JOURNAL OF PURE AND APPLIED ALGEBRA ELSEVIER
Journal of Pure and Applied Algebra
138 (1999)
39-44
Ranks of K, and G, of orders and group rings of finite groups over integers in number fields Aderemi 0. Kuku International Centre for Theoretical Physics, I-34100 Trieste. Italy Communicated
by CA.
Weibel; received
10 June 1997; received in revised form 28 October
1997
Abstract Let R be the ring of integers in a number field F, A any R-order in a semi-simple Falgebra C, r any maximal R-order containing A. We show in this paper that for all n 2 2 rank&(n) = rank G,(n) = rank K,(T) = rank K,(C). Hence if G is a finite group, rank K,(RG) = rank Gn(RG) = rankK,,(FG). @ 1999 Elsevier Science B.V. All rights reserved. AMS
clussifiicatiun:
19D50; 19F27
Introduction In this paper, we continue our study of higher K-theory of finite groups over integers in number fields.
of orders and group rings
In earlier papers, [7, 10-121, the author had proved that if R is the ring of integers in a number field F, and LI any R-order in a semi-simple F-algebra C, then &(A), G,(n) are finitely generated Abelian groups for all n 2 1. We now obtain information about the ranks of these groups and show precisely that if r is a maximal order containing A, then rank&(n) = rank G,(n) = rank K,,(T) = rankK,(C) for all II 22. It then follows that if /I = ZG, then rankK,,(ZG) = rankG,(ZG)=rankK,(T)=rankK,(QG). Notation. For any ring A, we write K,,(A) for the Quillen K-groups n,+l(BQP(A)) = z,(BGL(A)+) where P(A) is the category of finitely generated projective A-modules. If A is Noetherian, we write G,,(A) for rcn,+l(BQM_(A)) where M_(A) is the category of finitely generated A-modules. If R is the ring of integers in a number field F, and n is an R-order in a semi-simple F-algebra C, we write SK,(,4)=Ker(ZC,,(ci)--+K,(C)).
0022-4049/99/$ - see front matter @ 1999 Elsevier Science B.V. All rights reserved PII: SOO22-4049(98)00010-3
A. 0. Kuku I Journul of Pure and Applied Algebra 138 (I 999) 39-44
40
Theorem 1.1. Let R be the ring of integers in a number field F, A any R-order in a semi-simple F-algebra
C, r a maximal R-order containing A. Then for all n 2 2
rankK,(~)=rankK,(T)=rankG,(/i)=rankK,(C). The proof of Theorem 1.l will be in several steps 1.2-1.5 below. Theorem 1.2. Let R be the ring of integers in a number jield F, r a maximal Rorder in a semi-simple F-algebra C. Then the canonical map K,,(T)+ kernel and torsion cokernel for all n 2 2. Hence, rank K,,(f)
K,(C) hasJinite
= rank K,,(C).
Proof. Since r, .Z are regular, we have K,(T)? G,(T) and K,(C)? G,(C). So, we show that G,(T) t G,(C) has finite kernel and torsion cokemel. Now, SG,(T) = SK,(T) is finite for all n > 1 (being finitely generated and torsion see [6,9, lo]). Also the localisation sequence of Quillen yields . .
-Gn+1(r)--tGn+,(c)-t~Gn(r/pr)--t ..‘2
where p runs through the prime ideals of R. Now, for n > 1, each G,,(T/pT) is finite since F/pT) is finite (see [7]). So, $,G,(T/pT) is torsion. Hence, G,+t(Z)/lm (G,+t (f) is torsion, as required.
0
Lemma 1.3 (Serre). Let A +B EEK --) C @ L+ D be an exact sequence of Abelian groups. If A, B, C and D are finite (resp. torsion), then the kernel and cokernel of K -+ L are both finite (resp. torsion).
Theorem 1.4. Let R be the ring of integers in a number field F, A any R-order in a semi-simple F-algebra, r a maximal order containing A. Then, for all n > 1, the map G,,(T)-+G,,(A)
induced by the functor M_(T)-+M_(A) given by restriction of scalars
has finite kernel and cokernel. Hence,
rank G,(T) = rank K,,(T) = rank G,(A).
Proof. There exists a non-zero element s E R such that A c r c A( l/s). Let S = {s’}, i 2 0. Then As = A @‘R Rs ‘v r aR Rs = Ts. We show that for all n > 1, c(, : G,(T) + G,(A) has finite kernel and cokemel. Consider the following commutative diagram of exact sequences: . . . -+
Gn(rjsr)-3 G,(r)5 G,(rs)-
cn-l(r/sr)
-
..
