Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

This article was downloaded by: [University of Antwerpen]

On: 3 March 2010


Access details: Access Details: [subscription number 769143503]
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK

Communications in Algebra
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713597239

Abelian Group Gradings on Full Matrix Rings


Jang-Ho Chun a; Jaeun Lee a
a
Department of Mathematics, Yeungnam University, Kyongsan, South Korea

To cite this Article Chun, Jang-Ho and Lee, Jaeun(2007) 'Abelian Group Gradings on Full Matrix Rings', Communications
in Algebra, 35: 10, 3095 — 3102
To link to this Article: DOI: 10.1080/00927870701405116
URL: http://dx.doi.org/10.1080/00927870701405116

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Communications in Algebra® , 35: 3095–3102, 2007
Copyright © Taylor & Francis Group, LLC
ISSN: 0092-7872 print/1532-4125 online
DOI: 10.1080/00927870701405116

ABELIAN GROUP GRADINGS ON FULL MATRIX RINGS

Jang-Ho Chun and Jaeun Lee


Department of Mathematics, Yeungnam University,
Kyongsan, South Korea

For positive integers  and n, several authors studied  -gradings of the full matrix
ring Mn k over a field k. In this article, we show that every G × H-grading of
Mn k can be constructed by a pair of compatible G-grading and H-grading of Mn k,
where G and H are any finite groups. When G and H are finite cyclic groups, we
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

characterize all G × H-gradings which are isomorphic to a good grading. Moreover,


the results can be generalized for any finite abelian group grading of Mn k.

Key Words: Graded algebra; Matrix algebra.

Mathematics Subject Classification: Primary 16W50; Secondary 16S50.

1. INTRODUCTION
Let k be a field and G a finite group with identity g0 . A k-algebra R is

G-graded if R = g∈G Rg is a direct sum of k-subspaces such that Rg Rg ⊂ Rgg
for any g g  ∈ G. A nonzero element a ∈ Rg is called homogeneous of degree g,
denoted by degR a = g. For an abelian group G and a field k, several authors studied
G-gradings of Mn k.

A G-grading R = g∈G Rg of Mn k is good if each matrix unit eij is a
homogeneous element and it is fine if dim Rg ≤ 1 for each g ∈ G. It is clear that for
any good G-grading R, degR eii = g0 for each i = 1 2     n. i.e., dim Rg0 ≥ n. Hence
any good grading is never a fine grading.
Dăscălescu et al. (1999) showed that if k contains a primitive th root  of

unity, then R = −1 i=0 Ri is a  -grading if and only if there exists a matrix X ∈ Mn k
such that X  = aIn for some a ∈ k∗ and Ri = Ci X = A ∈ Mn k  XA = i AX
for all 0 ≤ i ≤  − 1. Moreover, the homogeneous component Ai of A ∈ Mn k is

Ai = 1 −1 −ri X r AX −r . Boboc and Dăscălescu (2001) showed that if a  -grading
−1r=0
R = i=0 Ri is good, then X can be chosen a diagonal matrix whose diagonal
entries are th roots of unity. Moreover, they determined all the isomorphic types

Received May 31, 2006; Revised May 31, 2006. Communicated by M. Ferrero.
Address correspondence to Jaeun Lee, Department of Mathematics, Yeungnam University,
Kyongsan, 712–749 South Korea; Fax: +82-53-810-4614; E-mail: julee@yu.ac.kr

3095
3096 CHUN AND LEE

of p -gradings of Mn k and counted them when p is prime. Bahturin et al. (2001)
showed that if G is an abelian group and k is an algebraically closed field, then there
exist a decomposition n = tq and a subgroup H ⊂ G such that Mn k is isomorphic
to the tensor product Mt k ⊗ Mq k as G-graded algebras, where Mt k is a fine
H-graded algebra and Mq k is a good G-graded algebra. In the case that k is not
algebraically closed, a cohomological method was given in Caenepeel et al. (2002)
to classify n -gradings of Mn k from n -gradings of Mn k̄, where k̄ is the algebraic
closure of k.
In this article, based on the articles Boboc and Dăscălescu (2001) and
Dăscălescu et al. (1999), we study G-gradings of Mn k when G is a finite product of
finite cyclic groups. In the case that G is a product of two groups, the structure of
G-gradings on Mn k is not known. The 2 × 2 -gradings of M2 k are classified
in Boboc (2003). In Section 2, if k contains a primitive mth root of unity, a

necessary and sufficient condition such that R = m−1 i=0 Ci X is (isomorphic to)
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

a good m -grading is given. In Section 3, for a field k having a primitive th



root  and a primitive mth root of unity, it is shown that R = ij Rij is
a  × m -grading if and only if there exist two matrices X and Y in Mn k
such X  = bIn , Y m = cIn , XY = aYX for some a b c ∈ k∗ with an = 1 and Rij =
Cij X Y = A ∈ Mn k  XA = i AX YA = j AY for all 0 ≤ i ≤  − 1, 0 ≤ j ≤ m − 1.

