Group rings of finite groups over p-adic integers (Q792445)

Who ever has tried to determine the structure of blocks or projective modules of group rings of a finite group - modularly or p-adically - knows about the great difficulties that arise. He therefore will appreciate very much the author's book - an extension of his Habilitationsschrift in Aachen (1980; Zbl 0502.20002) - which gives a first systematic approach to describe p-adic blocks of finite groups provided the decomposition numbers are zero or one. In Ch. V-VII he gives an explicit ringtheoretic description of the principal blocks - p- adically - of the following groups: 1) Blocks with cyclic defect - not necessarily principal - (among others he proves a generalization of Brauer's theorem of defect 1 to arbitrary cyclic defect). 2) \(G=SL(2,q)\) q an odd prime power at \(p=2.\) 3) Blocks with multiplicity one are almost described - in particular they are completely described if the Brauer characters are all real. For many other classes of groups the projection Be of a p-adic block B with respect to a central primitive idempotent e in the rational group algebra are described. These include 1) blocks with multiplicity one; 2) PSL(2,q) \(q\equiv 3(8)\) at 2; 3) SL(2,q) \(q\equiv 3(8)\) at 2; 4) \(S_{10}\) at 5; 5) PSL(3,3) at 5; 6) PSL(3,4) at 3; 7) \(M_{11}\) at 2 and 3 (always the principal block in 2)-7)). The general pattern is as follows: Let B be a block of \({\hat {\mathbb{Z}}}_ pG\), G a finite group, and denote by \(e_ 1,...,e_ n\) a complete set of central orthogonal idempotents in \({\hat {\mathbb{Q}}}_ pB\), and let \(P_ i\) 1\(\leq i\leq t\) be a complete set of indecomposable projective B- lattices. Step 1: Determine \(Be_ j\), 1\(\leq j\leq h\). It turns out that these orders become graduated orders (cf. below) under an unramified extension R of \({\hat {\mathbb{Z}}}_ p\), provided the decomposition numbers are 0 or 1. Step 2: Determine the upper (lower) amalgamating factors \[ u(P_ i)=\oplus^{n}_{j=1}P_ i\epsilon_ j/P_ i\quad(\ell(P_ i)=P_ i/(\oplus^{n}_{j=1}P_ i\epsilon_ j\cap P_ i)). \] After these steps B is determined; however, to carry them through in special examples requires deep insight into the characters, Brauer characters etc., which the author presents masterly. In Ch. II-IV the general theory is developed to carry through the above two steps in special examples. Let R be a complete local Dedekind domain with maximal ideal \({\mathfrak p}\) and field of fractions K. D is a finite dimensional skewfield over K, \(\Omega\) the maximal R-order in D, and \(\Lambda\) an R-order in a separable K-algebra A. \(\Lambda\) is called graduated (a more common name is tiled) if there are orthogonal primitive (in \(\Lambda)\) idempotents \(\epsilon_ 1...\epsilon_ t\), \(1=\sum^{t}_{i=1}\epsilon_ i\) with \(\epsilon_ i\Lambda \epsilon_ i\) maximal. \(\Lambda\) is called graduable if \(S\otimes_ R\Lambda\) is graduated for some unramified extension S of R. Such a graduated order in a simple A is up to isomorphism described by \(\tilde n=(n_ 1,...,n_ t),\quad n_ i\in {\mathbb{N}},\quad \sum n_ i=n,\quad M=(m_{ij})_{t\times t},\quad m_{ij}\in {\mathbb{Z}}\) as \[ \Lambda(\Omega,\tilde n,M)=\{(a_{ij}):\quad a_{ij}\in(rad \Omega^{m_{ij}})_{n_ i\times n_ j},\quad 1\leq i,j\leq t\}\subset(D)_ n=A \] by the structural invariants \(m_{ijk}=m_{ij}+m_{jk}-m_{ik},\quad 1\leq i,j,k\leq t.\) The graduable orders can be characterized internally as follows: Theorem II, 15: If \(\Lambda\) is an R-order in a simple A, then the following are equivalent: (i) \(\Lambda\) is graduable; (ii) the twosided \(\Lambda\)-ideals form a distributive lattice under \(+\), \(\cap\); (iii) For any irreducible \(\Lambda\)-lattice L with \(L\Omega =L\), the \(\Lambda\) composition factors have multiplicity at most 1 in a composition series of L/L rad \(\Omega\). (iv) If \(\Sigma\) is a maximal separable suborder of \(\Lambda\), then the full \(\Sigma\)-\(\Sigma\) sublattices of \(\Lambda\) form a distributive lattice under \(+\), \(\cap.\) The characterization of graduable orders is given by ''Galois descent''. Let \(\Lambda\) be graduable. Then the isomorphism type of \(\Lambda\) is uniquely determined by the type of \(ScR\Lambda\), S unramified and by the embedding \(\Lambda /rad \Lambda \to S\otimes_ R\Lambda /rad S\otimes_ R\Lambda.\) Let now \(\Lambda\) be an arbitrary R-order in A. Then a graduated hull \(\Gamma\) of \(\Lambda\) is a graduated overorder which is minimal w.r.t. \(\subset\). Then \(\Lambda\) in a simple A has a unique graduated hull if and only if the lattice of \(\Lambda\)-\(\Omega\) lattices in a simple A- module is distributive. In Ch. III the author turns to self-dual orders w.r.t. a generalized trace - these orders include group rings and some twisted group rings. He generalizes Jacobinski's formula of the conductor from a hereditary overorder \(\Gamma\) of the self-dual order \(\Lambda\) to the case where \(\Lambda\) is graduated. Next the author turns to the important question (cf. step 1): When is \(\Gamma =\prod^{h}_{i=1}\Lambda e_ i\) graduable, w.r.t. \(\{e_ i\}_{1\leq i\leq h},\) a complete set of primitive orthogonal idempotents of \(K\Lambda\). For this he defines the decomposition numbers: \(\Lambda \subset \oplus^{h}_{i=1}\Lambda e_ i\subset \oplus^{h}_{i=1}\Delta_ i\) where \(\Delta_ i\) a maximal order with \(\tilde S_ i\) the simple \(\Delta_ i\)-module. If \(S_ j\), 1\(\leq j\leq r\), are the simple \(\Lambda\)-modules, then \(d_{ji}\) is the multiplicity of \(S_ i\) in \(\hat S_ j|_{\Lambda}\). Then the author derives a Brauer-reciprocity: \[ f_ jd'\!_{ji}=d_{ji}\dim_{R/{\mathfrak p}} End_{\Lambda}(S_ i), \] where \(P_ i\) is the projective cover of \(S_ i\), and if \(V_ j\), 1\(\leq j\leq s\), are the simple \(K\Lambda\)-modules, then d'\({}_{ji}\) is the multiplicity of \(V_ j\) in \(KP_ i\), and \(f_ j=\dim_{R/{\mathfrak p}}\Omega_ i/rad \Omega_ i.\) (III, 11). He then proves that \(\Lambda e_ j\) is graduable if and only if \(d_{ji}\leq 1\) and the centre of \(\Lambda e_ j\) is maximal (III, 12). Having come this far, the author has to compute the structural invariants for \(\Lambda e_ i\) in step 1 and compute the amalgamation factors for the various blocks under consideration. The main ingredients for this are developed in the crucial but rather technical Ch. IV, where the author carefully analyses how the structural invariants and the amalgamation factors are built and how they behave under the various automorphisms and anti-automorphisms. This information is in most of the applications enough to determine the graduated hull of blocks of group rings. The book is of high standard, and is also carefully written. It is a valuable contribution to the p-adic representation theory of finite groups.