(1) -
G,,(A/sA) A
G,(A) L
where 6 is an isomorphism.
G,(As) i
G,_l(A/sA)
-.
..
41
A. 0. Kuku/ Journal of Pure and Applied Algebru 138 11999) 39-44
From (I) we extract the Mayer-Vietoris
Now, since T/ST and A/s/i finite [7] except Gn(T)
sequence
are finite and it 2 1 all the groups in (II) above are
and G,,(A). The result now follows from Lemma
q
1.3.
Theorem 1.5. Let R be the ring of integers in u number field F, A any R-order in u semisimple F-algebra, K,,(A)--+Kn(T)
T a maximal order containing A. Then for all n 2 1 the map
has finite kernel and cokernel. Hence rank K,,(A)=rankK,,(T).
To be able to prove Theorem
1.5, we first prove the following:
Theorem 1.6, Let R be the ring of integers in a number field F, A any R-order in a semisimple F-algebra
C, T a maximal order containing A. Then for all n 2 1 the
map K,,(A) --) K,,(T) (induced by the inclusion map A H r),
is un isomorphism mod
torsion.
Proof. First note that since every R-order is a B-order, there exists a non-zero s such that A c r c A(l/s). A
Put q - = sr. Then we have a Cartesian
-+r
I A/q --
(1)
1 r/q -
Now, by tensoring
(I) with Z(l/s),
if we write A(l/s)
group A, we have long exact Mayer-Vietoris
.
integer
square
sequence
for A@ Z(l/s)
for any Abelian
(see [2] or [18])
..K~+.(,,)(~)-K.o(j)‘K,(r)(~)@K&/g)($) :K,(r,q)
(-!)
-K,-,(A)
(-!)
+ ....
(II)
Now A/q and r/q are finite rings and so Kn(A/q) and K,(T/q) (see [l 1, 121). The Result is now immediate from Lemma 1.3. 0
Proof of 1.5. Let CL,: K,,(A)+Kn(T)
denote the map. By Theorem
are finite groups
1.6, the kernel and
cokemel of c(, are torsion. Also, for all IZ > 1, Kn(A) and K,(T) are finitely generated (see [ 11, 2.11). Hence, the kernel and cokemel of c(,, are finitely generated, hence finite. So rank K,(T) = rank Kn(A). As a fall out from the above, we now have the following that SK,(A)
is finite for any R-order A.
Cl
result which also proves
42
A.O. Kukul Journal of Pure and Applied Algebra
138 (1999)
39-44
Theorem 1.7. Let R be the ring of integers in a number field F, A any R-order in a semi-simple F-algebra C. Then the canonical map K,,(A) -+K,(.Z)
has jnite kernel
and torsion cokernel.
Proof. From the commutative diagram
we have an exact sequence K,(r) InWA)
0 -+kercc-+SK,(/i)-+SK,(T)-+ K,(C) K,(z) +o * ImK,(/1) 3 Im(K,(T) ’
Now, by Theorem 1.5 kercr is finite and by [S], SK,(T) =SG,,(T) is finite for all 12> 1. Hence, from the exact sequence (I) above, SK,(A) is finite. Also, by Theorem 1.5, K,,(r)/ImK,,(A) is finite, and by Theorem 1.2, K,(C)/Im(KJ) is torsion. Hence the result. 0 Remarks 1.8. (i) The above results hold for ,4 = RG where G is any finite group (ii) The ranks of K,(R) and K,(F) are well known and are due to Bore1 (see [I] or [41X More precisely, let ri be the number of embeddings of F in Iw and r2 the number of distinct conjugate pairs of embeddings of F in @ with image not contained in Iw. Then 1 CC
rank K,(F) =
0 rl +rz r2
1 q+r2--1
rank K,(R) =
rl + r2 r2
0
if if if if if
n =0, n= 1, n=2k
k > 0,
n=4k+
1,
n=4k+3,
if if if if if
n = 0, n=l, n=4k+ 1, n=4k+3, n=2k k > 0.
It then means that if Z is a direct product of matrix algebras over fields and f is a maximal order in C, then m&K,(T) = rank K,(C) is completely determined since C=17M,,(Fi) and f = nM,,(Ri) where Ri is the ring of integers in Fi. Also, by Theorem 1.2, this is equal to rank G,(n) as well as rank K,(A) if /1 is any R-order contained in r.
43
A. 0. Kuku I Journal of Pure und Applied Algebra 138 (I 999) 39-44
However,
E of F which splits
if C does not split, there exists a Galois extension
C, in which case we can reduce the problem
of ranks of K, of
to that of computation
fields. (iii) In [5], Jahren proves that if G is a finite group with r irreducible sentations,
c of them of complex r
if
n-_ l(4),
if
n~3(4),
{ 0
if
n is even.