Moreover, R = ij Rij is a good  × m -grading if and only if X and Y are
simultaneously diagonalizable. These are generalizations of the results in Boboc
(2003). We also extend our results when G is a finite product of finite cyclic groups.

2. GRADINGS BY CYCLIC GROUPS


In this section, we consider m -gradings of the full matrix ring Mn k for any
two positive integers m and n. Throughout this section, let k be a field containing a
primitive mth root  of unity.
First we will briefly introduce known results in Boboc and Dăscălescu (2001)

and Dăscălescu et al. (1999). Suppose that R = m−1 i=0 Ri is a m -grading of Mn k.
Then an additive map
R → R defined by
A = i A for each homogeneous
element A ∈ Ri is a graded algebra isomorphism on R. Hence, by the Noether–Skolem
Theorem, there exists a regular matrix X such that
A = XAX −1 for all A ∈ R.
Since
m is the identity map on R = Mn k,
m A = X m AX −m = A for all A ∈ R.
Therefore X m is an element of the center of R and hence X m = aIn for some a ∈ k∗ .
 such−ithat X−1 = aIn for some
Conversely, let X be a matrix in Mn k m
a ∈ k∗ . For anyA ∈
m−1 −m−1i m−1
Mn k, A = i=0 Ai , where Ai = m A +  XAX + · · · + 
1
X AX −m−1 for
m−1
0 ≤ i ≤ m − 1. Let Ri = Ai  A ∈ Mn k. Then R = i=0 Ri is a m -grading of Mn k.
Moreover, Ri = Ci X = A ∈ Mn k  XA = i AX, for all 0 ≤ i ≤ m − 1.

Remark. Let X be a matrix in Mn k such that X m = aIn for some a ∈ k∗ and
Y = bX for some b ∈ k∗ . Then Y m = bm X m = bm aIn and Ci X = Ci Y for all
0 ≤ i ≤ m − 1 and hence X and Y induce the same m -grading R of Mn k. In
particular, if k contains an mth root b of a, then there exists a matrix Y such that
Y m = In , and X and Y induce the same m -grading.
ABELIAN GROUP GRADINGS 3097

Theorem 2.1. Let X be a matrix with X m = aIn for some a ∈ k∗ and let k be a field

i=0 Ci X is a good m -grading of Mn k
containing an mth root  of unity. Then R = m−1
if and only if X is a diagonal matrix. Moreover, in this case, the matrix X is of the form
 
1 0 ··· 0
  
0 2   0 




   
  0
0 0 0 n

where is an mth root of a in k and 2  3      n are mth roots of unity.



Proof. Let X = xij . Suppose that R = m−1 i=0 Ci X is a good m -grading of Mn k.
Then eii ∈ C0 X for all 1 ≤ i ≤ n. Therefore, Xeii = eii X for all 1 ≤ i ≤ n. It implies
that all entries in the ith row and those in the ith column of X are zero except xii
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

and hence X is a diagonal matrix. Conversely, let X = xij  be a diagonal matrix.


Since X m = aIn for some a ∈ k∗ , xiim = a for each i = 1 2     n. Let i = xxii for each
11
i = 2     n. Then m
i = 1 for each i = 2     n. By taking = x11 , we get

 
1 0 ··· 0
  
0 2   0 
X = 



   
  0
0 0 0 n

Let Y = −1 X. Then, by the above remark, Ci X = Ci Y for all 0 ≤ i ≤ m − 1.



Now, to complete the proof, it suffices to show that R = m−1 i=0 Ci Y is a good
m -grading of Mn k. Note that the i j element of Y eij is i and the i j element
of eij Y is j for each 1 ≤ i j ≤ n. Since i is an mth root of unity for each
i = 1 2     n, i =  j for some . That is, for each 1 ≤ i j ≤ n, eij is an element

of C X for some  and hence R = Ci Y is a good m -grading. 

Theorem 2.2. Let Y be a matrix with Y m = aIn for some a ∈ k∗ and let k be a

field containing an mth root  of unity. Then R = m−1 i=0 Ci Y is isomorphic to a good
m -grading of Mn k if and only if Y is similar to a diagonal matrix.

