whose matrices contain in every row and in every column exactly one nonvanishing entry. This is valid for F* as well as for ~" and this is a property invariant under column permutation, Kronecker product multiplication and the inducing process. Hence it is a property of F* and F" and ~* F' and F = (~* P') t G~H. From a well known theorem (see Huppert [I], V, 18.9) it follows, that since E is irreducible, ~ is induced by a onedimensional representation of a suitable subgroup. q.e.d. E.C. Dade has proved this for the special case H = Cp = <(1...p)> Sp (p a prime number, see Huppert [I], V, 18.10) as an important part of his proof, that every solvable group can be imbedded in sa M-group (see Huppert [1], V, 18.11). The last 4 theorems are corollaries of the construction of the irreducible representations of G~H we have given above. Another theorem concerning generalized decomposition numbers will be given

113

in

section

8.

With the aid of the results of this section we would like now t o desoribe the application of this theory of wreath products %o the representation theory of symmetric and alternating groups.

Chapter III

Application to the representation theory

of symmetric and alternating groups

Two examples, the theory of the symmetrized outer products and the theory of the generalized decomposition numbers will show how the representation theory of wreath products can be applied to the representation theory of symmetric and alternating groups. More precisely we apply the representation theory of wreath products of the form G%S n. In the first case the used subgroups are of the form Sm%Sn, in the second case of the form Cm~S n. Although the representation theory of G%S n has been derived with the aid of the representation theory of the symmetric group,our argument is not circular since for the applied subgroups G~S m ~ Sn we have m<n. Hence this application is actually a recursion pro-

115

tess. Thus for eTample the evaluation of the (strictly) generalized decomposition numbers of Sn is reduced to the evaluation of decomposition numbers of symmetric groups of lower degrees m<n.

There are probably other ways of applying this theory of representations of wreath products to the theory of the symmetric group. Presumably we can illuminate in this way the concept of the socalled "star-diagram" or "p-quotient" of a Young-diagram satisfactorily, but this will be discussed in the following parts of this paper.

The first section of this chapter contains the theory of the symmetrized outer products of irreducible ordinary representations of symmetric groups. Then as a preparation for the third section which contains the theory of the generalized decomposition matrix of symmetric and alternating groups we s1~mmarize in the second section some known and some new results about decomposition numbers of symmetric and alternating groups.

116

6. Symmetrized

outer products of irreducible

C-representations

of symmetric groups

A chance to apply the representation the ordinary representation

theory of wreath products to

theory of the symmetric group arises n from the trivial fact, that the normalizer NSmn( Sm) of a direct n
product Sm := S m . . . S m ( n f a c t o r s ) n of n subgroups isomorphic to

Sm and in Smn lies between x Sm and Smn: 6.1 n n X Sm _< gSm n(x S m) -< Smn

Por as we have seen in section 2 (of. 2.33) this normalizer is a

n

faithful p e r m u t a t i o n basis group.

representation

o f Sm'~ n w i t h

x Sm a s

its

Hence the problem to derive the reduction of the representation n [a]...[~] = [~]~...~[~] t Smn induced from Sm can be divided into two problems. For using the transitivity of the inducing process and 6.1 we obtain: n

6.2

Sm(X

Sm)

The first one of the two remaining problems is to derive the reduction of n

6.3

[~3

~...~

[~2 f NSmn(X S m)

The second one is the problem to give the reduction of the repre-

117

sentations of Smn induced by the irreducible constituents of 6.3. Let us first consider the reduction of 6.3. n We know that NSm n(x Sm) is a faithful permutation representation of Sm~Sn, hence the problem is to give the reduction of n 6.4 ~ [~] $ Sm~S n But with the reciprocity theorem of Probenius and 5.18 we have at once the solution of this problem:
n

6.5

# [~] f Sm~.S = P. fP(e;~), n

if the sum is taken over all the partitions of n and if fP denotes the dimension of [p]. Applied to our starting problem we obtain from 6.5: n

6.6

~ [=] f Stun = z fPC(~;P) f Stun) , P

Thus we have reduced this problem to the reduction of the t rized outer prqducts (cf. 5.26)

6.7

[~][~]

: : (~;P)

t Smn

Gathering up we obtain the equation

6~s .

[~2...[~]

: z fP([~][p]) p

which originally directed the attention to certain in general reducible representations of Smn which were called symmetrized outer products. A hint to clarify their theory with the aid of the representation theory of wreath products we owe to Robinson

118

who pointed out that they are induced by certain irreducible ren presentations of Ngm n(x S m) (see Robinson [3],[4],[5], Kerber [4]).

To derive results of the theory of symmetrized outer products it is useful to describe the ordinary representation theory of Sm~S n in more detail. This we shall do now using the results of the sections 4 and 5. But we shall not only consider the irreducible representations of the form (~|~).

Prom the results of the sections 4 and 5 we obtain, that all the irreducible C-representations of Sm~S n are of the form $ Sm%S n

6.9

P = (( ~ _ [~]j) (P~m)[~]k)) '

j=1
with irreducible C-representations

k=1
[~]j of Sm and [~]k of Snk,

if ~[~]j is of type (nl,...,np(m)). The number of irreducible C-representations of Sm%S n is (cf.5.21):

f~tQ

(n)

P(nl)...p(np(m))

, (n) = (nl,...np(m))

if the sum is taken over all the p(m)-tupels so that O~n i E S, Z n i = n.

We have given an example for the evaluation of the representing matrices in the last section. p(m)p(n) of these representations 6.9 are of the special form

C~).
The next question is, how we can evaluate the character table of

119

Sm~S n. Of course we can obtain the characters by evaluating the matrices for a complete set of representatives of the conjugacy classes under each irreducible ordinary representation and checking the traces of these matrices. But this is the most complicated way to get the character table and it c a ~ e simplified

very much by evaluating the matrices only for certain representations and using symmetries of these tables. These symmetries are of the same kind as these of the character table of Sn which arise from the fact described by 4.14 from what follows that we get the row of [a'] from the row of [~] by changing the sign in the columns of conjugacy classes which belong to Sn~A n. The reason for these symmetries of the character table of Sn is the fact that [a] and [a'] form a pair of irreducible C-representations of Sn which are associated (in the sense of Clifford's theory of representations of groups with normal subgroups) with respect to t~e normal divisor A n of index 2. This can be applied also to Sm~Sn, even with more success: The n normalizer NSmn( Sm), a faithful permutation representation of S m ~ n ~ C O n t a i n s for m,n>1 two different and nontrivial normal subgroups of index 2, the subgroup 6.11 SmaSh + := Sm~S n n Amn

of the even permutations, and the subgroup

120

6.12

Sm~A n .

(Por the sake of simplicity we write Sm%Sn + instead of (Sm%Sn)+, hoping that this subgroup will not be confused with Sm%S ~ = Sm%An.) Set us denote by F + the representation associated with F with respect to Sm~Sn+ and by FA the representation associated with F with respect to Sm%A n. Thus F + (resp. pA) means the inner tensor product of P and the alternating representation of Sm%S n with respect to Sm%Sn+ (resp. Sm~An). If we denote these alternating representations by A+Sm%S n resp. A-S m %S n we have obviously A

A+Sm~'gn = { (lm;n), i f m is even

6,1~ (lm;ln),if m is odd

AASm%S n =

(m;1n) .

Thus P+ = f P @ (Im;n) 6.14 , if 2 I m

tF

(lm;In), if 2 t m ,

FA = P @ (m|1 n) . To describe F + and FA explicitly in the form 6.9 we take into aceount that for inner tensor products the following is valid if H ~ G, D I is a representation of H, D 2 a representation of G:
(D 1 f G) @ D 2 = (D 1 @ (D 2 ~ H)) f G .

Using this it is easy to verify the following theorem (Kerber [4]):

The irreducible C-representations

of Sm~SnaSSociated with the

representation 6.9 with respect to Sm~Sn + resp. Sm~A n are:

121

~+ :=

I((uK~']J)~ (~[~]k)') t Sm~Sn , if 2 1 m

(C~[~']j) @ (~[~']k)') t Sm~S n, if 2 t m

resp. pA := ((~[~]j)
, J

(#liB,]k),)

1~ Sm~S n ,
. .

if [a']j resp. [~']k denotes the representation associated to [aSj resp. [~]k with respect to Am resp. An~r, i.e. if [~'Sj := [~]j @ [ImS ' [~'Sk := [~]k lInkS " Special cases are

{C~';~) , if 2 I m
(~|~)+ = C~';~'), if 2 ~ m ,

(a;~)A = (~;~,) .
This indicates how we can get the rows of P+ and pA from the row of F by changing the sign in certain collnmns. P is called selfassociated with respect to SmaSh + (Sm~A n) if F = F + (P = FA). The character of such a selfassociated representation vanishes outside of Sm~Sn + (Sm~An). Finally we mention special character values which can be obtained by specializing 5.28 and 5.29: 6.16 If in 6.9 the representation ~[~]~ of the basis group is of type (nl,...,np(m)) and if the conjugacy class._ of ~ in Sn has an empty intersection with S(n), then we have for the

122

~P of F: ~F(f~) : O, V f.

If ~ is an n-cycle, then ~F(f;~) = 0 if F is not of the form (a;~), and (-1)kf ~, if [~] = [n-k,1 k] (O<_k<n)

~=~(e;~)
0 elsewhere (For the last statement of. 4.42) As an example we give the character table of $3~S 2. The distri-

bution of the elements of S3%S 2 into conjugacy classes we gave in section 3. The ordinary irreducible representations of this group have been described in section 5, it remains to evaluate the characters. The first row of this table on page 123 is the trivial row of the identity representation (3;2) = [3]#[3] [2]'. The second and third row can be obtained from the first row according to 6.15 and the fourth row can be obtained from the second one. We evaluate the matrices representing elements of the conjugacy classes under (2,1;1) and get therfrom the fifth row and from this the sixth row according %o 6.15. Similarly we obtain the seventh row from which we get the ninth one. The eigth row belongs to a representation which is selfassociated with respect to $3~$2 + as well as S3%A 2 such that for this row there are only 5 matrices to

123

DO

DO

1"..3

.-~

'.,J,I

U
e_Jo

r~

~1

h) II

~1

.--IX)
+

II

~-~

II

,'-v

II

O~
o "4

m
U ,=-it~

I'0 v +

t~
m

4
o

I'0

(1,1;1)
I
0 1~ 0 0

I
-"~ "~

I
"-~ -"~ v

((12),1;1)

ro

I'O

f*o

--',

-.x

.-~

i,,3 v

((123),1;I)

t.,o

I
0 ro 0 0 0 --' -'~ ~" "-' JD

((12),

(12) ;4)

Po v 4~ 'kTt v

((123),(123);1)

v
0 --~ 0 0 .-~ ..L 4~ v

((12),(123);1)

'
0 0 0 Ix) DO "~

'

4~ v DO

(1,1;(12))

I
0 0 0 ~ ~ ~

I
-" ~

I
4~ C~ v

((123),(123);(12))

k.lrt 0 0 0 0 0

'

'~

._,

((12),(123);(12))
,#,.

