• Keine Ergebnisse gefunden

Chapter 8. Finite Groups

N/A
N/A
Protected

Academic year: 2021

Aktie "Chapter 8. Finite Groups"

Copied!
6
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

The aim of this chapter is to prove several central results of the theory of finite groups: Theorems of Schur and Zassenhaus and Burnside’s transfer theorem (aslo known as Burnside’s normal �-complement theorem).

Throughout this chapter, unless otherwise stated, G denotes a finite group in multiplicative notation.

References:

[Bro94]

K. S. B����,Cohomology of groups, Graduate Texts in Mathematics, vol. 87, Springer-Verlag,

New York, 1994.

[Rot09] J. J.

R�����,An introduction to the theory of groups. Fourth ed., Graduate Texts in Mathema-

tics, vol. 148, Springer-Verlag, New York, 1995.

23 Cohomology of Finite Groups

To begin with, we collect in this section a few general results about the cohomology of finite groups.

Lemma 23.1

If G is a finite group, then |G| ¨ H

pG� Mq “

0

for every

•1.

Proof : Let 1 denote the trivial group. Because 1 is a cyclic group of order one Theorem 16.2 yields Hp1�Mq – 0 if • 1 (whereas H0p1�Mq – M). Now, the composition of the restriction with the transfer

HpG�Mq resG1 // Hloooomoooonp1�Mq

–0if�•1 trG1

//HpG�Mq

equals multiplication by the index of|G:1| “ |G|by Proposition 21.3 and factors through0if•1by the above. Therefore multiplication by|G|is zero inHpG�Mqif•1.

Proposition 23.2

If G is a finite group and M is a finitely generated

ZG-module then

H

0

pG� Mq is a finitely generated abelian group and H

pG� Mq is a finite abelian group of exponent dividing |G| for all

•1.

Proof : FixPZ•0.

Claim 1: HpG�Mqis a finitely generated abelian group.

79

(2)

Indeed: Using the fact that ZG is a noetherian ring as G is finite, we may construct a projectiveZG- resolutionP of Z in which all the modules are finitely generated abelian groups. Now, applying the functor HomZGp´�Mq to P we again obtain complexes of finitely generated abelian groups since for each •0, HomZGpP�Mq “HomZpP�MqG ÑHomZpP�Mq, which is a finitely generated abelian group if bothP andM are. The cohomology groups of this complex is again finitely generated.

Claim 2: HpG�Mqis a finite group if•1.

Indeed: SinceHpG�Mqis a finitely generated abelian group by the first claim and|G| ¨HpG�Mq “0 by Lemma 23.1,HpG�Mqmust be torsion, hence finite.

Exercise [Exercise 1, Exercise Sheet 12]

If M is a

ZG-module which is induced from the trivial subgroup, then

H

pG� Mq “

0

for all

•1.

Deduce that H

pG� Mq “

0

for all

•1

if M is a projective

ZG-module.

Exercise [Exercise 2, Exercise Sheet 12]

Let be a prime number, let G be a finite group of order divisible by �, and let P be a Sylow

�-subgroup of G. If M is an

F

G-module, then the restriction map

resGP :

H

pG� Mq

Ñ H

pP

ResGP

pMqq is injective for all

•0.

24 The Theorems of Schur and Zassenhaus

In this section we prove two main results of the theory of finite groups, which are often considered as one Theorem and called the

Schur-Zassenhaus Theorem. Beacause of the methods we have developed,

we differentiate between the abelian and the non-abelian case.

Theorem 24.1 (Schur, 1904)

Let G be a finite group and let A “ pA� ¨� ˚q be a

Z

G-module such that there exists P

Z•1

with

1

for all P A. Suppose that

`

|G|� �

˘

1. Then the following hold:

(a) Every group extension

1 //

A

//

E

//

G

//1

inducing the given G-action on A splits.

(b) Any two complements of A in E are E-conjugate.