Proof. Suppose that R = m−1 i=0 Ci Y is isomorphic to a good m -grading
m−1
S = i=0 Ci X. Then there exists a graded isomorphism S = Mn k →
R = Mn k. By the Noether–Skolem Theorem, there exists a regular matrix P
such that A = PAP −1 for all A ∈ S = Mn k. Since C0 X = C0 Y, for each
i = 1 2     n,

eii ∈ C0 X if and only if Peii P −1 ∈ C0 Y


if and only if YPeii P −1 = Peii P −1 Y
if and only if P −1 YPeii = eii P −1 YP
if and only if P −1 YP is a diagonal matrix.
3098 CHUN AND LEE

Conversely, suppose that there exists a regular matrix P such that P −1 YP is a



diagonal matrix. Let S = Ci P −1 YP and define R → S by A = P −1 AP for
all A ∈ R. Then is an automorphism on Mn k and Ci Y = Ci P −1 YP for all

i = 0 1     m − 1. By Theorem 2.1, S = Ci P −1 YP is a good m -grading and
hence R is isomorphic to a good grading. 

3. GRADINGS BY FINITE ABELIAN GROUPS



Let G and H be two finite groups and let R = gh∈G×H Rgh be a G × H-
 
grading of Mn k. Let Sg = h∈H Rgh and Th = g∈G Rgh for each g ∈ G and
 
h ∈ H. Then S = g∈G Sg and T = h∈H Th are G-grading and H-grading of Mn k,
respectively. We call them the gradings associated with R.
It is clear that Rgh = Sg ∩ Th for each g ∈ G and h ∈ H. Therefore, every
G × H-grading of Mn k induces a natural G-grading and H-grading of Mn k.
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

But, the converse is not true in general. That is, there exists a pair of G-grading
 
S = Sg and H-grading T = Th of Mn k such that S and T are not the natural
gradings which are induced by a G × H-grading of Mn k. Indeed, if we consider
two 2 -gradings S and T as follows: Let



a 0 0 c
S0 = a b ∈ k  S1 = c d ∈ k 
0 b d 0



a b − a d c
T0 = a b ∈ k  T = c d ∈ k 
0 b 1
d −d

S = S0 + S1 and T = T0 + T1 , then M2 k = S0 ∩ T0 + S0 ∩ T1 + S1 ∩ T0 + S1 ∩ T1



and hence Si ∩ Tj is not a 2 × 2 -grading of M2 k. In this case, S0 ∩ T0 + S0 ∩
T1 = S0 . To avoid this situation, we define the compatibility of two gradings as
follows.

Definition 3.1. Let G and H be two finite groups. A G-grading S = Sg and an
 
H-grading T = Th of Mn k are compatible if Sg = h∈H Sg ∩ Th  for all g ∈ G and

Th = g∈G Sg ∩ Th  for all h ∈ H.
 
Lemma 3.2. Let G and H be two finite groups and let S = Sg and T = Th be
G-grading and H-grading of Mn k, respectively. Then S and T are compatible if and
only if there exists a G × H-grading R of Mn k whose associated G and H-gradings
are S and T , respectively.

Proof. Let R be a G × H-grading of Mn k. Then its associated G- and


H-gradings are compatible. For converse, suppose that S and T are compatible.
 
For each g h ∈ G × H, let Rgh = Sg ∩ Th . Then R = gh Rgh = gh Sg ∩ Th 
is a G × H-grading of Mn k. Indeed, it is clear that Rgh · Rg h  ⊂ Sgg ∩ Thh =

Rgg hh  . Let A ∈ Sg . Since Sg = h∈H Sg ∩ Th , there exists Agh ∈ Sg ∩ Th for each

h ∈ H such that A = h∈H Agh . Since every element A in Mn k can be expressed as
a sum of elements in Sg , hence every element A in Mn k can be expressed as a sum
of elements in Sg ∩ Th = Rgh . It completes the proof. 
ABELIAN GROUP GRADINGS 3099

Theorem 3.3. Let  and m be positive integers and let X and 


−1 Y be matrices

X  = bIn and
with  m−1Y m
= cIn for some b c ∈ k . Let S = S
i=0 i = −1
i=0 Ci X and
T = j=0 Tj = j=0 Cj Y be  and m -gradings of Mn k, respectively. Suppose
m−1

that k contains a primitive th root  and a primitive mth root of unity. Then two
gradings S and T are compatible if and only if XY = aYX for some a ∈ k with an = 1.
m−1
Proof. Suppose that S and T are compatible. Let A ∈ Si . Then A = j=0 Aij ,
where Aij ∈ Si ∩ Tj for each j = 0     m − 1. We observe that
   