124.

be constructed. In the same way we get the zeros of the second, fourth and sixth column and the fifth and sixth row. The box containing zeros in the right and lower corner we get from 6.16 (cf. Kerber E4]). This table has also been evaluated by Littlewood (Littlewood [2],

p. 275) with the aid of the table of S 6, while for this procedure to get the table of $3~S 2 (in the general case: Sm~S n) we used only the representations of S 3 and S 2 (resp. Sm and Sn and sub-

groups of Sn in the general case) and not the table of S 6 (resp. Smn ). In Littlewood's book we can find also the tables of $4~$2 (p. 277), $2~S 4 (p. 278) and $3~S 3 (p. 280). son (Robinson In a paper of Robin-

E4]) the table of $2~S 3 can be found, he also did

not use the table of S 6. In a later paper Littlewood used this more direct method, too (Littlewood L3]).

Having derived these detailed results on the ordinary representation theory of Sm~S n we return to the theory of symmetrized outer products [~](~)[~] of irreducible ordinary representations

of symmetric groups. A problem which is up to now only incompletely duction of [ a ] Q [ ~ ] Boerner solved is the re[5] and

(see the references given in Robinson

[3]). But using our theorems we can easily obtain the

most important results of this theory. At first we get at once from 6.15 for the multiplicity

125

([aS(~[p],[y])

of the irreducible representation

[y] of Smn in

[ ([,~'20[~1,[~,'3) ([~][p],[])

, if

2 I

(E,~'lO[~'],r~,'-I),

if 2 1 m ,

the so-called T, ittlewood's theorem

of con~u~ates (Littlewood E1 ]).

Another theorem, perhaps the most important one can also be obtained easily. With its help our problem can be reduced to the problem to give the reduction of symmetrized outer products of the special form

[~](~[r]

(r_<n). The assertion is analogous to 4.41.

If n = n1+n 2 we denote at first

6.18

([c~]Q[nl])([a:]Q[n2])

:= ( ( [ ~ ] O [ n l ] ) # ( [ ~ ] O [ n 2 ] ) ) 1 "

Bmn

and prove that the following is valid:

6.19

([~2Q[n12)([~20[n22)

= [~2(~([nll[n22)

Proof: Using a well known theorem about induced representations (see Curtis/Reiner 512,(43.2)) and the transitivity of the inducing process we obtain for the left hand side, using 6.18: ((a;nl)~(~;n2)) ~ 8m~(SnlXBn2) ~ Sm~8 n t Smn Because of (Sm~Snl)X(Sm~Sn2) = Sm~(SnlXSn2 ) this is equal to

(~;[nl]tl[n2])

'~ Sm',,,S '~ Smn = [o~]Q([nl][n2]) n

q.e.d.

Applying 4.41 to [P2:

126

[~]

= l[~+j-i]l

we obtain the announced result (Robinson [5], 3.531):

[~]O[~] = l[~]O[~i+J-i]l
For example

[~][2]

[a][3]
[~]~)[I]

[]012,1]

=
1

= ([~]O[2])([~][1]) = ([~][2])[a] [~]Q[~i+J-i]

- [~]013]
Ana i~ we know the

- [~]O[3] ~n.

is o~ the ~orm [ ~ ] [ r ] ,

reauctions o~ these [~]Q[~i+J-i] we get the reaucton o~ [~]O[~]

with 6.20 and the Littlewood-Richardson-rule 4.51. It is remarkable that with the reciprocity theorem of Frobenius we obtain

6.21

([o:]C)[r],[y])

= ([]

~ Sm~Sr,@[a])

which reduces the problem to the evaluation of the character of @[~] (cf. Kerber [4]).

Concluding this section we would like to derive two equations concerning the matrix 6.22 of the multiplicities 6.23
r ~

for a fixed partition ~ of m and for the partitions ~ of n, y of

127

mn. Thus R a is a p(mm)xp(n)-matrix. Let [6] be an irreducible representation of Sn_ I and 6.24 r~ := ([~]~[6],[E])

Using the Littlewood-Richardson-rule 4.51 we get at once the matrices S ~ := ( s ~ ) 6.25 and 6.26 e~6 -- ([6][i],[~])
.

and E := (e~&), whose entries are defined by

s~

yE

::

([E][=],[~])

A special c~se of 6.19 is

6.27

([=]016])[=]

-- [=]([6][i])

We can compare the multiplicities of the irreducible constituents on both sides of 6.27. For the left hand side we obtain from 6.24 and 6.25: 6.28 ([~]Q[6])[=] = ~ ra~ [a] = *
to ~y

n svErE6[V] -

And since 6.26 is valid the right hand side of 6.27 is

6.29

[=]Q([~][1])

=ne~[=]Q[~]

= ~,r~e~6[~]
~,~

Comparing 6.28 and 6.29 we obtain"

6.50

R~

= S~R a*

(Robinson [5], 3.548) as a necessary condition for the wanted matrix R m. A second condition for R ~ can be derived using the theorem 6.16 concerning the characters of the irreducible representations of

128

Sm~S n We

consider the equation 6.23. Using the reciprocity theorem of

Frobenius we obtain
6.3 Sm,S n = + ... ,

if we gather under the snmmation sign exactly those irreducible constituents of [y] ~ SmaSh, which are of the form (~;~). Since 6.16 it suffices to regard only these constituents of the form (~;~) if we take the character of (e;(1...n)) on both sides of this equation. We obtain C(n m) = ~ r ~ ; ~ ( e ; ( 1 . . . n ) ) . (The partition of (e;(1...n)) is P(e;(1...n)) = (n,...,n) =(nm).). Using 6.16 we get therefrom: 6.~2 C (nm) = ~,~:#=(n-k,1 k ) r~6 f~(-1)k

~ as a second necessary condition for the coefficients r ~ of the wanted matrix R ~ (see Kerber [4]). At the end of this section we give three of the most important rules for these symmetrized outer products. They are well known but now we can obtain these rules much more directly with the aid of the representation theory of wreath products. A generalization of 6.19 is

([~]O[~])([~]Q[~])

= [~]o([~][~])

the proof is quite analogous to the proof of 6.19.

129

A special case of 5.27 is

6.34

[~]([~]

+ [D

= [~][~]

[,,]0[~,1

and from the associativity of the wreath product multiplication we obtain

(Ea]C~l)

1"~,3 --- [ a ] ( C # ' l [ ~ , l ) .

130

7. Block-structure and decomposition numbers

of symmetric and alternating groups

I -, '"

illil

The development of the modular representation theory of the symmetric group began in 1940 with the publication of the two parts of T. Nakayama's paper "On some modular properties of irreducible representations of symmetric groups" (Nakayama [I],[2]). The second part of this paper concludes with a conjecture about the p-blockstructure of the symmetric group which has been proved first by Brauer and Robinson in 1947 (Brauer [I], Robinson [2]) and which is the foundation for all the following papers concerning this theory. Beyond this fundamental theorem which is still called "Nakayama's conjecture" we know a lot of results but not the wanted general results (i.e. independent of n and p) about the decomposition numbers of Sn with respect to p. Purthermore we know how the evaluation of the generalized decomposition numbers Can be reduced to the evaluation of decomposition numbers of symmetric groups of lower degrees and we know some analogous theorems about the alternating group, e.g. we know the block-structure of An and how we can get the decomposition matrix of An as far as the decomposition matrix of Sn is known.

131

We would like to describe some of these newer results beginning with the theorem about cribe the present the block-structure of An We shall des-

situation of the theory of decomposition num-

bers of symmetric and alternating groups To formulate Nakayama's conjecture we need the following definition:

7.1 Def.: Let [a~ be a Young-diagram and q a natural number. If we cancel successively parts of the rim which belong to hooks of length q, then a (uniquely determined) subdiagram [~] remains which we call the q-core of [u].

Por example if q = 3:

//
-.

-.

,r

I"

[32,2,1]

= [01

po p

pl j

/ If p is a prime number, then the fundamental theorem of the modular representation theory of the s.ymmetric group reads as follows:

~2

("Naka2ama's conjecture,') The irreducible ordinary representations of Sn which form the p-block of Sn to which ra~ belongs are exactly the representations E ~ with
-

of Sn with the same p-core as [~], i.e. the r~]

132

As has been said above: this is no longer a conjecture, theorem has been proved in 1947 by Brauer and Robinson! To prove 7.2 the well known characterization the class multipliers (see Curtis/Reiner [I],

since this

of the p-blocks by 85,86) can be used:

[~] and [~] belong to the same p-block if and only if their characte~ a, ~ satisfy the following congruence modulo p:
:=

73

(Loal/f )

(ic iIf )

=:

mod p,

on all the conjugacy classes Ca of Sn. An important role in the proof is played by the defect of a block. It turns out, that the defect d~ of the block containing [~] and

the number b~ of p-hooks which can be removed from [~] to yield [~] satisfy the equation d~ : b~ + ep(b~!) if as in section I ep(m) denotes the exponent of the maximal power of p which divides m. b ~ is called the p-weight of [a]. Actually the following is valid (Brauer [1]): ~.5 The defect group of the block of [~] is isomorphic to a p

p-Sylow-subgroup__ of Cp~Sba -< Spb~, i.e. isomorphic to the wreath product Cp~Pb~, if Pba denotes a p-Sylow-subgroup of

Sb~"
This provides an opportunity to apply the theory of representations

133

of wreath products. Actually some parts of the known proofs of Nakayama's conjecture can be simplified in this way. But this remark may suffice here, a more detailed discussion is left to the later parts of this paper, since we shall have a similar application to the theory of generalized decomposition numbers which will be discussed in full detail.

Before doing this we would like to consider the alternating group. The theorem concerning alternating groups and analogous to Nakayama's conjecture is (Puttaswamaiah [1], Kerber [3]):

[1],

Puttaswamaiah/Robinson

7,6 (i) If [a] = [~], then every irreducible constituent of [ulnA n forms its own p-block and each modular representation associated with such a constituent is irreducible. (ii)If [~] ~ [~], then to the p-block of an irreducible constituent of [~] $ A n belong exactly the irreducible constituents of restrictions [~] $ An of such representations [~] of Sn for which

Proof: We shall repeatedly use the theorem (of. Curtis/Reiner (86.3)) that an ordinary irreducible representation forms its

[1],

134

own p-block and is modular irreducible if p is contained in its dimension as often as in the order of the group.

(i)

We assume first, that [ ~ ] = [ ~ ] .

a) I f

~ $ a',

then ( c f .

#.54)

[ ~ ] ~ An i s

irreducible.

[a] = [~] implies, that [a] contains no hook of length p, hence in this case (cf. 4.46): ep(f [a]~An) = ep(f u) = ep(n!) ~ ep(n!/2) = ep(JAnJ), if n>1. Using the theorem mentioned above, this part of the statement

i s proved f o r

the case [ ~ ] = [ ~ ] ~ [ a ' ] .

b) If on the contrary ~ = a', then by 4.54 the restriction decomposes:

[a] ~

An

[=]+ + [a]-,

and [a]~ are irreducible ordinary representations of dimension fa/2. [a] = [~] implies again ep(n!) = ep(fa). Thus

e p ( f ~-+) = e p ( f a / 2 ) and the s t a t e m e n t (ii)We i s proved f o r [~3 + [ ~ ] .