Proof : We prove that theHpG�Aqis trivial for all•1. For convenience, writeAadditively in this proof.

Thus by assumption we have¨A“0. By Lemma 23.1, we know that|G| ¨HpG�Aq “0for all•1.

Since¨A“0, we also have¨CpG�Aq “0, and thus¨HpG�Aq “0. Now, by the Bézout identity there exists�� �PZsuch that

¨ |G| `¨“1 and hence

HpG�Aq “�¨ |G| ¨loooooooomoooooooonHpG�Aqq

“0

`�¨loooooomoooooon¨HpG�Aq

“0

“0

(3)

Since H2pG�Aq vanishes any extension splits and since H1pG�Aq vanishes all complements of A are E-conjugate, by Theorem 19.3 and Theorem 18.3(b) respectively.

Theorem 24.2 (Zassenhaus, 1937 )

Let

1 //

A

//

E

//

G

//1

be an extension of finite groups (where A is not necessarily abelian). If

`

|A|� |G|

˘

1, then the extension splits.

Proof : W.l.o.g. we may assume that |G|•1. Then we proceed by induction on the size ofA.

¨ If|A| “1or|A|is prime, thenAis abelian. Moreover,|A| “1for allPA. Thus Schur’s Theorem applies and yields the result.

¨ Suppose now that |A| P Z•2zP. Let P P be a prime number dividing |A|, let P be a Sylow

�-subgroup ofA, and setN:“NEpPqfor the normaliser ofP inE.

Claim 1: EAN.

Indeed, ifPE, thenP and�P�´1 are Sylow�-subgroups of A, henceA-conjugate, so that there exists PA such that�P�´1�P�´1. Thus `

´1�qP`

´1˘´1

P, i.e. ´1PNEpPq “N, and therefore`

´1˘ PAN.

Claim 2: Ahas a complement inE.

We split the proof of this claim in two cases:

Case 1: NE.

In this case, restrictingtoN yields the group extension

1 //AXN //N �|N //G //1

whereAXNàAbecauseGN{pAXNq –AN{AE{A. Thus, by the induction hypothesis, this extension splits. Hence, by Proposition 17.6, there exists a complementH ofAXN in N. Since

|H| “ |G|, we have HXA“ t1u, and thereforeH is a complement ofAinE.

Case 2: NE.

LetZ :“ZpPqbe the centre ofP, which is a non-trivial subgroup ofP becauseP is a�-group.

SinceZ is characteristic inP (i.e. invariant under all automorphisms ofP), and sinceP is normal inN, we deduce thatZ is normal in E. Thus, by the universal property of the quotient,�induces a group homomorphism:E{Z›ÑG� �Z ›Ñ�p�q, whose kernel isA{Z. In other words, there is a group extension of the form

1 //A{Z //E{Z //G //1�

Now |Z| ‰ 1implies that |A{Z| †|A|, hence, by the induction hypothesis again, this extension splits. So letF be a complement ofA{Z inE{Z. By the Correspondence Theorem, there exists a subgroupFr §E containingZ such thatFFr{Z. In other words, there is a group extension

1 //Z //Fr //F //1 SinceFGandZ §P§A, we have`

|Z|�|F|˘

“1and therefore this extension splits by Schur’s Theorem. Thus, there is a complementH of Z inFr. But|H| “ |G| impliesHXA“1, so thatH is also a complement ofA inE. The second claim is proved.

We conclude that the extension splits using Proposition 17.6 .

Remark 24.3

Notice that both Schur’s and Zassenhaus’ Theorems can be stated in easy terms not involving

cohomology, but their proofs rely on cohomological methods.