−1 −1 −1

m−1
−1

m−1
YXAX Y = Y AY
i
= i
YAij Y =
i j
Aij
j=0 j=0
 

m−1
=X Aij X −1 = XYAY −1 X −1 
j

j=0
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

It implies that YXAYX−1 = XYAXY−1 for all A ∈ Si . Hence YX−1 XYA =


AYX−1 XY for all A ∈ Mn k. Since the multiplicative center of Mn k is the
set aIn  a ∈ k of all scalar matrices of Mn k and XY is regular, we find
YX−1 XY = aIn for some a ∈ k∗ . By comparing the determinants of the both
sides of YX−1 XY = aIn , we can see that an = 1. For converse, suppose that
XY = aYX for some a ∈ k∗ . Let A ∈ Si and let A =  j=0 Aj , where Aj ∈ Tj for each
m−1

1 m−1 −kj k −k
j = 0     m − 1. Since Aj = m k=0 Y AY ,
 
1 
m−1  
XAj X −1 = −kj X Y k AY −k X −1
m k=0
 
1 
m−1  
= −kj Y XAXk −1
Y −k 
m k=0

Since A ∈ Si , XAX −1 = i A and hence XAj X −1 = i Aj . Therefore Aj ∈ Si ∩ Tj and


 m−1 
Si ⊂ m−1
j=0 Si ∩ Tj . It is clear that Si ⊃ S ∩ Tj  and hence Si = m−1
j=0 Si ∩ Tj .
−1 j=0 i
Similarly, we can show that Tj = i=0 Si ∩ Tj . It completes the proof. 

Theorem 3.4. Let  and m be positive integers and let R = ij∈ ×m Rij
 m−1
be a  × m -grading of Mn k. Let S = −1 i=0 Si and T = j=0 Tj be two
gradings associated with R such that Si = Ci X i = 0 1      − 1 and Tj = Cj Y
j = 0 1     m − 1 for some matrices X and Y with X  = bIn and Y m = cIn for some
b c ∈ k∗ . Suppose that k contains a primitive th root  and a primitive mth root of
unity. Then
 
 
 1  
−ri −sj r s −s −r
Rij =  X Y AY X A ∈ Mn k = Cij X Y

 m 0≤r≤−1 


0≤s≤m−1

for each i = 0 1      − 1 and j = 0 1     m − 1, where Cij X Y = A ∈ Mn k 


XA = i AX YA = j AY.
3100 CHUN AND LEE


Proof. It comes from Lemma 3.2 and Theorem 3.3 that Si = m−1 k=0 Rik i = 0
−1
1      − 1, Tj = k=0 Rkj j = 0 1     m − 1 and XY = aYX for some a ∈ k
with an = 1. It is known in Section 2 that
 
1  −ri r
−r
Si =  X AX A ∈ Mn k
 0≤r≤−1

and
 
1 
−sj s −s
Tj = Y AY A ∈ Mn k 
m 0≤s≤m−1

Since XY = aYX, Y s X r AX −r Y −s = X r Y s AY −s X −r for all A ∈ Mn k and all integers r


and s, we have
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

 
1  −ri r
−r
Rij = Si ∩ Tj =  X AX A ∈ Tj
 0≤r≤−1
 
1 
−sj s −s
= Y AY A ∈ Si
m 0≤s≤m−1
 
 
 1  
−ri −sj r s −s −r
=  X Y AY X A ∈ Mn k = Cij X Y

 m 0≤r≤−1 


0≤s≤m−1

for each i = 0 1      − 1 and j = 0 1     m − 1. 



Lemma 3.5. Let G and H be two finite groups and let R = gh∈G×H Rgh be a
G × H-grading of Mn k. Then R is a good grading if and only if the two gradings S
and T associated with R are good gradings.

Proof. We only prove the nontrivial implication. Suppose that S and T are good
gradings. Then for each 1 ≤ i j ≤ n, eij ∈ Sg for some g ∈ G, and eij ∈ Th for some
h ∈ H. Hence, eij ∈ Sg ∩ Th for some g ∈ G and h ∈ H. Hence, R is a good grading.


Theorem 3.6. Let  and m be positive integers and let R = ij∈ ×m Rij
 m−1
be a  × m -grading of Mn k. Let S = −1 i=0 Si and T = j=0 Tj be two
gradings associated with R such that Si = Ci X i = 0 1      − 1 and Tj = Cj Y
j = 0 1     m − 1 for some matrices X and Y with X  = bIn and Y m = cIn for some
b c ∈ k∗ . Suppose that k contains a primitive th root of unity and a primitive mth
root of unity. Then R is isomorphic to a good grading if and only if X and Y are
simultaneously diagonalizable.