= ep(n!/2)

= ep(lAnl),

the case [ i f ] = [ ~ ] = [ ~ ' ] .

assume now t h a t

a) At first we would like to show that under the additional assumption p ~ 2 the restriction [a] ~ An cannot contain an irreducible constituent which belongs to a block of defect 0. If f aA denotes the dimension of such a supposed constituent of a block of defect 0, we have ep(f aA) = ep(n!/2) = ep(n!) ,

135

the last equation is valid since we have assumed p + 2. Because of f aA = fa or =

fa/2 this implies

,

epCn!) = ep(f aA) = ep(f a) in contradiction to Ca] ~ [~].

b) If p = 2 we have for the dimension of a constituent

[a]A of

Ca] $ A n which belongs to a block of defect 0 and which is therefore modular irreducible:

e2(f aA) = e2(n!/2 ) = s2(nX ) - 1 . With e2(f a) ~ e2(f

aA

(1)

) we obtain
e2(f a) = e2(n! ) -I (2)

since [ a] + [ ~ ] . Comparing (I) and (2) we get

e2Cf aA) = e2(f a) ,

such that because of f aA =

f~ or = fa/2 we obtain

[a] ~ A n = [a]A Since

on

accoumt of (2) Ca] contains

exactly one 2-hook,Ca] be-

longs to a 2-block corresponding 2-cores are selfassociated: for every 2-core

= Col

to a 2-core with n-2 nodes. for every 2-core since

C~] = [ ~ ]

[~] we have
or = lj

Hence the diagram Ca] containing exactly one 2-hook is of one of the two following forms:

Ca] = [ ~ 1 . 2 , ~ 1 - 1 , . . . , 2 , 1 J

or Ca] =

1,a1-1,...,2,13]

136

These two representations

are associated with respect to An, and

their restrictions to An are equal. Hence only one of them yields a row of the decomposition matrix of A n . Thus [~]A forms its own block and this block consists actually of the constituents of restrictions [~3 ~ An with [~] = [~] or [~3 = [~']

since the last consideration shows that there is up to equivalence only one such constituent: [m]A = [~I+2'$I-I'''''2'I] $ An = KSI'SI -I'''''2'13] $ An "

Thus the statement is proved in case that [~] $ E~] and [~]~A n contains a constituent of a block of defect 0 and if [~] $ K~] this is only the case if p = 2. c) Now let [~]A be an irreducible constituent of [~] $ A n so that
=

or

a) and b) imply that we can assume that [~]A and the constituents [~]A of [~3 ~ An belong to blocks with defects >0. We have to show that [~]A and [~]A belong to the same block. To prove this we use the class multipliers. Since [~] belongs to the Sn-block of [~] or of [~'3 we have (Curtis/Reiner Is], (85.12)):

for all the p-regular Sn-classes Ca which satisfy Ca ~ A nWe would like to show the validity of the analogous congruences for ~ ~A and ~ A , this would complete the proof of this part of

137

our statement (see Curtis/Reiner The theorem of Probenius

[I],(86.19)).

(4.55) implies that ~

=A

resp.

~A

agree with Ca resp. C ~ or with ~C ~ resp. C~ on non-splitting classes of S n. The same is valid for the dimensions, hence for such non-splitting classes ca_cAn we have also ~A 8A Wa ~ a

(P) "

The elements of splitting classes consist of cycles of pairwise different lengths (see 1.23). Hence the order of the centrali-

zer of a permutation which belongs to a splitting class is the product of its cycle lengths (see 2.32). If such a permutation is a p-regular one, p does not divide any one of these cycle lengths and therefore p doesn't divide the order of the centralizer such that the defect of such a p-regular splitting class is O. Hence from a well known theorem (Curtis/Reiner (86.27)) we get: ~A ~A a ~ 0 m a [I],

(P)

for all the p-regular splitting classes, since [~]A and [~]A belong to blocks of defects >0. Thus for all the p-regular An-classes Ca we have

~A ~A a ~ ~a

(P) '

hence E~SA and [~JA belong to the same block. d) It remains to show, that the block of [~]A does not contain a constituent [~]A of a representation [~] ~ A n with [~] + [~]

138

and

+
belonging to a certain

The ordinary irreducible representations

block B can be characterized as follows: two ordinary irreducible representations belong to B if and only if there exists beginning with

a chain of ordinary irreducible representations one of the two representations

and ending with the other one associated to any two

and so that modular representations

neighbours of the chain have an irreducible constituent in

c ommon.

Hence it suffices to show that modular representations and [8]A associated with irreducible constituents

[~]A [~]A

[~]Aand

of [~] 4 A n and [8] $ A n such that [~] + [S] and [~] + [S'] have no irreducible constituent in common. Prom Clifford's theory of representations of groups with

normal divisors we get that an irreducible modular constituent P~ of [~] ~ A n is a constituent of the restriction P~ ~ A n of a modular irreducible representation F~ belonging either to the p-block of Sn with p-core [~] or to the associated p-block with p-core IS']. But [8] is not contained in one of these two blocks of Sn. If now ~ dikFk [~] k describes the modular decomposition of [~], then [~] ~ A n has

139

the same decomposition as the restriction of the modular representation on the right hand side:

k Thus the modular decomposition of this right hand side contains the decomposition of [~]A. But since [U]A is contained in quite another block, this right hand side and hence the decomposition
i i ,i i

An =

dik(

An)

of [~]A' too, cannot contain a modular representation in common with [~]A" q.e.d. 7.2 and 7.6 enable us to evaluate the distribution of the ordinary irreducible representations of Sn and An into p-blocks in a very simple way and for every n and p. But the problem of finding the decomposition numbers is still far from a satisfactory solution. We would like to describe this now. We shall treat the case p = 2 and report upon the known results. Thereafter we shall discuss what can be said about the decompo~ sition matrix of A n if the decomposition matrix of Sn is known. In certain cases, for example if n~2p, it is easy to get the decomposition numbers of Sn (see Robinson [5], P. 122). But for the general case, there is only the method of explicitly reducing the representing matrices which is possible without a vast amount of calculations only for very small dimensions (see Robinson [5])The use of an induction process seems to be much more promising.

140

As we soon shall see such a process together with some results on the decomposition numbers of special representations allows the

evaluation of the decomposition numbers of Sn with respect to p=2 up to n = 9. We would like to show how this can be done.

Per a fixed prime number p let D I := I Dn, p = (d~k)

denote the decomposition matrix of Sn with respect to p. (The upper index I will be shown to be of use in the following section where the generalized decomposition matrix will be considered whose first col,,m~s are built up by DI.) If the ordinary irreducible representation the i-th row of D I i.e. if [~]i ~ Z dik k
[a] i of sn belongs to

F~ ,

the modular decomposition of Z d~k [a] i is the same as the decomi position of the k-th principal indeeomposable U k of Sn: Z dik [ ]i " i The induced representation U k ~ Sn+ I is a direct sum of principal I of Sn+1, hence Z dik([~] i $ Sn+ I ) has the same dei composition as a certain sum of principal indeoomposables Sj of indecomposables Sn+1: 7.8 Z dik( [~]i ~ Sn+ 1) i Z aj~j . j 7.7 Uk ~

17j

The branching theorem 4.52 implies, that the irreducible constitu-

141

ents of [~]i t Sn+l are exactly the [~] arising from [~]i by adding one node. Restricting these additions in such a way that the arising diagrams [#] have the same p-core, we can assure, that the decomposition of the arising representation of Sn+ 1 contains principal indecomposables belonging to one block only.

The node in the i-th row and j-th column of [6] is called an r-node (with respect to p) if j -i~r By (p) .

[6] t Sn+l
we shall denote the representation r-induced by [6] and consisting exactly of the representations multiplicity [#] (each with

1) of Sn+ I whose diagrams arise from [~] by

adding an r-node. Analogously we define the r-restriction of E~] to Sn_ I and call the procedure the r-inducin6 resp. r-restriotin6 process.

Using Nakayama's conjecture we obtain that the constituents of r [ 6 ] ~ Sn+ 1 h a v e t h e same p - c o r e s o t h a t t h e y a l l b e l o n g t o t h e same p-block of Sn I and that all the diagrams with this p-core r and out of [6] t Sn+ I are contained in [a] ~ Sn+ 1 . Hence we have (see Robinson C5],6.11):

7.10 If [~] and [ y ] arise from [6] by adding a node, then [ ~ ] = [ ~ ]

142

if and only if the added nodes are of the same residue class r modulo p.

This implies for the decomposition matrix:

7,11

~ z ~k([a]i

. ,

f Sn+ 1)

z bsU S , j
_ _

such that the U~ with b~ $ 0 belong to the same block of Sn+ 1.

-

With this fumdamental result we can start with our example p = 2. The only principal indecomposable of S I is [1], this is trivial. The residue class modulo 2 of the only node of the diagram [I] is O. Thus 0 o 1 1

U 1 ~ S2 = [ 1 ] $ S2 = ~ ,

[1 ] t S2 = [ 2 ] + [ 12] ~ U 1 t S 2"

S 2 is a 2-group, hence it possesses only one principal indecomposable and the last equation together with 7.11 implies that this principal indecomposable has the same decomposition as [2] + [ 1 ~ . Thus

2 =

[12]

is the decomposition matrix of S 2 for p = 2. To proceed with the r-inducing process we replace the nodes of the diagram by their residue classes modulo 2: 0

[2]:
We obtain

01

[12]:

14.3

U1 ~ 8 3

[2]

+ [12 ] ~ S3 = [32 + [132

U1 ~ S 3 .~. [ 2 ] + [12 ] ~' S3 = 212,12

Nakayama's conjecture implies that [ 2 , 1 ] is modular irreducible. Besides this block S 3 possesses only one further principal indecomposable, hence this one has the same decomposition as [3]+[I 3] and we obtain the decomposition matrix of S 3 for p = 2: forms its own 2-block and

D~,2 = t

[13-] r2,1]

(Por the sake of simplicity we often omit the O's.) Then

[3]=

010,

[2,1]:

01,

[13] .

0 1 0

Thus

~1 ~ s4 ~ [32 + [13] $ s4 _- ~
1 1

U1 I' S4 ~. [ 3 ] + [132 I S4 = [42 + [3,12 + [2,122 + [142 , '7.14

o U2 ~' S4 ~ [ 2 , 1 ] 1 U2 t 84 ~- [2,1] o f 84. = [ 3 , 1 ] 1 ~ 84. = ,~

+ [22 ] + [ 2 , 1 2 ] ,

S 4 has exactly two 2-regular classes so that D1,2~ consists of 2 columns. 7.14 implies, that the columns of the matrix

144

1 1
7.15 R4, 2 :=
o 1

[4] 1
I I

[3,1] [2 2] [2,1 2 ] [14. ]

of

1
are linear combinations the columns of D;, 2.

