Idempotents in Group Algebras and Applications to Units
Vrije Universiteit Brussel - Faculty of Science and Bio-engineering Sciences
Group representations: idempotents in group algebras and applications to units
Graduation thesis submitted in fulfillment of the requirements for the degree of Doctor in Sciences
Author: Promotor: Inneke Van Gelder Prof. Dr. Eric Jespers Copromotor: Prof. Dr. Gabriela Olteanu
march 2015
ACKNOWLEDGMENTS
Ik zou graag iedereen willen bedanken die, elk op zijn manier, heeft bijgedragen tot de realisatie van deze doctoraatsthesis. Eerst en vooral zou ik graag mijn promotor Eric Jespers willen bedanken. Hij heeft zijn enthousiasme voor groepsringen op mij overgedragen door zijn interessante cursussen en boeiende lezingen. Door zijn ruime kennissenkring kwam ik in contact met verschillende vriendelijke wiskundigen waar ik veel van heb kunnen leren. I owe a very special thank you to my copromotor Gabriela Olteanu. I would like to thank her for her hospitality during my stays in Romania, for her support, both professional and personal. Thanks for the nice cooperations and the very careful proofreading of our joint works. Also my gratitude is expressed to my co-authors Andreas B¨achle, Mauricio Caicedo, Angel´ del R´ıo,Florian Eisele and Ann Kiefer for the pleasant and enriching collaborations. Thanks to Allen Herman and Alexander Konovalov for the very interesting discussions about GAP and wedderga. Vervolgens zou ik ook graag mijn collega’s Andreas, Ann, Mauricio en Sara willen bedanken voor het nalezen van en hun kritische opmerkingen op een eerste versie van mijn doctoraatsthesis. Graag zou ik alle collega’s van de Vrije Universiteit Brussel, en in het bijzonder Philippe, Sara, Ann, Karen en Timmy, willen bedanken voor de fijne lunches, ontspannende koffiepauzes, toffe babbels, filmavonden, wiskundige nevenactiviteiten en zoveel meer! Verder ben ik ook dank verschuldigd aan het Fonds voor Wetenschappelijk Onderzoek om mij gedurende vier jaar financieel te ondersteunen. Hierbij moet ik ook Stefaan Caenepeel bedanken om mij gedurende een jaar een assistenten- positie aan te bieden in afwachting van een aanstelling door het FWO. Mijn familie en vrienden wil ik bedanken om mij onvoorwaardelijk te steunen. Mijn laatste, maar zeker niet de minste dank gaat uit naar mijn man Giel, om zijn steun en vertrouwen. Ook een heel dikke dankjewel om dit werk van een mooie omslag te voorzien.
Inneke Van Gelder maart 2015
CONTENTS contents i introduction iii summary ix publications xix samenvatting (summary in dutch) xxi list of notations xxxiii
1 preliminaries 1 1.1 Fixed point free groups ...... 2 1.2 Quaternion algebras ...... 3 1.3 Normal bases ...... 3 1.4 Number fields ...... 4 1.5 Crossed products ...... 10 1.6 Group rings ...... 11 1.7 Wedderburn-Artin decomposition ...... 14 1.8 Z-orders ...... 24 1.9 Congruence Subgroup Problem ...... 26 1.10 Finite subgroups of exceptional simple algebras ...... 31 1.11 Cyclotomic units ...... 33 1.12 Bass units ...... 34 1.13 Bicyclic units ...... 36 1.14 Central units ...... 39
2 wedderburn decomposition and idempotents 41 2.1 The Wedderburn decomposition of FG ...... 41 2.2 Primitive idempotents of QG ...... 55 2.3 Primitive idempotents of FG ...... 61 2.4 Conclusions ...... 66