(iv) Even though p-adic
orders
cokemel
type, then for all n > 1, we have
c
rankK,(ZG)=
we do not have finite generation
A, the following
real repre-
results,
Theorem
results
(1.8),
for K,(A),
concerning
G,(A),
finite kernel
of and
are quite interesting.
Theorem 1.8. Let R be the ring of integers in a p-adic field F, A any R-order in a semi-simple F-algebra
C, r a maximal R-order containing A. Then, for all n > 2,
(i) the canonical map K,,(T)+
K,,(C) has jinite kernel and cokernel,
(ii) the canonical map G,(A)-+ G,(C) has jnite kernel and cokernel, (iii) a, : G,,(T) + G,,(A) has jinite kernel and cokernel where c(,, is the map induced by the functor M_(T)-+M_(A) given by restriction of scalars. Proof.
Since r and C are regular,
(i) is a special case of (ii) and (iii) follows from
both (i) and (ii) as the map in (i) factors as K,(T)=G,(T)*G,(A)+G,(C)=K,(C). So, it suffices to prove (ii). Now, C = A[ l/p] and so, there is an exact sequence G,(A/PA>-G,(/~)~G,(C)-,G,-I(AIPA)
Since n > 2, and A/pA kernel and cokemel
is finite, the end terms are finite. Now by Lemma
of G,,(A) + G,(Z)
are finite.
1.3, the
0
Acknowledgements I would like to thank Chuck Weibel for useful conversations Science Foundation
and the Swiss National
for support during a short visit to the University
part of this work was done. I also thank the Institute
of Lausamre where
of Mathematics
at Lausanne
for
hospitality. References [I] A. Borel, Stable real cohomology of arithmetic groups, Ann. Sci. Ecole. Norm. Sup. (4), Serie 7 (1974) 235-272. [2] R. Charney, A note on excision in K-theory, Proc. Bielefeld Conf., Springer Lecture Notes, vol. 1046, 1984, 49-54.
A.O. Kukul Journal of Pure and Applied Algebra 138 (1999)
44
3944
[3] A. Dress, A.O. Kuku, The Cartan map for equivariant higher K-groups, Commun. Algebra 9 (7) (1981) 727-746. [4] D. Grayson, On the K-theory of fields, in: Algebraic K-Theory and Algebraic Number Theory, Contemp. Math. 83 (1989) 31-55. [5] B. Jahren, On the rational K-theory of group-rings of finite groups, preprint. [6] A.O. Kuku, Some finiteness theorems in the K-theory of orders in p-adic algebras, J. London Math. Sot. 2 (13) (1976) 122-128. [7] A.O. Kuku, SK, of Orders and G, of Finite Rings, Springer Lecture Notes, vol. 551, Springer, Berlin, 1976, pp. 60-68. [8] A.O. Kuku, SG, of orders and group-rings, Math. Zeit. 165 (1979) 291-295. [9] A.O. Kuku, Higher algebraic K-theory of group-rings and orders in algebras over number fields, Commun. Algebra 10 (8) (1982) 805-816. [IO] A.O. Kuku, K-theory of group-rings of finite groups over maximal orders in division algebras, J. Algebra 91 (1) (1984) 18-31. [l l] A.O. Kuku, K,, SK, of integral group-rings and orders, Contemp. Math. 55 (1986) 333-338. [12] A.O. Kuku, Some finiteness results in the higher K-theory of orders and group-rings, Topology Appl. 25 (1987) 185-191. [13] R. Oliver, Whitehead Groups of Finite Groups, Cambridge University Press, Cambridge 1988. [14] D.G. Quillen, Higher Algebraic Theory I, Springer Lecture Notes, vol. 341, Springer, Berlin, 1973, pp. 77-139.
[ 151 I. Reiner, K. Roggenkamp,
Integral Representations,
1979. [ 161 E. Spanier, Algebraic Topology, [17] A. Suslin, Stability in Algebraic pp. 304-333. [18] C. Weibel, Mayer-Vieloris
Springer Lecture Notes, vol. 744, Springer,
McGraw-Hill, New York, 1966. K-Theory, Springer Lecture Notes, vol. 966, Springer,
sequences
and module structures
Lecture Notes in Maths, vol. 854, Springer,
on NK,,, Proc. Evanston
Berlin (1981) pp. 466-493.
Berlin,
K-theory
Berlin,
1982, Conf.,