(with nonnegatve integral coefficients)

R4, 2 is a matrix of the same shape as D~, 2 and it is a lower triangular matrix. Hence its first col!~mn contains the first col~mn of D~ ,2 and its second column contains the second collm~ of D 1 ,2" The triangular form implies furthermore, that the second column

of R4, 2 cannot contain the first colum# of D I 4,2 so that the second ool1~m~ of R4, 2 is equal to the second column of D 1 4,2' since decomposition numbers are integers. This second col1~mn of R4, 2 cannot be subtracted from the first column of R4, 2 without getting negative entries such that the first column of R4, 2 is equal to the first column of D 1 4,2: 1 1 1 1 1 [4] [3,1] [2 2 ] [2,1 2 ] [1 4 ]

? .1,,~

D1

4,2

0 1 1

Then [4]: 0 1 0 1 , [.3,1]: 0 1 0 , [22]; 1 0 1 , [2,12]: 1 0 0 1 , [14]: 1 0 0 1 0

1

145

And we obtain:

[4]+[3,1]+[2,12]+[14] "
7.17

~ 85 = [5]+[3,2]+213,12]+[22,1]+[15] f S5 = 2 ( [ 4 , 1 ] + [ 2 , 1 3 ] ) ~ 85 = 3 ( [ 3 , 2 ] + [ 3 , 1 2 ] + [ 2 2 , 1 ] ) f S5 = [ 4 , 1 ] + [ 2 , 1 3 ]
1 1

(1) (21 (3) (4)

[3,1]+[22]+[2,12] "

Nakayama's

conjecture implies, that so

[4,1] forms a 2-block together

has

with [2,13 ]

that we obtain from (4), that [ 4 , 1 ] + [ 2 , 1 ~

the same decomposition as a principal

indecomposable of S 5 since (the number

this block contains only one principal indecomposable of modular irreducible representations known). There are two further principal indecomposables, and (3) suggest that we investigate,

in a given block of Sn is

hence 7.17 (I)

how the matrix

1
1 2

r3,2]
[3,12 ]

7.18

R5, 2 .-

2 1 1

2 2

E22,1 ]
[1 ]

[4,1] [2,13 ]
should be corrected to yield D I, 1 5
~

Here as in the following examples we observe,

that this matrix

Rn, p which we would like to correct to obtain Dn,p, has the I following important properties: (i) Rn, p is a direct sum of submatrices of the same shape as the

146

I submatrices of Dn, p (there is a general formula for the number of modular irreducible representations in a given block of Sn (see Robinson [5~,6.37) so that Rn, p has the same number of columns as Din,p" (ii) The columns of Rn, p are rational integral combinations (with nonnegative coefficients)of the columns of Dn,p" I (iii)The submatrices of which Rn, p is a direct sum are lower triangular matrices with nonzero entries along the leading diagonal. I Let us now assume that we have obtained by r-inducing on Dn_1, p a matrix Rn, p which satisfies (i)-(iii). We have to describe what can be done to get D n,p from Rn,p" I We fix attention on one of the SlJmmands of Rn, p (see (i)). Since (iii) is valid this summand is a lower triangular matrix.This implies that for a certain rearrangement of the columns of D n,p the I corresponding submatrix of Dn, pl is also lower triangular. We

assume, that this rearrangement has been carried out so that it remains to obtain from the i-th column of this s~mmand of Rn, p the i-th column of the corresponding s1~mmand of Dn, pl (we can be

sure, that the i-th column of the s1~mmand of Rn, p includes the I i-th column of the s11mmand of Dn, p at least once and this wanted i-th column of Dn,p is not included in a column farther to the I right in Rn,p).

147

Hence the column to the extreme right of the considered summand of Rn, p (cf. example 7.18) is an integral multiple of the corresponding col1~m~ of D 1 Thus we have to check whether this column n,p" can be divided by a natural number to yield a new column over ~.

~.~!~If the column to the extreme right of the considered summand of Rnl P can be divided by a natural number, this division has
, , _ ,., , , . , .

to be carried out to yield the corresponding column of Dn, p" I

Proof: We know that the Cartan matrix 1 Cn, p := t~l ~1 ~n,p~n,p

(tDn,p the transposed matrix) has a determinant whose value is a I power of p. Purthermore if D p arises from Dn, pl by adding a column of Dn, p

to another column of Dn,p, it is easy to see that I det(t 1, DI, ) = det 1 Dn,p n,p" Cn,p " With this in mind it is not too difficult to check, that det(tRn,pRn,p) = r~. ..rkdetOn, p , 2 I (I)

if r i E N is the multiplicity of the i-th column of Dn,p in the I i-th col~mn of Rn, p (after the suitable rearrangement columns of Dn,p has been carried out). 1 1 Finally we recall that Dn, p has maximal p-rank (see Curtis/ Reiner [1 ],(83.5)). Hence if it is possible we have to divide the considered column of the

148

to

the extreme right by p. That we have to divide these entries

by other common factors we see from (I), recalling that detC~,p is a power of p. q.e.d. Thus we obtain one col!3mn of D 1 anyway. n,p Por example the second column of 7.18 has to be divided by 2: In
1

1 2 1 1

1 1 1

7.20

R~, 2 :=

[5] [3,2] E3,12 ] [22,1] [15 ] [4,1] [2,13 ]

at least the second and third columns agree with the corresponding oolum~s of D 1
5,2"

It remains to decide whether the second col1~m~ has to be sub~raoted from the first one in R~, 2 to obtain D~, 2. This can be decided using a result of ~.H. Peel on the deoompositiom numbers of [n-2,2]. Before we mention Peel's result we give the other general result on decomposition numbers of Sn (Farahat [I], cf. also Kerber [1]) and which is of equal importance:

7~21 If p does not divide n, then [n-1,1] is irreducible.

m ,

If p divides n and n>2, then there are exactly 2 irreducible constituents of [n-1,1], each of them with multiplicity 1, and one of t h e m is the identit~ representation.

149

Peel's result (Peel

[1]) reads as follows:

7,22 1) If p does~ot divide n-2 we distinguish the cases p=2 and

p~2s
a) In case that p~2 and (i) p ~ n-l, then [n-2,2] is irreducible. If

(ii)p ] n-l, then [n-2,2] has exactly ~wo different irreducible constituents, each with multiplicity I,

one of them is the identity representation. b) If p=2 and (i) n=2a+1, 2 ~ a, [n-2,2] is irreducible. (ii)n=2a+1, If

2 I a, then [n-2,2] contains exactly ~wo iteach with multiplicity I,

reducible constituents,

one of them is the identity representation.

2) If on the other hand p divides n-2, and a) p=2 we have to distinguish the following two cases: (i) If n=2a, 2 I a and a>2 ([2] and ~ are irreducible

as we have seen above), then [n-2,2] possesses exactly two different irreducible constituents, each

one with multiplicity I, one of them is of dimension

n-2.

(ii)If n=2a, 2 t a, then [n-2,2] contains exactly three

150

different irreducible constituents, each one with multiplicity I, one of them is the identity representation, another one is of dimension n-2. b) If p$2, En-2,2] possesses exactly 2 different irreducible constituents, every one with multiplicity 1, one of them is the irreducible representation [n-1,1]

(of. 7.21).
Prom 7.22 I) b) (ii) we obtain with 7.20:

1
7.23 D1 5,2 = 1

[51 [3,2] !

1I

[3,12 ] I2 2 , 1 ] [ [15 ] [2,1 ~] .

The r-inducing process, Nakayama's conjecture and 7.21/7.22 yield

1

D~, 2 =

1 1 2 1 1 2 1 1 1

1 1 1 0 0 1 1 1

[6] [5,1] [4,2]

[4,12 ]

[3 2 ] [2 3 ] [3~13] [2z,12] [2,14 ] [16 ] [3,2,1]

The application of the r-inducing procesz to D 1 6,2 yields the following matrix (we used 7.19 and 7.22):

151

1 0 1 1
I

1 1 I
0

[7] [5,2]
[5,12 ]
[4,2,1

1
1

0
1

[32,1] [3,2 2 ]
[,2,1 2

7.25

R7, 2 :=

1 0
1

1 1
1

[3,1&] [22,13
[17 ]

[6,1]

[4,3]
2 1 1 1 1

[4,13 ] [23,1]

J [2,15 ]

To check, that this matrix is the decomposition matrix of S7 we use a nice trick: 7.24 implies that the difference of the characters of ~ and [6] is the Brauer character of F 3, the third of S 6 which belongs to the

2-modular irreducible representation third column of D~,2. ~1r short :

Hence
7.26

we may induce to get

~z ? s7 ~ ([3-N~])

t s7 ~ ~

t s7 - [6] t s7
.

[4,'3] * [3-~,1] - [7] - [6,1]

Now we look at this decomposition and 7.25.

The right hand side of 7.26 contains each 2-modular irreducible representation of S7 with nonnegative multiplicity. From 7.25

we know, that [7] as well as [6,1 ] are irreducible representa-

152

tions ( s e e a l s o

7.21)

such that (see 7 . 2 6 )

[4,~+[3-~,I] contains

[7] as well as [6,1] at least once. Hence in 7.25 neither the third eol,mn can be subtracted from the first one nor the fifth colnmn from the fourth one:

7.27

R7, 2 = D$, 2

Applying the r-inducing process to this matrix we obtain:

1 1 1

0 1 0
1

1 1 1
2

1 1 1
1

[8] [7,1] [6,2]

[6,12 ]

[5,3]
[5,15 ]

7 28

R 8

92

" -o

0 2 2 2 1 2 2 2 0 0 1 0 1 1

1 1 0 2 2 0 0 1 1 1 1 1 1

0 1 1 2 1 0 1 1 0 1 1 1

[4.2 ]
[4,5,1]

[4,2 2 ] [4,2,12 ] [4,14 ] [32,2] [32,12 ]

[3,22,1]

[2 4 ]
[23,12 ]

[3,15 ]
[22,14 ] [2,16 ]

1 1

[18 ] [5,2,1] [3,2,13 ]

7.21 and 7.22 imply, that at most the fifth column of this matrix has to be subtracted from the first one and/or the fourth column from the second one.

155

Once more we consider an induced character: 7.29

t 88 =

([4,3]-[6,1])

s8
.

Since the first three rows of R8, 2 agree with the first three rows of DI,2 (cf. 7.21/7.22), [4,3,1] contains at least twice the irrecontains

ducible representation [8] as well as [ 5 , ~ + [ 4 ~ + [ 4 , 3 , 1 ]

at least thrice the second irreducible representation [7,1]-[8]. This implies D1 8,2 "

R8,2

r-inducing on the columns of D I yields some possibilities for 8,2 the col~mns of D I (see 7.32 on page 154) which we would like to 9,2 investigate now. We consider at first the lower right hand submatrix for to obtain the box of D I 9,2 which includes [8,1]. That the column to the extreme right has to be divided by 3 we obtain from 7.19. Since the first entry of the second possibility for the first colnmn is I, the first possibility for this column contains twice the first column of the corresponding box of D I 9,2: I 0
I

7.31

3 3 3 I 3 0 I

--q

I~.

~3

o
i,o

C~

o o

I~ i -~.

r~

~1

o ~ ~ 0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 ~ ~ ~

~ H

o"
~ ~ ~ ~ ~ ~ ~ ~ 0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 0 ~ ~ ~ ~ 0 ~ ~ ~ ~

(D

~ ~

~ ~

Is" o o

o
~ ~ 0 0 ~ ~ 0 0 ~ ~ ~ ~ 0 ~ 0 ~ 0 ~ 0 0 ~ ~

~
~ ~ 0 ~ t~ ~o-~ I"o ~ ~ ~ ~ ~ ~ o ~
I

r~ r~

~1 +

o ,,n

I~ H

~ o

Iii ill

I11

el-

H.

~ 0 ~

I::

,~

,~-,~
DO ~ ~ PO~...~ ~ DO~, ~ '~,~1~--,,~ ~

el,

-~ ~

~ ~

0'~

I~ .1~ -,~

-,~ ~

'~ ~

.,1~, DO

I~,)~ ~

",~

I~t,~,~

I'~

et-

155

7.33

would contain the eigth columm of D~, 2. This implies, that

1

1 01
11

7.34

11
1 0

01 1 are the two columns to the extreme right of D 1 9,2" Hence there remains to examine the following two possibilities for the sixth colnmn and the corrected seventh and eigth colnmn: 1 1 0 2

131
7.35 31 01 3 3 1 1 33 11 I 3 10 3 1 0 02 1

111
The combination of these two possibilities for the first colnmn gives, that

7.36

contains the first column of the corresponding box of D~, 2.