i contents
3 exceptional components 67 3.1 Group algebras with exceptional components of type EC2 . . . 68 3.2 Group algebras with exceptional components of type EC1 . . . 72 3.3 Examples ...... 93
4 central units 101 4.1 Abelian groups ...... 101 4.1.1 A new proof of the Bass-Milnor Theorem ...... 103 4.1.2 A virtual basis of Bass units ...... 107 4.2 Strongly monomial groups ...... 110 4.3 Abelian-by-supersolvable groups ...... 113 4.3.1 Generalizing the Jespers-Parmenter-Sehgal Theorem . . 114 4.3.2 Reducing to a basis of products of Bass units ...... 120 4.4 Another class within the strongly monomial groups ...... 124 4.5 Conclusions ...... 133
5 applications to units of group rings 135 5.1 A subgroup of finite index in U(Z(Cqm o1 Cpn )) ...... 135 5.2 A method to compute U(ZG) up to commensurability . . . . . 137 5.3 Examples ...... 142 + 5.3.1 U(ZD16) up to finite index ...... 142 5.3.2 U(ZSL(2, 5)) up to commensurability ...... 143 5.4 Conclusions ...... 144 bibliography 147 index 155
ii INTRODUCTION
The notion of a group algebra already appeared in a paper of Arthur Cayley from 1854. However, only after the influential works of Richard Brauer (1901- 1977) and Emmy Noether (1882-1935) on representation theory, the subject gained attention because of the correspondence between modules of group alge- bras and group representations. In 1940, Graham Higman posed the following question in his Ph.D. thesis, for finite groups G and H:
Does ZG ' ZH imply that G ' H? This problem is referred to as the (integral) isomorphism problem. It was anticipated for a long time for this conjecture to be true. In 1987, Klaus W. Roggenkamp and Leonard L. Scott showed that this indeed is the case if G is a nilpotent group. It was a surprise when Martin Hertweck gave a counter example to the isomorphism problem in his Ph.D. thesis in 1998. Nowadays, it is still an important problem to decide for which classes of groups the con- jecture does hold. In all these investigations, the unit group U(ZG) of ZG plays a fundamental role. It is essential to consider ZG as a Z-order in the (semisimple) rational group algebra QG and to have a detailed understanding of the Wedderburn decomposition of QG. If one proves the equality of two numbers a and b by showing first that ‘a is less than or equal to b’ and then ‘a is greater than or equal to b’, it is unfair. One should instead show that they are really equal by disclosing the inner ground for their equality — Emmy Noether
The Wedderburn-Artin Theorem states that a semisimple ring R is isomor- phic to a product of finitely many ni-by-ni matrix rings over division rings Di, for some integers ni. However, in the mindset of Emmy Noether, such a classification is unfair. One should instead aim to construct an explicit isomor- phism between R and the product of matrix rings. To do this, a first important step is to calculate the primitive central idempotents e of R to distinguish the different matrix rings. Secondly, one needs to construct elements in each com- ponent Re, which play the role of a complete set of matrix units. In particular, one has to construct a complete set of orthogonal primitive idempotents.