(4)

25 Burnside’s Transfer Theorem

Throughout this section, we let H be a subgroup of G of index |G

:

H | “: and A be a

trivial Z

G- module. Our first aim is to understand the action of the transfer homomorphism on H

1

pG� Aq. So first recall that H

1

pG� Aq “ Z

1

pG� Aq “

HomGrp

pG� Aq by Example 10, and hence we see the transfer as a homomorphism

trGH : HomGrp

pH� Aq

Ñ

HomGrp

pG� Aq Lemma 25.1

Let P

Z•0

and let

ZG�`1

be the �-th term of the bar resolution of

Z

as a

ZG-module. View it as

a projective resolution of

Z

as a

Z

F -module by restriction. Fix a right transversal S “ t�

1

� � � � � �

u of H in G. Then the comparison maps between this and the bar resolution of

Z

as a

Z

H-module are given by the canonical inclusion

:ZH�`1

ÑZG

�`1

and by the map

:

Z

G

�`1

Ñ

Z

H

�`1

p�

0

� � � � � �

q

ބ

p�

0

� � � � � �

q , where, for every

§

�,

for some

P H and some

P S.

Proof : Using the definition of the differential maps of the bar resolution, we see that there are commutative diagrams

ZH�`1

✏✏

//

ö

ZG�`1

✏✏

ZH

ε

✏✏

0 //

ö

ZG

ε

✏✏ZH �´1 //ZG Z Id //Z

and ZG�`1

✏✏

//

ö

ZH�`1

✏✏

ZG

ε

✏✏

0

//

ö

ZH

ε

✏✏ZG �´1 //ZH Z Id //Z

(where•1). Thus the Comparison Theorem yields the result.

Proposition 25.2

Fix a right transversal S “ t�

1

� � � � � �

u of H in G. Then the transfer for H

1

is described as follows:

trGH : HomGrp

pH Aq

Ñ

HomGrp

pG� Aq

fi›

Ñ

˜

trGH

p�q

:

fi›

Ñ

ÿ

�“1

p�

q

¸

where

with

P H for all

§

�.

(5)

Proof : On the one hand

HomGrppH�Aq “H1pH�Aq –H1`

HomZHpZH2�Aq˘

via the bar resolution. On the other hand,

HomGrppH�Aq “H1pH�Aq –H1`

HomZHpZG2�Aq˘

via the bar resolution forGrestricted toH. Now, transfer is defined using the second resolution, therefore, we need to compare these two resolutions. But

H1pH�Aq “Z1pH�Aq�

becauseB1pH�Aq “0sinceH acts trivially onA, and

Z1pH�AqÑC1pH�Aq –HomZH`

ZH2�A˘

If for a given PHomGrppH�Aq,˜ denotes the image of inHomZH`

ZH2�A˘

, then forPH set

˜ :ZH2 ›ÑA p1� �q “ r�sfi›Ñp�q and thus for each PH,

˜`

p�� ��q˘

“˜`

¨ p1� �q˘

¨p�q “p�q

becauseH acts trivially on A, and we extend this map by Z-linearity to the whole of ZH2. Using the comparison map1:ZG2›ÑZH2 of the previous lemma yields˜˝1:ZG2›ÑA, which isZH-linear.

Now, computing the transfer using its definition yields for everyPZG2: trGH˝1˘

p�q “ÿ

�“1´1˝1˘

p��q ““ÿ

�“1

˝1˘ p��q because ´11 � � � � � �´1 (

is a left transversal ofHinGandAis trivial. We want to view this as a1-cocycle forG, that is evaluate this on an elementr�s “ p1� �q PG2ÄZG2 :

trGHp�qp�q “trGH˝1˘

p1� �q “ÿ

�“1

˝1˘

p�� ��q “ÿ

�“1

˜` 1� �˘ because“1¨ and σp�q. So we obtain

trGHp�qp�q “ÿ

�“1

˜p1� �q “ ÿ

�“1p�q as required.

Lemma 25.3 (Choice of transversal for a fixed

P

G

) Fix P G.

(a) There exists a right transversal of H in G of the form

S

1

� �

1

�� � � � � �

1

1´1

� �

2

� �

2

�� � � � � �

2

2´1

� � � � � �

� �

�� � � � � �

´1(

with

1

` � � � `

“ |G

:

H| “ and

´1

P H for all

§

�.