156

Thus it remains to decide whether 1 0

1

[8,1] 1
1

7 -37

0 0 0
1

0 1 1
1

1 1 I 1

[6,3~ [6,1J] [4,3,2]

[4,3,1 2 ]

[4,22,1] [32,2,1]
[23,13 ]

0 [4,1 5 ]

0 0 1

0 1

[2,17 ]

can be a submatrix of D I 9,2 or not. To decide this we consider again a special induced character:

7.38

F~ t

s9

= ([~,2]-218'i)

s9

= [4,3,2]+[3"~'~+[32,2,1"]-219]-218,1]

This decomposition implies, that [4,3,2]+[32,2, 1 ] contains at least twice the irreducible constituent [8,1] so that 7.36 describes

the first column of the considered submatrix of D~, 2. It remains to investigate the submatrix of 7.32 which contains the identity representation. That the only possibility for the fourth column has to be divided by 2 we obtain from 7.19 since the fifth coll,mn cannot be subtracted. Hence 7.31 suggests to examine the matrix 7.39 (see the following page), whose i-th column contains the i-th col~3mn of D~, 2 (I~i~5). Obviously it remains to check the first and the second col~mn of 7.39. The other columns are columns of D I 9,2 as they stand since

157

from these columns not any other one oan be subtracted without yielding negative entries.
1 1 1

2
1

1
1 1

[9] [7,2] [7,12 ]

[6,2,1] 1 1 0 1 1 1 1 0 0 1 1
0

7.39

0 2 2 3 2 2 3 2 2 2 2
1

1 1 0 2 2 1 2 0 0 1 1
1

0 1 1 1 0 0 1 0 1 0 1
1

[5,4] [5,3,1]
[5,22 ] [5,2,12 ]

[5,14 ] [42 , 1]
[4-,2,13 ]

[33 ] [32,13 ] [3,23 ]

[3,22,12 ]

2 0
1

1 1
1

0 1

[3,2,14 ] [3,~6] [24,1]

[22,15 ] [1 9 ]

To examine the first and second column of this matrix we consider

7.4o

~4

t s 9 --

([8]+[5,3]-[6,~1)

t s9

: [9]+[8,1]+[6,3]+[5,4]+[5,3,1 ]-[7,1--~-[6,2,'I"]'-[6,1-~] Since the first three rows of 7.39 are correct as they stand (cf. 7.22), the representation ducible representation [7,1~+[6,2,1i contains the second irre-

[7,2]-[9] with multiplicity 2 (notice that

the third colnmn of 7.39 cannot be subtracted from the second

o n e ) . Hence 7.40 i m p l i e s t h a t [ 5 , 4 ] + [ 5 , 3 , 1 ]

contains [7,2]-[9]

at least twice such that neither the fourth nor the fifth column

158

has to be subtracted from the second col~mn of 7.39. The secon~ col1~mn of 7.39 is therefore correct as it stands. It remains to consider the first column and to decide whether the third and/or fifth coll,mn have to be subtracted or not. The first three entries of this col~Imn are correct as they stand. We consider the decomposition 7.41 ~3 ~ $9 ~ ([8]+[6,2]-[7,1]) ~ S 9
=

[8,1]+[7,2]+[7,1--~ contains thrice the irreducible representation [9], hence at least this is valid for [9]+[7,2]+[6,2,1]. Thus the third column has not to be subtracted from the first one so that even the first four entries of this col1~mn agree with D I 9,2" With this we return to 7.40: [7,1~+[6,2,1] contains thrice the irreducible constituent [9] such that this representatiau is contained in [9]+[5,3,1] at least with multiplicity 3. Hence also the fifth column has not to be subtracted from the first one. Hence 7.42 (on page 159) is the decomposition matrix of S 9 with respect to p=2. Thus we have verified without any an explicit reduction of a representation the decomposition numbers of Sn for p=2 and n_(9 which Robinson gave (Robinson [5],[6]). We have shown, how a combination of the r-inducing process together with a use of Farahat's and Peel's results yields these far reaching results. But it

h'

~.~

~.

h~

PO

~"

o
~ ~ 0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 ~ ~ ~ ~

o
~ ~ ~ ~ ~ ~ 0 0 ~ ~ ~ ~

I~

o
0 ~ ~ ~ ~ ~

0
~ 0 0 ~ ~ 0 0 ~

~.

c~

0 P~
0 ~ ~ ~ ~ ~ ~ 0 ~

-4 --

~+ ~.

0 0

C.
~D

4~
~ 0 ~ 0 ~

,...a

I~

ro

0
k24~ ~ --~ ~ ~, --~ --~ ~ 0 fOL-~ ~

0
k/~L.._~ ~ --~ ,, ~ 4::,.--.~ k2~ ~ --~ ~ k,W ~ ..a ~ ~ --.% ~ ,.~

(D

II

I~

c~

~+

(~

~. k

O]

k~

160

1 11

7.43

111 211 1011 2111 00111 302111 301001 13111 212110 001010 4.110111 5131211 5121111 2121100 2110101 2111101 2100001 5121111 2111101 5131211 4.131110 2111 000 2100001 2110101 4110111 301001 302111 21100 00101 00111 1011 111 11 1

(cf. the table 2-10 in Robinson [5], Appendix, which has to be corrected). The last for columns of 7.43 agree with the last four col~mns of the corresponding submatrix of D! 10,2" These methods have been used also for the case p=3, where the first difficulty arises at n=8. The question is whetherE5,3] contains [8] or not. The answer is negative: [5,3] does not contain [8] and using this result, the decomposition matrices have been calculated up to n=10 (Kerber/Peel [I]). The reader can find
o

there a necessary and sufficient condition that [n-3,3] contains

161

InS. Por further interesting results especially on the decomposition of hook-representations In-r,1 r] the reader is referred

to Peel [2].
Concluding these considerations of the decomposition numbers of Sn and gathering up our experiences we dare to give a conjecture:

7.44 Conjecture: The submatrices of which the decomposition matrix Dn,pl of Sn for p is a direct sum are for a suitable rearrangement of the col1~mns lower triangular matrices with 1's along the leading diagonal, if the first rows of the considered submatrix correspond to diagrams with no p rows of equal length in their natural order.

Parts of this conjecture but not the full statement have been proved by Robinson and 0.E. Taulbee (Robinson [5], Taulbee [I]).

Concluding this section we would like to investigate what can be said about the decomposition matrix D I of A n if D 1 is known. An,P Sn,P 7.6 yields the distribution of the ordinary irreducible representations of A n into p-blocks Hence for to obtain D I it remains An, p

to describe how we can get its columns from the columns of D I Sn,P and to describe what happens with the corresponding entry of D Sn,P , if the representations are restricted to A n . I

162

As in the ordinary case we can apply Clifford's theory of representations of groups with normal divisors. Our aim is to describe modular irreducible representations F S of Sn whose restriction is reducible resp. irreducible. From Clifford's theory we concludethat the irreducible representations of Sn over an algebraically closed field K (of any characteristic) can be obtained in the following way: Take an irreducible representations FA of An and find its inertia group. Since ISn:Anl!2 this inertia group is A n or Sn. If A n is the inertia group, then PAfSn is an irreducible representation of Sn. If Sn is the inertia group of FA, the FA can be extended to an irreducible representation~ A of Sn, and with this representatien we can construct two irreducible representations of Sn: FA = ~A @ [ ] ' @

(which need not be different, e.g. if char K = 2: i2] = [I--~, and hence ~A = ~ A @ [1--1~).

And in this way we obtain all the irreducible representations of Sn This implies:

If F S is an irreducible 2-modular representation of Sn (n>1),

.,. ,,

then the following is valid: (i) F S & A n reducible (ii)Fs ~ P8 @ [ln] ~ ~ Z FA: F S = FA ~ S n. = (P8@[ln]) $ An irreducible.

PS $ ~

163

If a=~', then [ ~ ] ~ An splits into two mutually conjugate and irreducible representations of An: [a] ~ An = [u]+ + [u]so that [~]+(a) ~ [~]- , V a E 8n\A n . Thus also

y(a) ~ i1 i
and hence the constituents of [~]- are conjugates of the constituents of [~]+. This implies ~ If PS ~ PS @ [1'hi' and the multiplicity of PS in [~] is odd and ~ = ~'
9

then Ps~An is reducible

Suppose now that using 7.46 we have succeeded in picking out the columns of D I which belong to modular irreducible represen8n,P tations PS whose restriction to An is reducible (i.e. PS is selfassociated with respect to An). Then under certain circumstances (which are fulfilled in all the known cases D~n,p) we are able to evaluate D I at once. An,P Let us consider the row of D I which belongs to [u] and the Sn,P column which belongs to PS" We denote by a resp. b the multiplicity of ~S resp. PS @ [ln] in i~] such that D I contains the Sn,P following submatrix:

164

~S [G]
[~t]

Ps [ '1'n]

...
..

(i) If ~ + ~', PS ~ FS [In] :

In this case the row of [5'] and the column of PS @ [Ini has to be cancelled, and in the row of [5] ~ An we have in the column of PS ~ An the deoomposition number a+b, since [~i " a~ S + b(F S @ [In]) +--.

[5] ~ An .

a(~ S ~ An) + b(F S @ [I n] ~ An) +...

= s An +
. . . .

(ii) a + m', PS " FS " [In]' FS & An irreducible:

It is trivial, that a is the multiplicity of FS ~ A n in [5] ~ An . (iii) ~ + 5', PS ~ An reducible (~ PS N ~S @ [ln]):

Then FS ~ An ~ FS + FS with two mutually conjugate and irreducible representations ~ of A n . Obviously [~] I An contains F~ as well

as F~ with multiplicity a.
(iv) a -- 5', F s ~ FS [In] :

a = b is the multiplicity of PS ~ An in [5] + as well as in [~]-

165

since FS $ An is selfcomjugate. (v) ~ = ~', PS N PS [1hi, FS ~ An irreducible:

PS ~ An is selfconjugate again. Thus a = b is even and a/2 is the multiplicity of PS ~ An in [m]+ as well as in [~]-.

The theory provides an answer only if a = O or a = I. In this case we have the submatrices

Co~]-

o I
o

or

[,,I-

I
I

(resp. IO ~]if we use another denumeration). Gathering up we have obtained (see Puttaswamaiah maiah/2obinson

[1],

Puttaswa-

[1],

Kerber

[3]):

If D I contains the submatrix Sn,P

FS Fs ~ [1 ~ ]

[e,]

then if (i) ~ ~ ~' PS ~ FS @ [In]' D1 contains the submatrix An,P

166

~sSAn
[=] ~, An
[a+b] (ii) If ~ + =', PS N F s @ [sni, PS ~ An irreducible:

Fs~An
[~]~A n [a ]

(iii)~ ~ ~,, Ps ~ An " Ps + PS: +

[~s ~ An [a
(iv) ~ = ~', PS + PS @ '['ln']':

a]

PS ~ An [=]+

[a]-

[:1
F8 ~ An

(v) a = ~', PS N p S @ [In]' PS ~ An irreducible:

E~

r~/2 ]

[-]+
resp.

o]
I .