iii introduction
A classical method for obtaining the primitive central idempotents of a semi- simple group algebra FG involves computations using the irreducible charac- ters of G over an algebraic closure of F . However, the known methods to com- pute the character table of a finite group are very time consuming. Therefore, in practical applications, the classical description of primitive central idempo- tents sometimes is of limited use. One would like a character-free description that only depends on the lattice of subgroups and the characteristic of the field, i.e. a description completely within FG. Such a description has been obtained by Aurora Olivieri, Angel´ del R´ıoand Juan Jacobo Sim´onin 2004 for the primitive central idempotents of QG when G is a strongly monomial group, for example an abelian-by-supersolvable group. This method relies on pairs of subgroups (H,K) of G satisfying some conditions which can be checked in- side the rational group algebra QG. Such pairs are called strong Shoda pairs of G. It turns out that each primitive central idempotent is the sum of the distinct conjugates of ε(H,K) (corresponding to a natural idempotent in the rational group algebra Q(H/K)) for a strong Shoda pair (H,K), which we de- note by e(G, H, K). Furthermore, each simple component in the Wedderburn decomposition is a matrix ring over a crossed product of the finite abelian group NG(K)/H over a specific cyclotomic field for some strong Shoda pair (H,K). In 2007, Osnel Broche and Angel´ del R´ıotransfered those results to the case of semisimple finite group algebras FG for strongly monomial groups G. For arbitrary semisimple group algebras FG, it remains an open problem to give a character-free description of the primitive central idempotents and the Wedderburn decomposition of FG. For a rational group algebra QG of a finite nilpotent group G, a complete set of matrix units of an arbitrary simple component QGe(G, H, K) was given, in 2012, by Eric Jespers, Gabriela Olteanu and Angel´ del R´ıo.In joined work with Gabriela Olteanu, we gave a similar result for semisimple finite group algebras FG of nilpotent groups G. Moreover, examples were given to show that the method can not be extended to, for example, finite metacyclic groups. Chapter 1 is a preliminary chapter. In Chapter 2, we first study the primitive central idempotents and the Wedderburn decomposition of group algebras FG with F a number field and G a strongly monomial group (Theorem 2.1.6). This is a generalization of the results of Aurora Olivieri, Angel´ del R´ıo,Juan Jacobo Sim´onand Osnel Broche. Next, we focus on a complete set of matrix units in the Wedderburn components of QG and FG, with F a finite field, for a class of finite strongly monomial groups containing some metacyclic groups (Theorems 2.2.1 and 2.3.4).
iv introduction
I have never done anything ‘useful’. No discovery of mine has made, or is likely to make, directly or indirectly, for good or ill, the least difference to the amenity of the world — Godfrey Harold Hardy
Regardless of his sayings, much of the work of Godfrey Harold Hardy (1877- 1947) did find applications in different branches of science, other than mathe- matics. Hardy was a number theorist and exactly number theory is the elected area in pure mathematics to have many applications to other areas, such as coding theory and internet security. In 1974, Donald Knuth formulated this as follows: “Virtually every theorem in elementary number theory arises in a natural, motivated way in connection with the problem of making compu- ters do high-speed numerical calculations”. Finite group algebras and their Wedderburn decomposition have applications to coding theory as well. Cyclic codes can be realized as ideals of group algebras of cyclic groups and many other important codes appear as ideals of group algebras of non-cyclic groups, see Section 2.3 for references. A concrete realization of the Wedderburn de- composition also allows applications to many other topics, for example to the investigation of the group of automorphisms of group rings, as shown by Au- rora Olivieri, Angel´ del R´ıoand Juan Jacobo Sim´onin 2006. In this thesis, we focus on the applications to the group of units of RG, where R is the ring of integers of a number field F . The main example is the group of units of integral group rings. Only for very few finite non-commutative groups G, a presentation of the group U(ZG) is known. However, Carl Ludwig Siegel, Armand Borel and Harish-Chandra showed, in a much more general setting, that U(RG) is always finitely generated, if G is finite. Therefore, we are satisfied with finding finitely many generators of U(RG), and in particular of U(ZG). If E is a complete collection of primitive central idempotents of FG, then M M RG ⊆ RGe ⊆ F Ge = FG e∈E e∈E
and each F Ge ' Mne (De) for some integers ne and some division rings De. L Since both RG and e∈E RGe are Z-orders in FG, we know that U(RG) is L of finite index in e∈E U(RGe). If we choose an order Oe in each De, then
also GLne (Oe) and U(RGe) have a common subgroup which is of finite index in both. This means that first, we have to find generating sets of units in
GLne (Oe), which is generated (up to finite index) by SLne (Oe) and the matri- ces with diagonal entries in U(Z(Oe)). So, the problem reduces to describing
SLne (Oe) and U(Z(Oe)).
