Computer Computations in Representation

Theory II:
Root Systems and Weyl Groups
Jerey Adams University of Maryland
September 11, 2002
www.math.umd.edu/jda/minicourse
1
6 GL(n, R)
Recall G = GL(n, R) is the group of n n
invertible matrices over R.n
Let V = M
nn
(R) R
n
2
(as a vector space),
and dene a representation of G on V by
(g)(X) = gXg
1
(g G, X V )
The subspace of matrices of trace 0 is an ir-
reducible representation of G.
Now let T G be the subgroup of diagonal
matrices. Then V is a representation of T by
2
restriction:
(t)(X) = tXt
1
(t T, X V )
Problem: Decompose V , as a representation
of T, into a direct sum of irreducible represen-
tations.
Note that T is abelian; in fact T R
n
. Here
are some one-dimensional representations of T.
3
Let
Z
n

k = (k
1
, . . . , k
n
) : T R

be the map
diag(x
1
, x
2
, . . . , x
n
) x
k
1
1
x
k
2
2
. . . x
k
n
n
This is a group homomorphism.
Let E
i,j
be the matrix with a 1 in the i
th
row and j
th
column, and 0 elsewhere. Let t =
diag(x
1
, . . . , x
n
). Then
(t)E
i,j
=
x
i
x
j
E
i,j
That is C < E
i,j
> is a one-dimensional repre-
4
sentation of T given by

k =

(0, . . . , 0) i = j
(0, . . . , 0, 1
i
, 0, . . . , 0, 1
j
, 0, . . . , 0) i ,= j
with 1 in the i
th
place and 1 in the j
th
place.
Write e
1
, . . . , e
n
for the standard basis of R
n
(or
Z
n
). Then (i ,= j)
(t)E
i,j
= (e
i
e
j
)(t)E
i,j
This gives the solution to the Problem: V is
the direct sum of one-dimensional representa-
tions
e
i
e
j
(1 i ,= j n)
and the trivial representation

0 with multiplic-
5
ity n.
The set
R = e
i
e
j
[ 1 i ,= j n R
n
is an example of a root system.
Now let
W = Norm
G
(T)/T
For example any permutation matrix is con-
tained in Norm
G
(T), and acts on T (by conju-
gation) by the natural permutation action. For
example
6

0 1 0
0 0 1
1 0 0

x
1
x
2
x
3

0 1 0
0 0 1
1 0 0

1
=

x
2
x
3
x
1

In fact W S
n
. This is an example of a
Weyl group.
Note that the action of W on T by conjuga-
tion gives an action of W on R, again by the
natural permutation action.
6.1 Other Groups
Now let G be Sp(2n, R) or SO(n, R). Recall
G = g GL(m, R) [ gJg
t
= J
with J as in Lecture I.
Let
g = X M
mm
(R) [ XJ + JX
t
= 0
The g is a representation of G by
(g)(X) = gXg
1
(g G, X g)
7
(check this).
Let T be the diagonal subgroup:
T = diag(x
1
, . . . , x
n
,
1
x
1
, . . . ,
1
x
n
)
(with an additional 1 in the case of SO(2n+1)).
Again T is isomorphic to R
n
, and we have
the onedimensional representations

k Z
n
of
T. As before, decompose g as a representation
of T.
We get e
i
e
j
[ 1 i ,= j k as before.
Let
F
i,j
=

0
0
.
.
.
0
0
0
1
i,j
1
j,i
0
0
0
.
.
.
0
0
0
0
.
.
.
0
0

8
Then
(t)F
i,j
= (e
i
+ e
j
)(t)F
i,j
(if i = j this is 2e
i
(t)F
i,i
).
Let R Z
n
be the non-zero elements which
occur. We dene W as before
W = Norm
G
(T)/T
This is a nite group.
9
GL(n, R):
R = (e
i
e
j
) [ 1 i < j n
W S
n
consists of all permutations in n
coordinates
Sp(2n, R):
R = e
i
e
j
[ 1 i < j n2e
i
[ 1 i n
W consists of all permutations and sign changes
in n coordinates.
SO(2n, R):
R = e
i
e
j
[ 1 i < j n
W consists of all permutations and an even
number sign changes in n coordinates.
SO(2n + 1, R):
R = e
i
e
j
[ 1 i < j ne
i
[ 1 i n
W consists of all permutations and sign changes
in n coordinates.
10
6.2 Formal Denition of Root Systems
Basic references are [1] and [2]. A root systemR
is a nite subset of V = R
n
with the following
properties. Write (v, w) = vw for the standard
inner product on V .
Let
v
be the reection in the plane orthogo-
nal to v. Then

v
(w) = w < w, v > v
where
< w, v >= 2(w, v)/(v, v)
The requirements are:
1. 0 , R and R spans V
2. if R then are the only multiples of
in R
3. If , R then < v, w > Z
4. If R then