(b) If P

HomGrp

pH� Aq and P G, then

trGH

p�qp�q “

�“1

`

´1˘

.

(6)

Proof :

(a) The element acts on the right on right cosetsH�, viaH�fi›ÑH��. Decompose the set of right cosets into�-orbits. Let�be the number of�-orbits and letH�1� � � � �H�be representatives of the

�-orbits. We get all the right cosets of H by applying powers of to each H� and we suppose that H�´1H� (that is, is the cardinality of the orbit). With this choice, we obtain a right transversal with the required properties.

(b) Proposition 25.2 together withpaqyield for1§§�:

�� �2� � � � � �´1 belong to the right transversalS, so that`

˘

¨“1¨`

�`1˘

PH¨S for0§ §´2and

`´1˘

“`

´1˘

PH¨S. Therefore trGHp�qp�q “ÿ

�“1`

´1˘

because all the other elements ofH appearing are 1and �p1q “0.

Theorem 25.4 (Burnside’s transfer theorem (or Burnside’s normal p-complement theorem), 1911) Let G be a finite group, let be a prime number such that | |G| and let P be a Sylow �-subgroup of G. If P is abelian and C

G

pPq “ N

G

pPq, then there exists a normal complement N to P in G, i.e.

GN ¸ P .

Proof :

Claim: If there existPG andPP such that�� ���´1PP, then���´1�.

Indeed: using the assumptions, we haveP´1P�, which is abelian, and therefore bothP and´1P�

are Sylow �-subgroups of CGp�q. Thus P and ´1P� are conjugate in CGp�q, so that there exists

PCGp�qsuch that�P�´1´1P�, that is��Pp��q´1P and hence��PNGpPq “CGpPq. Finally,

���´1“ p��q�p��q´1becausePP, as required.

Now consider the identity mapIdP PHomGrppP�Pqand trGPpIdPq PHomGrppG�Pq. The previous lemma yields for a fixedPP,

trGPpIdPqp�q “π

�“1IdP`

´1

looomooon

PP

˘

Now using the claim yields´1 for each1§§�, hence trGPpIdPqp�q “π

�“1|G:P|

In particular, this proves that trGPpIdPq: GP is a surjective group homomorphism, because for each

PP, there exists�PP such that|G:P| by the Bézout identity. (Indeed, Bézout implies that there exist�� �PZsuch that�|P| `�|G:P| “1, hence1“ p�q|G:P| and we choose.)

Finally, setN:“ker´

trGPpIdq¯

, so that we have a group extension

1 //N //GtrGPpIdPq//P //1

with a section given by |G:P|1 ¨�, where : P ›Ñ G is the canonical inclusion. It follows that N is a normal complement ofP inG.

Referenzen

ÄHNLICHE DOKUMENTE

In the last Section, we use his mass formula and obtain a list of all one- class genera of parahoric families in exceptional groups over number fields.. Then k is

Motivated by recent results on the minimal base of a permutation group, we introduce a new local invariant attached to arbitrary finite groups.. Building on results of Halasi–Maróti,

(In fact, Brou´ e’s abelian defect group conjecture, and hence the blockwise Alperin weight conjecture, holds for principal 3-blocks with elementary abelian defect groups of order

Remark: It is known that G-sets are, in a certain way, analogous to representations of the group G: Every G-set U canonically defines a permutation representation of G on the

Most proofs of this fact can actually be used to prove Theorem 1 in all its generality, so there is not much need to provide another proof here.. But yet, let us sketch a proof

Primitive central idempotents of finite group rings of symmetric and alternating groups in characteristic 2.. Harald Meyer August

Primitive central idempotents of finite group rings of symmetric and alternating groups in characteristic 3.. Harald Meyer August

We discuss various (still open) questions on approximation of infinite discrete groups.. We focus on finite-dimensional approximations such as residual finiteness