Proof. Suppose that there exists an isomorphism  R → U , where U is a good


 × m -grading of Mn k. Then Rij  = Uij for all i = 0 1 2      − 1 and
ABELIAN GROUP GRADINGS 3101

j = 0 1 2     m − 1. By the Noether–Skolem Theorem, there exists a regular


matrix P such that A = PAP −1 . Let V and W be the two gradings associated
with U . Then Si  = Vi and Tj  = Wj . Therefore,  S → V and  T → W are
isomorphisms as  - and m -gradings, respectively. We observe that


−1 
−1   
m−1 
m−1  
V = Vi = Ci PXP −1 and W = Wj = Cj PYP −1 
i=0 i=0 j=0 j=0

Since U is a good grading, by Lemma 3.5, both V and W are good gradings. By
Theorem 2.1, both PXP −1 and PYP −1 are diagonal. That is, X and Y are simultaneously
diagonalizable. Conversely, suppose that X and Y are simultaneously diagonalizable.
Then there exists a regular matrix P such that both PXP −1 and PYP −1 are diagonal. By
 −1 −1
m−1
Theorem 2.1, the two gradings V = −1 i=0 Vi = i=0 Ci PXP  and W = j=0 Wj =
m−1 −1
j=0 C j PYP  are good   - and  m -gradings, respectively. Since two diagonal
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

matrices commute, by Theorem 3.3, the two gradings V and W are compatible. Now,
 
it comes from Lemma 3.5 that U = ij Vi ∩ Wj = ij Ci PXP −1  ∩ Cj PYP −1  is a
good grading. We define  R → U by A = PAP −1 for all A ∈ Mn k. Then  is an
isomorphism. It completes the proof. 

Let R = i1 i2 i  Ri1 i2 i  be a m1 × m2 × · · · × m -grading of Mn k,
where m1 , m2 ,    , and m are natural numbers. For each 1 ≤ t ≤ , we set

mt −1 mt −1
  
St = Sti = Ri1 i2 it−1 iit+1 i  
i=0 i=0 i1 i2 it−1 iit+1 i 

Then St is a mt -grading of Mn k which is called the mt -grading associated with R.
For each t = 1 2     , let us take a regular matrix Xt such that

Sti = Ci Xt  = A ∈ Mn k  Xt A = t i AXt 

where t is a primitive mt th root of unity. Now, the following corollary comes from
Lemma 3.5 and Theorem 3.6.

Corollary 3.7. Let m1 , m2     , and m be positive integers and let R = i1 i2 i 
Ri1 i2 i  be a m1 × m2 × · · · × m -grading of Mn k. For each t = 1 2     , let
mt −1 mt −1 m
St = i=0 Sti = i=0 Ci Xt  be the mt -grading associated with R, where Xt t =

at In for some at ∈ k . Suppose that k contains a primitive mt th root of unity for each
t = 1 2     . Then R is isomorphic to a good grading if and only if X1  X2      and
X are simultaneously diagonalizable.

ACKNOWLEDGMENT
The authors would like to express their gratitude to the referee for valuable
comments and constructive suggestions of the manuscript.
Jaeun Lee is supported by Com2 MaC-KOSEF (R11-1999-054).
3102 CHUN AND LEE

REFERENCES
Bahturin, Yu. A., Sehgal, S. K., Zaicev, M. V. (2001). Group gradings on associative
algebras. J. Algebra 241(2):677–698.
Boboc, C. (2003). Gradings of matrix algebras by the Klein group. Comm. Algebra
31(5):2311–2326.
Boboc, C., Dăscălescu, S. (2001). Gradings of matrix algebras by cyclic groups. Comm.
Algebra 29(11):5013–5021.
Caenepeel, S., Dăscălescu, S., Năstăsescu, C. (2002). On gradings of matrix algebras and
descent theory. Comm. Algebra 30(12):5901–5920.
Dăscălescu, S., Ion, B., Năstăsescu, C., Rios Montes, J. (1999). Group gradings on full
matrix rings. J. Algebra 220(2):709–728.
Knus, M. A. (1969). Algebra graded by a group. In: Hilton, P., ed. Category Theory,
Homology Theory and Their Appllications II. Lecture Notes in Math., Vol. 92. Berlin:
Springer-Verlag, pp. 117–133.
Downloaded By: [University of Antwerpen] At: 09:29 3 March 2010

You might also like