As

an example we give the decomposition matrix of A 9 which arises for p = 3 given in Kerber/

from the decomposition matrix of S9 Peel [1],

N o
et-

(~ N N
ul ~ CD ~

U
-

O O N N O O ro -A -'~ ~ O ~ O ..A -A O O N ..A ro O O O O .,A O ..~ I~O O O "-A O ro -~ O O -'~ O O .-.L

-..~ - A

ro

-.~

...~ -A ~ ~

O -~

O -~ O -A ,..A

.~ -A

..L

o~ ~

~,

~'

el-

168

8. Generalized decomposition numbers of

symmetric and alternating groups

l|

If p is a prime number and G a finite group with ordinary irreducible characters Ci, Brauer characters k of the irreducible re-

I p-modular representations, and decomposition mumbers d i k w i t h spect to p we have for a p-regular element gEG: ~i(g) = Z d ~ k ( g ) k This can be generalized to general group elements g. 8.1

As is well known an element gEG is a product of a uniquely determined p-element x with a uniquely determined p-regular y which commutes with x: 8.2 g = xy = yx .

Let us call x the p-component, y the p-regular component of g. g rums through a complete system of representatives of the conjugacy classes of G if in 8.2 x runs through a complete system of the p-classes of G and y - while x is fixed - runs through a complete system of representatives of the p-regular classes of the centralizer CG(X ) of x in G. Since CG(1) = G, the following result of Brauer generalizes 8.1:

If xEG is a p-element of order pr and if ~k are the Brauer

169

characters o f 0G(x) with respect t o p, then there exist algebraic integers d~k in Q(~) ( a primitive pr-th root of unity) depending only on x and satisfying ciCxy) = Z d~k~Cy)
k ,

yECG(X) , y p-regular.

(of. Ourtis/Reiner

[I ], $ 90A)

If now D x indicates the matrix of these algebraic integers

8.4 Dx == ,

it can be shown, that for an x' conjugate to x the matrix D x' arises from D x by a permutation of the columns. Thus for an investigation of these algebraic integers d x k W e i need only consider

the matrices D xj for a complete system, say for [Xl:=1,...,Xu], of representatives of the p-classes of G.

For such a fixed system of representatives we denote for short

X

8.5

Dv

==

l v<_u.

The matrix 8.6 D := (DI,...,Du),

which is therefore uniquely determined up to a column permutation is called the generalized decomposition matrix of G with respect to p. Its entries are called the Reneralize d decomposition numbers of G with respect to p. Its first col~3m~s contain D I If yj are the representatives the decomposition matrix of G. of the p-regular classes of CG(X v)

170

we indicate as follows:

8.?

z v .= (ci(xvyQ),

~v

:=

(~i( y Q )

and hence the matrices

U

8.8

# :=

$ jv , Z := (Z 1,...,z u) v=l

satisfy the equation

8.9

Z =D#

We would like to evaluate these matrices D of generalized decomposition numbers of the symmetric group. At first we notice that since a matrix of Brauer characters is not singular:

8.10

Dv = z V ( J V ) - 1 .

Z v is known from the character table, thus it remains to derive the matrices @v of the Brauer characters of centralizers of p-elements of S n. E Sn is a p-element if and only if the lengths of all the cyclic factors of ~ are powers of p, this is implied by 1.11. Prom 2.32 we obtain, that the centralizer of such a p-element is a direct product of symmetries of cyclic p-groups:

8.11 If ~ E Sn is a p-element of type T~ = (al,...,an), then we have for the centralizer of ~:

CSn

(~) = (c pi~Sa pi ) i

171

This implies (of. also Osima [4]) that the wanted matrix ~

of the

Brauer characters of the centralizer of the p-element ~ is a direct product of the matrices a i P
@x= X i a

of the Brauer characters of the Sa pi

.

pl

Since Q and hence also GF(p) is a splitting field for Sn, the api and therefore @~ is rational integral. Hence (0sima [4]):

8.!3 The Brauer characters with respect to p of the centralizers of p-elements of symmetric groups are rational integral.

Henoe (@~)-I is a matrix with rational entries and therefore the same is valid for D v (see 8.10) since the ordinary irreducible characters of Sn are rational integral valued. And since the generalized decomposition numbers are algebraic integers in Q(E) (see 8.3), they are not only rational but even rational integral (Kerber [I], see also Osima [4]):

The generalized decomposition numbers of symmetric groups are rational integral.

Por an example we evaluate D6,2, the generalized decomposition matrix of S 6 for p = 2. (i) The 2-classes of S6 are represented by ~I = I, ~2 = (12),

~3 = (12)(34),

~4 = (12)(34)(56),

~5 = (1234),

~6 = (1234)(56).

172

The centralizers

of these elements are of the form

0(~1) = S 6 , 0(~2) = 02 x 84 , C(~ 3) = O2~S 2 x S 2 . 0(~4) = 02%S 3 , 0(~5) = 04 x S 2 , C(~ 6) = 04 x 02 . (,,)Hence we need only the matrices of Brauer characters $2, S 3 and S 4. These matrices are of Si,

@1 = @2 = (1)
as can be evaluated

and

'3 = '4 =

[: 1]
-1 '

easily with the known character tables matrices D I 2,2' D I and D I (see 3,2 4,2

and the decomposition

section 7). Thus we obtain

,2 = ' 4 X t I = '4 = '3 =

[12 - i ]

, ,3 '5 = '1X

= ,2 x , 2 = (1) , ' 2 = (1) = ' 6

in section

[12 - : ] '

(,1 has been omitted since D~, 2 has been evaluated 7). Because

[: 1]_1__ ii/3 ,j31

-~ [ 2/3 -~/3 .I
r

and 3
3
=

0 -~

3 -2

o I

z~ _i _~

z,

0 _~

3 0 2 -1

o
0

-1 -1

173

as

can be read off from the character table of $6, we get decomposition matrix of S 6 with respect

for the generalized to p = 2:

1 1 1 2 1 1 2 1 1 1

D6,2 =

1 1 1 0 0 1 1 1

1 1 1 1 1 1

I 1 I I 0 1 I 0 -I 0 0-1 -1 -I -I -1 -I

1 1 -12

1 0 -t -1 -1 0 1 -1 -1 1 -1

t ,1

,o

1 o -1

-1
,1 ,1

0 i-1 1

t 1 1

(the columns belonging to @3 = 15 = ~6 = (1) agree with the concerning columns of the character table). The generalized decomposition numbers of A n can be evaluated of p-elements simi-

larly as long as the appropriate centralizers direct products of generalized

are

alternating groups C i%Aa . or P p~ have (I) as matrix of Brauer characters (such that the columns agrees with a column of the character table). generalized decomposition matrices In this way the

of A n with respect to p=3 have

been calculated for n~7 (Kerber

[1]). than in the case numbers of

During these calculations we see that~otherwise of the symmetric group~the generalized

decomposition e.g.

A n are not in general rational I 8.16

integers,

,3

93

"-

(-1-iW)/2
decomposition matrix of A 3 for p = 3. Thus

is the generalized

174

(Kerber

[I S) :

8.17 The generalized

decomposition mumbers

of alternating

groups

are not in general rational integral.

Nevertheless

(Osima [5]):

8.18 The generalized are rational

decomposition numbers

of alternating groups

integral for p = 2.

Proof:

The decomposition

numbers

of A n are rational

integral by

definition,

thus D I is a matrix over 2. to show that @v and the Z v are matrices over Z

Hence it suffices if v>1.

Thus it is enough to prove that the Brauer characters with respect to p=2 of centralizers of 2-elements ~I of A n are rational inte-

gral and that this is valid also for the values of the ordinary irreducible characters of 2-singular elements of A n-

The last statement theorem 4.55, ly 2-regular.

is valid as can be seen from Prcbenius' of splitting classes are obviousthat the values of the Brauer in

since permutations

It remains to prove,

characters with respect to p=2 of centralizers A n are rational integral. Thus the proof of the following lemma proof of 8.18:

of 2-elements

(Osima [5]) completes

the

175

8.19 If ~$I is a 2-element of A n , then the irreducible Brauer characters of CS n (~) with respect to p=2 remain irreducible
_ _ _

if they are restricted to C An(~) such that the values of the Brauer characters of CAn(~) are rational integral as well

(of. 8o13).

Proof:We shall show that there is a subgroup of CAn(~) which has the same matrix of Brauer characters as CSn(~). Let ~ be an even 2-element of Sn such that ~ I and hence

CSn(~) = i (Ci%Sai) = j (C2 J~%Sa2 j) = j (C*~S~2 2j) . J ~1 implies that there is a k>1 so that a k>0. Let us fix such
-2

a k. A subgroup of CSn(~) with the same matrix of Brauer characters as CSn(~ ) is al X .... G := S' X...X S' (C2k~Sa2k) S' a2k-1 a2k+1 Let us consider the subgroup G + := G G An ~ CAn(~)

We would like to show, that G + has the same irreducible Brauer characters as G and hence as CSn(~), what implies that the statement 8.19 is valid. ~rom Clifford's theory of representations of groups with normal

divisors we know, that a normal divisor N of index 2 in B has the same irreducible Brauer characters as B if no 2-regular conjugacy class of B splits into conjugacy classes of N.

176

Hence it suffices to show that no 2-regular class of G splits into two 2-regular classes of G +. If CG is a 2-regular class of G it is a product 0 G = C10204... ' of 2-regular classes C j of the direct factors Sa2 j if j~k, and C k is a 2-regular class of 0 k~S a . 2 2k ' Prom the results of section 3 we obtain that Ck E S a k' such that 2 all the factors C i are contained in G + as well. It is obvious, i>I. Hence that the factors C i don't split into G+-classes if

it remains to show that C I does not split, which can be

proved as follows. al Set p be any odd permutation out of S' ; since k>O there is an odd permutation ~ E C k~Sa and we have pa E G +. Hence C I does 2 2k not split. This completes the proof of 8.19 and so also of 8.18. q.e.d. Hence (Osima

[5]):
is a 2-element of type T~ = (al,...,a n) of An,

8.20 If ~ 1 then

CSn (~) as well as CAn (~) have

~al x ~a2 x ~a x .-.

as matrix of Brauer characters with respect to p=2 if ~a 2j is the matrix of the 2-modular Brauer characters of S a . 2~

177

Using this lemma M. 0sima evaluated the generalized decomposition matrices of A n with respect to p=2 for n=6,7,8,9 (Osima [5]), the matrices for n~7 had already been known (Kerber [I]).

This shows, that for to prove that the generalized decomposition numbers of a finite group G with respect to p are rational integers it suffices to show that the values of the Brauer characters of centralizers of p-elements as well as the values of ordinary irreducible characters on p-singular elements are rational integral. We would like to check wreath products G~S n. whether this is true for certain

8.21 If GF(p) is a splitting field for the centralizers of the p-elements ~I of the finite group G and if the values of the ordinary irreducible characters with respect to p of G~S n on p-singular elements are rational integral, then the generalized decomposition numbers of G%S n with respect to p are rational integral.