v introduction
In Chapter 3, we classify the finite groups G such that, for a fixed abelian number field F , for all Wedderburn components Mn(D) in the group algebra FG, the corresponding SLn(O), for any Z-order O in D, is generated by the elementary matrices over a two-sided ideal in O (Theorems 3.1.2 and 3.2.21). The components Mn(D) where this is not possible are the so-called exceptional components. This investigation is a generalization of a result from Mauricio Caicedo and Angel´ del R´ıowho dealt with QG. It involves deep results from Hyman Bass, Bernhard Liehl, Leonid N. Vaserˇste˘ınand Tyakal Nanjundiah Venkataramana related to the Congruence Subgroup Problem. In Chapter 4, we study the central units Z(U(ZG)) for finite groups G. Due to Hyman Bass and John Willard Milnor (1966) it is well known that, for a finite abelian group G, the Bass units of the integral group ring ZG generate a subgroup of finite index in U(ZG). We give a new constructive proof of this result (Proposition 4.1.1). For non-abelian groups, some constructions of central units of ZG have been given by Eric Jespers, Guilherme Leal, Michael M. Parmenter, Sudarshan Sehgal and Raul Antonio Ferraz. This was done mainly for finite nilpotent groups G. We construct generalized Bass units and show that they generate a subgroup of finite index in Z(U(ZG)) for finite strongly monomial groups G (Theorem 4.2.3). For a class within the finite abelian-by-supersolvable groups G, we can do more and describe a multiplica- tively independent set (based on Bass units) which generate a subgroup of finite index in Z(U(ZG)) (Theorem 4.3.8). For another class of finite strongly monomial groups containing some metacyclic groups, we construct such a set of multiplicatively independent elements starting from generalized Bass units (Theorem 4.4.4). In Chapter 5, we combine the results of the previous chapters to construct a generating set of U(ZG) up to finite index. This work is a continuation of a result from Eric Jespers, Gabriela Olteanu and Angel´ del R´ıofrom 2012, that described the unit group of ZG up to finite index for finite nilpotent groups G. We also continue works of J¨urgenRitter and Sudarshan Sehgal, and Eric Jespers and Guilherme Leal who showed that under some conditions the Bass units together with the bicyclic units generate a subgroup of finite index in U(ZG). If QG does not contain exceptional components, if one can construct matrix units in each Wedderburn component of QG and moreover, if one knows a generating set of Z(U(ZG)), then it is possible to describe U(ZG) up to finite index. We demonstrate this for metacyclic groups Cqm o1 Cpn , for different prime numbers p and q (Theorem 5.1.1). However, if QG has only exceptional components of type M2(D), then it turns out that SL2(O) can still
vi introduction
be generated by elementary matrices for a special (i.e. left norm Euclidean) Z- order O of D (Proposition 5.2.1). This allows us to construct the group of units of ZG up to finite index for finite groups G, such that QG has only exceptional components of one type and such that one knows non-central idempotents in the non-commutative non-exceptional components of QG (Proposition 5.2.2). Those non-central idempotents are needed to imitate the elementary matrices with (generalized) bicyclic units in ZG.
vii
SUMMARY
In this summary, we present our main results. For the convenience of the reader, Chapter 1 is devoted to a preliminary ex- position on quaternion algebras, number fields, crossed products, group rings, Z-orders, cyclotomic units, Bass units and bicyclic units. In Chapter 2, we give a concrete realization of the Wedderburn decomposi- tion of group algebras FG of finite strongly monomial groups G over number fields F . This description is mainly based on the fact that, for rational group algebras QG of finite strongly monomial groups G, the Wedderburn decompo- sition is completely described using strong Shoda pairs.
Corollary 2.1.7 [8] If G is a finite strongly monomial group and F is a number field, then every primitive central idempotent of FG is of the form eC (G, H, K) for a strong Shoda pair (H,K) of G and C ∈ CF (H/K). Furthermore, for every strong Shoda pair (H,K) of G and every C ∈ CF (H/K),