Proof: Set (f;~) be a p-element of G~S n and of type (aik). For its centralizer we have (cf. 3.25):
8.22 ~
-

x
i,k k

ik

Hence it suffices to consider the Brauer characters of the factors

8.23

OG~Sk(f~k;~k)-Sai k

We would like to show that the values of their Brauer characters

178

with respect to p are rational integral. From the results of section 5 we obtain that it suffices to show that the p-modular representations of 8.24 i . j CG,,~k(fik,~ik)

can be written over GP(p). But this is valid since 8.24 is an extension of the centralizer of a p-element of G with a cyclic p-group (cf. 3.19). Thus GF(p) is a splitting field for the subgroups 8.24 and hence for the basis group of 8.23 as well. This implies (see section 5) that GF(p) is a splitting field for the groups 8.23, too, since it is a splitting field for symmetric groups Sai k as well. Hence GP(p) is a splitting field for OG~Sn(f;~) (see 8.22) such

that the values of its Brauer characters are rational integral. The decomposition numbers of G%S n are rational integral by definition, the values of the ordinary irreducible characters of G%S n on p-singular elements are rational integral by assumption, hence (of. 8.10) the D v are matrices over Z. q.e.d. 8.11 and the results of section 5 imply that GF(p) is a splitting field for the centralizers of p-elements in Sn, thus as a special case of 8.21 we obtain:

8.25 The generalized decomposition numbers of Sm~S n with respect to p are rational integral for all m, n and p.

179

As we have seen, the p-modular irreducible representations of the centralizers CSn(~) of p-elements ~+I such that ~ E An remain irreducible if they are restricted to CAn(~) such that GP(p) is a splitting field for CAn(~) as well. Hence also 8.18 is a special case of 8.21 (G:=Am, n=1). For G := [I] we obviously obtain 8.14.

Multiplying DSn,P resp. DAn,P with the transposed of its complexconjugate we get the matrices CSn ,p := 8.26 CAn,P := At~n,p'DAn,P = C pl ~ ,.. ~ C pv , the ~eneralized 0artan matrix of Sn resp. An, consisting of submatrices 0 ~i resp. C Pj along the leading diagonal which are

S+~n,p.DSn,P = C ~1

~ ... ~ C

~u

the Cartan matrices of the centralizers of the p-elements ~i resp. pj (cf. Curtis/Reiner [I], ~ 90A). For example

%6,2 =
4

[i] , ,

, [16] ,

,[8] , [8].

The ordinary irreducible characters of the centralizer n (~) = X (C i~Sa ) CSn i=0 p pi of a p-element ~ of Sn of type T~ = (al,...,a n) are of course the products

CC1--.C n

of irreducible ordinary characters ~i of the direct factors

180

0 i~Sa p pi (C i := I, if api=0 ). Since n 0Sn(~)/O1~Sal = x (C i~S a ) i=I p pi possesses only one p-block (cf. 5.31) we have (0sima [4]): 8,27 Two ordinary irreducible characters C=C...C n and C'=C'...C n' of OSn(~). (~ a p-element of type T~=(al,...,an)) belong to the same p-block if and only if C and C ' belong to the same p-block of Sal, i.e. if they bo~h have the same p-core. According to this we denote the diagram of C as the diagram of C such that now 8.27 reads as follows: 8.28 Two ordinary irreducible representations of the centralizer of a p-element of Sn belong to the same p-block if and only if their diagrams have the same p-core.

Two corollaries are (Kerber [I], Osima [4]): 8~ If ~ is a p-element of type (al,...,an), then CSn.(X) has only one p-block if a1~I and p~2 resp. if a1~2 and p=2.

An application to the character table is:

181

8.30 If B is the block of Sn which contains the identity representation [hi, then Ca(xy) = 0, V [aS E B if aI~I (x a p-element of type (al,...,an)) and p~2 resp. if ai~2 and p=2.

These results follow from a result of Brauer (Brauer [2], main theorem), that every column of the generalized decomposition matrix of a finite group G contains nonvanishing entries only in the rows of a certain p-block (8.30 can be obtained from the Murnaghan-Nakayama-formula as well). 8.29 suggests that we conclude with a hint at an important result of 0sima (0sima [4]) which establishes the connection between the p-blocks of Sn and the p-blocks of the centralizers of p-elements (see Brauer [2], (6A), also Curtis/Reiner [ I ] , ~ 87, 90A) :

8.31 The block ~ of CS~(~) with p-core [~] determines the p-block B of Sn with the same p-core IV].

References
m

N.N. Aizenberg:

[I] On the representations of the wreath product of finite groups. Ukrain Mat. ~. 13, no. 4 (1961), 5-12. (Zbl 109, MR 25)

E. Bayar:

[1] Eine neue EinfGhrung in die Darstellungstheorie symmetrischer Gruppen. Mitt. math. Sem. Univ. Giessen 81 (1969), 1-45. (Zbl 197,~R 39)

J.L. Berggren:

[1] Finite groups in which every element is conjugate to its inverse. Pacific J. Math. 28 (1969), 289-293. (Zbl 172, MR 39)

R.L. O.N. P.R. M.B.

Bivins/ Metropolis/ Stein/ Wells:

[I] Characters of the symmetric groups of degree 15 and 16. ~ath. Tables Aids Comput.

(195#), 212-216. (Zbl 56, ~

16)

H. Boerner:

[I] Representations of Groups. North-Holland Publishing Company, Amsterdam, 1963. [2] Representations of Groups. Second revised edition. North-Holland Publishing Company, Amsterdam, 197@ [3] Darstellungstheori~ der endlichen Gruppen. Enzyklop~die der math. Wiss. Band I,I, Heft 6, Teil If. Teubner-Verlag, Stuttgart, 1967.

R. Brauer:

[I] On a conjecture by Nakayama. Trans. Roy. Soc. Canada, Sect. III (3) 41 (1947), 11-19. (Zbl 29, MR 10) [2] Zur Darstellungstheorie der Gruppen endlicher 0rdnung II. Math. Z. 72 (1959), 25-46. ~Zbl 93, MR 21)

183

R. Brauer/O. Nesbitt: [1] On the modular characters of groups. Annals of Math. (2) 42 (1941), 556-590. (Zbl 27, MR 2) M. Burrow: [I] A generalization of the Young diagram. Can. J. Math.

(1954), 498-5o8. (Zbl 56, m~ 16)

[2] Representation theory of finite groups. Academic

Press, New York/London, 1965. A. Cauchy: [1] Exercisem d'analyse et de physique math~matique. Band III, Paris 1844. A.H. Clifford: [1] Representations induced in an invariant subgroup. Annals of Math. (2) 38 (1937), 533-550.

(Zbl I?)
A.J. Coleman: [1] Induced representations with applications to Sn and GL(n). Lecture notes prepared by 0.J. Bradley. Queen's Papers in Pure and Applied Mathematics, no. ~. Queen's University, Kingston, Ontario, 1966. (Zbl 141, MR 34) S. Comet: [I] Improved methods to calculate the characters of the symmetric group. Math. Oomput. I.~ (1960), 104-117.

(Zbl 103, MR 22)

H.S.M. Ooxeter/W.O.J. Moser: [1] Generators and relations for discrete groups. Springer-Verlag, Berlin-GSttingen-Heidelberg 1957.

O.W. Curtis/l. Reiner: [I] Representation theory of finite groups and associative algebras. Interscience Publishers, New York 1962. H.K. Parahat: [I] On the natural representation of the symmetric groups. Proc. Glasgow Math. Assoc. ~ (1962), 121-136. (Zbl 107, MR 25)

184

J.S. Frame: [1] Orthogonal group matrices of hyperoctahedral groups. Nagoya Math. J. 27 (1966), 585-590. (Zbl 145, MR 33)

[I] The hook graphs of the symmetric group. G. de B. Robinson/ Canadian J. Math. ~ (1954), 316-324. R.M. Thrall: (zbl 55, MR 15) R. Erucht:

J.S. Prame/

[I]

Coronas of groups and their subgroups, with an appli~ cation to determinants. (Spanish) Revista Union Mat. Argentina ~ (1942), 42-69. (Zbl 61, MR 4)

Y. Gtindttzalp: [1 ]~ber die gewGhnlichen irreduziblen Oharaktere der symmetrischen Gruppe. Mitt. math. Sem. Univ. Giessen 8_!I no. 2 (1969), 1-53. (Zbl 196, MR 39) , B. Huppert: [I] Endliehe Gruppen I. Springer-Verlag, Berlin-GSttingen-Heidelberg-New York 1967. H. Hutzelmeyer/P. Kramer/T.H. Seligman: [I] Antisymmetric states of the cluster model and their ~-particle parentage. International Conference on Clustering Phenomena in Nuclei. Bochum 1969.

O. Jordan: [1] Trait~ des substitutions et des ~quations alg~briques. Gauthier-Villars, Paris, 1870. L. KalouJnine: [I] Sur les p-groupes de Sylow du groupe sym~trique du degr~ pm. C.R. Acad. Sci. Paris 22__!1(1945),

222-224. (Zbl 61, MR 7) [2] La structure du p-groupe de Sylow du groupe

sym~trique du degr~ p2. C.R. Acad. Sci. Paris ~22 (1946), 1424-1425. (Zbl 61, MR 8) [3] Sur les p-groupes de Sylow du groupe symetrlque du degr~ pm.(suite centrale ascendante et descendante.) C.R. Acad. Sci. Paris 223 (1946), 703-705.

(MR 8)
[4] Sur lee p-groupes de Sylow du groupe symetrlquel du " degr6 pm. (Sous-groupes caracteristiques, sousgroupes parallelotopiques.) C.R. Acad. Sci. Paris 224 (1947), 253-255. (Zbl 30, ~R 8)

185

[5S La structure des p-groupes des Sylow des groupes sym6triques finis. Ann. Sci. Ecole Norm. Sup. (3) 65 (1948), 259-276. (Zbl 34, MR 10) [6] Sur la structure des p-groupes de Sylow des groupes symetrlques finis et de quelques generalizations infinis des oes groupes. Seminaire Bourbaki 1948/1949 (Zbl 100) [7] ~ber sine Verallgemeinerung der p-Sylowgruppen symmetrischer Gruppen. Acta Math. Acad. Sci. Hungar. ~ (1951), 197-221. (Zbl 44, MR 14) A. Kerber: [I] Zur modularen Darstellungstheorie symmetrisoher und alternierender Gruppen. Mitt. math. Sem. Univ. Giessen 6.~8(1966), iii+80 8. (Zbl 139, MR 33) [2] Zur Darstellungstheorie yon Kranzprodukten. Can. J. Math. 2_~0(1968), 665-672. (Zbl 157, MR 58) [3] Zurmodularen Darstellungstheorie symmetrischerund alternierender Gruppen II. Archly d. Math. 19 (1969), 588-594. (Zbl 175, MR 39) [4] Zur Darstellungstheorie yon Symmetrien symmetrischer Gruppen. Mitt. math. Sem. Univ. Giessen 8_~0(1969), 1-27. (Zbl 167, MR 40) [5] On a paper of Fox about a method for calculating the ordinary irreducible characters of symmetric groups. J. Combinatorial Theory ~ (1969), 90-93. (Zbl 165,

MR 38)
[6] Zur Theorie der M-Gruppen. Math. Z. 1!~ (1970), 4-6.

(Zbl 186, MR 41)

[7] Zu einer Arbeit yon J.L. Berggren ~ber ambivalente Gruppen. Pacific J. Math. (Zbl 182) A. Kerber/ [I] On the decomposition numbers of symmetric and M.H. Peel: alternating groups. Mitt. math. Sem. Univ. Giessen 9_!I (1971), 45-81. K. Kondo: [I] Table of characters of the symmetric group of degree 14. Proc. Phys. Math. Soo. Japan (3) 2_~2 (1940), 585-593. (Zbl 23, MR 2)

186

[2] Decomposition of the characters of some groups I. Proc. Phys.-Math. Soc. Japan (3) 23 (1941), 26~-271. (MR 3) P. Kramer/T.H. Seligman: [1] Origin of Regge Symmetry for Wigner Coefficients of SU(2). Z. Physik 219 (1969), 105-113.

M. Kretzschmar:

[I] Gruppentheoretische Untersuchungen zum Schalenmodell I. Die Mathematische Theorie des HamiltonOperators. Z. f. Physik 1~7 (1960), 433-456. [2] Gruppentheoretische Untersuchungen zum Schalenmodell II. Z a m P r o b l e m der Translationsinvarianz. Z. Physik I~8 (1960), 284-303.

D.E. Littlewood:

[I] Invariant theory, tensors and group characters. Phil. Trans. Roy. Soc. London Ser. A 239 (1944) 305-365. (Zbl 60, MR 6) [2] The theory of group characters and matrix representations of groups. Oxford: At the Clarendon Press 1940, 2 nd ed. 1958. [3] The characters and representations of imprimitive groups. Proc. London Math. Soc. (3) (1956), 251-266. (Zbl. 70, MR 17)

A. Loewy:

[1]~ber abstrakt definierte Transmutationssysteme oder Mischgruppen. J. reine angew. Math. I~7 (1927), 239254. [I] Applications of finite groups. New York/London: Academic Press 1959.

J.S. Lomont:

J. Malzan:

[I] Real finite groups. Thesis, University of Toronto 1969. [I] Induced monomial representations, Young elements, and metacyclic groups. Proc. Amer. Math. Soc. 19, 453-458 (1968). (Zbl 155, MR 36)

H.J. Munkholm:

H. Nagao:

[1] On groups with the same table of characters as symmetric groups. J. Inst. Polytechn. Osaka City Univ. Ser. A ~ (1957), I-8. (MR 19)

187

T. Nakayama:

[I] On some modular properties of irreducible representations of a symmetric group I. Jap. J. Math.

17 (1940), 165-184. (Zbl 61, MR 3)

[2] On some modular properties of irreducible representations of symmetric groups II. Jap. J. Math.

1'7 (1941), 411-423. (Zbl 61, MR 3)

E. Netto: [1] Substitutionentheorie und ihre Anwendungen auf die Algebra. Teubner-Verlag, Leipzig 1882.

B. Neumann:

[1] Die Automorphismengruppe

Ann. 107 (1932), 367-386.

der freien Gruppen. Math.

O. Ore: [I] Theory of monomial groups. Trans. Amer. Math. Soc. 51

(1942), 15-64. (Zbl 28, MR 3)

M. Osima: [I] On the representations of the generalized symmetric group. Math. J. Okayama Univ. ~ (1954), 39-56. (Zbl 58, ~ 16) [2] Some remarks on the characters of the symmetric group II. Can. J. Math. ~ (1954), 511-521. (Zbl 58, ]~[R 16) [3] On the representations of the generalized symmetric group II. Lath. J. Okayama Univ. ~ (1956), 81-97. (Zbl 72, MR 18) [4] On the generalized decomposition numbers of the symmetric group. J. ~ath. Soc. Japan 2_~0 (1968), 289-296. (MR 36) [5] On the generalized decomposition numbers of the alternating group. (Japanese) Symposium on Group Theory, Kyoto Univ., 1968, 51-55. T. Oyama: [I] On the groups with the same table of characters as alternating groups. Osaka J. Math. 1 (196&), 91-101. (Zbl 139, MR 29) [I] On the second natural representation of the symmetric groups. Glasgow Math. J. 1_~0(1969), 25-37. (Zbl 175, ]~ 39)

M.H. Peel:

188

[2] Hook representations of the symmetric groups. (to appear in the Glasgow Math. J.) S. Piocard: [1] Sur les bases du groupe sym6trique et les couples f de substitutions qui engendrent un groupe reguller. Mem. Univ. Neuch~tel, Tome I__9(1946).(Zbl 61, MR 8)

[2] Sur les bases du groupe syme~rique II. Paria, Librairie Vuibert, 1948. (Zbl 31, NR 10) [3] Sur les bases d~s~roupes d'ordre fini. Mem. Univ. Neuch~tel, Tome ~ (1957). (Zbl 77, ~ 20) G. P61ya: [I] Kombinatorische Anzahlbestimmtmgen for Gruppen, Graphen und chemische Verbindungen. Acta math. Uppsala 6_~8 (1937), 145-254. B.M. Puttaswamaiah: [I] Group representations (alternating and generalized symmetric groups). Thesis, University of Toronto 1963. [2] Unitary representations of generalized symmetric groups. Can. J. Math. 2!I (1969), 28-38. (Zbl 169, MR 38) B.M. Puttaswamaiah/G. de B. Robinson: [I] Induced representations and alternating groups. Can. J. Math. 16 (1964), 587-601. (MR 29)

A. Radzig: [I] Die Anwendung des Sylow'schen Satzes auf die symmetrische und die alternirende Gruppe. Diss. Berlin 1895. G. de B. Robinson: [I] A geometrical study of the hyperoctahedral group. Proc. Cambridge Philos. Soc. 26 (1930) 94-98. [2] On a conjecture by Nakayama. Trans. Roy. Soc. Canada Sect. III (3) 4_!I(1947), 20-25. (Zbl 29, MR 10) [3] On the disjoint product of irreducible representations of the symmetric group. Can. J. Math. ! (1949), 166-175. (MR 10)

189

[4] Induced representations and invariants. Oan. J. Math. ~ (1950), 334-343. (Zbl 39, MR 12) [5] Representation theory of the symmetric group. University of Toronto Press, Toronto 1961.

[6] Modular representations of 8n. Can. J. Math. 16 (1964), 191-203. (Zbl 117, ~R 28) G. de B. Robinson/0.E. Taulbee: [I] On the modular representations of the symmetric group VI. Proc. Nat. Acad. Sci. USA 41 (1955), 596-598. (Zbl 66, MR 17) A. Scholz: [I] Ein Beitrag zur Zusammensetzung endlicher Gruppen. Math. Z. 32 (1930), 187-189. G.M. Seitz: [I] M-Groups and the supersolvable residual. Math. Z. 110 (1969), 101-122. P.R. Smith: [1] The use of plethysm in the study of configuration of equivalent electrons. M. Sc. thesis, University of Canterbury, New Zealand 1967. P.R. Smith/B.G. Wybourne: [I] Selection rules and the decomposition of the Kronecker square of irreducible representations. J. Math. Phys. ~ (1967), 2434-244e. [2] Plethysm and the theory of complex spectra. J. Math. Phys. ~ (1968), 1040-1051.

W. Specht: [1] Eine Verallgemeinerung der symmetriechen Gruppe. Schriften Berlin I (1932), 1-32. (Zbl 4) [2] Eine Verallgemeinerung der Permutationsgruppen. Math. Z. 37 (1933), 321-341. (Zbl 7) O.E. Taulbee: [I] Modular representations of the symmetric group. Thesis, Michigan State University 1957. A.J. Weir: [I] The 8ylow subgroups of the symmetric groups. Proc. Amer. Math. Soc. 6 (1955), 534-541. (Zbl 65, MR 17)

190

T. Yokonuma: [I] On a property of some generalized symmetric groups. J. ~ac. Sol. Tokyo 12 (1965), 193-211.

(Zbl 139, M~ 32)

A. Young: [I] On quantitative substitutional analysis (Pifth paper) Proc. London Math. Soc. (2) 31 (1930), 273-288. [2] On quantitative substitutional analysis (Sixth paper) Proc. London Math. Soc. (2) 34 (1932), 196-230.

(Zhl 5)
M. Zia-ud-Din: [1] The characters of the symmetric group of order 11!. Proc. London Math. Soc. (2) 39 (1935), 200-204. (Oorrection: Proc. London Math. Soc.

(2), 2 ,

558)

[2] The characters of the symmetric groups of degree 12 and 13. Proc. London Math. Soc. 42, 340-355 (1937). (Zbl 15)

Subject-Index

Alternating
-

group group cycleproduct

11 39 9 79 40 80 20 94 20 76 25 85 94 79 6 6

Generalized

alternating matrix

group

39 179 169 169 39 11 9 25 39 39 22 22 5 5 79 79 79 79 79 79

- , generalized

- Cartan m a t r i x
-

ambivalent associated

decomposition - humbers

a r m of a h o o k - part of the r i m - partition

-

- symmetric group,

group

alternating

- , ambivalent
-

representations Young-diagrams

, basis , complete monomial permuations , hyperoctahedral permutations

axial distance Basis group theorem representations

- of h o r i z o n t a l - of v e r t i c a l - , symmetric Hand hook

-

branching Conjugate corner cycle - , order

- , permutation

(of a hook) of a

(of a hook)

, a r m of a , corner of a of a
a

-s, d i s j o i n t cycle-notation cycleproduct, cyclic Degree associated permutation of a p e r m u t a t i o n group determinantal irr. dictionary disjoint f o r m (of of S n) characters order

7 7 40 6 5 5 77 72 11 7

- , hand
,

leg of

, leglength

of a

79 79 22 22 39 90 90

- , length horizontal - -s, group

of a permutation of group

hyperoctahedral Inertia - group Last letter factor

difference-product cycles

sequence

75 79 79 67

Even permutations Frobenius' theorem

11 86

leg

(of a hook) (of a hook) of J. v N e u m a n n

leglength lemma

192

Littlewood-Richardson-rule Littlewood's conjugates M-group ~urnaghan-Nakayama-formula Nakayama's conjecture theorem of

84. 125 111 80 131 11 6 168 8 8

Selfassociated
-

partition groups

20 20 29 14 71 5 5 39 102

Young-diagram permutation

similar

standard-conjugator standard-tableau symbols symmetric

-

group outer p r o d u c t s ordinary

- , generalized

Odd p e r m u t a t i o n order of a cycle

symmetrized (general)

(of irr. P-component p-element p-regular p-regular p-regular p-singular p-weight partition
-

repr.

of S n) 39

117

symmetry Tableau

component partition (of a diagram)

168 96 8 132 8 20
8

21 6 10 40 90 of G~S n of'the

transposition type
-

of a p e r m u t a t i o n repr.

- an element group

- - - irr. basis

, associated

- of a p e r m u t a t i o n permutation - , cyclic
-

5 6
of a

Vertical

permutation of

- -s, group Wreath

22 22
24

, degree , even

5 11 5 5 29 22 11 10

product symmetric groups

-s w i t h

39 20 20 79 17 21

- group
-

- , degree

of a

Young-diagram - , associated - , r i m of a Young-subgroups Young-tableau , standard Young's natural

-

- -s, s i m i l a r - , horizontal
-

, odd of a

- , type of a - , partition - , vertical -s, group -s, group R-induction r-node r-restriction rim (of a diagram) of h o r i z o n t a l of v e r t i c a l

22 22 22 14.1 141 141 79

form of S n)

71 (of i r r . 7 5 75 75 80

ordinary

repr.

- orthogonal - seminormal Zero-diagram

form form