: R R
That is , R implies < , >
R.
11
Given R let W be the group generated by the
reections

[ R.
Thus W acts on R.
In the case of a root system coming from a
group the Weyl group is isomorphic to Norm
G
(T)/T.
Example: In the root system of GL(n, R),
Sp(2n, R) or SO(2n + 1, R), = e
i
e
j
gives
the transposition (i j) in S
n
. These generate
S
n
. In the case of GL(n, R) this is all of W.
Example: In the case of Sp(2n, R) or SO(2n+
1, R), if = e
i
or 2e
i
then

is the sign change

in the i
th
coordinate. These generate all permu-
tations and sign changes, i.e. the Weyl group of
type B
n
.
12
Now root systems are very rigid. In fact the
possible angles between roots are 2/n with
n = 1, 2, 3, 4, 6. Note that 2/5 is not allowed.
Theorem 6.1 The irreducible root systems
are: A
n
, B
n
, C
n
, D
n
(n 1) and E
6
, E
7
, E
8
, F
4
and G
2
.
The root systems of GL(n, R), SO(2n+1, R),
Sp(2n, R) and SO(2n, R) are the classical
root systems of type A
n1
, B
n
, C
n
and D
n
, re-
spectively.
The root systems E
6
, E
7
, E
8
, F
4
and G
2
are
called the exceptional root systems.
13
Rank 2 root systems
A2
C2
B2
G2
The Weyl groups are
W(A
1
) Z/2Z Z/2Z
W(B
2
) W(C
2
) D
4
W(G
2
) D
6
14
The exceptional root systems and Weyl groups
E
8
: e
i
e
j
, 1 i < j 8
1
2
(
1
, . . . ,
8
),
i
= 1,

i
= 1
F
4
: e
i
e
j
, 1 i < j 4
1
2
(
1
, . . . ,
4
),
i
= 1
G
2
: e
i
e
j
, 1 i < j 3
(2, 1 1), (1, 2, 1), (1 1, 2)
Type [R[ Order(W) realization
E
6
72 51,840 O(6, F
2
)
E
7
126 2,903,040 O(7, F
2
) Z/2Z
E
8
240 696,729,600 W
2
O(8, F
2
)
F
4
48 1152
G
2
12 12 D
6
15
6.3 Lie groups and root systems
Recall the groups GL(n, R), Sp(2n, R) and SO(m, R)
each give rise to a root system and its Weyl
group. The converse holds:
Theorem 6.2 (Fantastic Theorem:) Let R
be a root system. Then there is a Lie group
G for which this is the root system.
More precisely:
There is a subgroup G of some GL(m, R),
and
a subspace g of M
mm
(R), such that
G acts on g by (g)(X) = gXg
1
The diagonal subgroup T is isomorphic to
R
n
The onedimensional representations of T
on g are the root system R Z
n
R
n
The Weyl group of Ris isomorphic to Norm
G
(T)
16
The Lie groups of type E
6
, E
7
, E
8
, F
4
and G
2
are among the most fascinating objects in math-
ematics.
Example: E
8
(F
2
) The preceding construction
works over any eld (this is one of the remark-
able things about it). The group E
8
(F
2
) is a
nite simple group of order
337, 804, 753, 143, 634, 806, 261, 388, 190, 614, 085, 595, 079, 991,
692, 242, 467, 651, 576, 160, 959, 909, 068, 800, 000 10
75
17
7 More on Root Systems
Let R be a root system of rank n, i.e. the ambi-
ent vector space is of dimension n. Then there
is a basis
1
, . . . ,
n
R of V .
For example if R is of type A
n1
then we
can take
i
= e
i
e
i+1
. The corresponding
reections are the transpositions (i, i +1) in S
n
,
which generate S
n
. This is an example of a basis
with further nice properties: a set of simple
roots.
Denition 7.1 A set
1
, . . . ,
n
is a set of
simple roots if it is a basis of V and every
root R can be written
=

i
a
i

i
with all a
i
0 or all a
i
0.
18
Example: B
2
B2

+
2+
19
Given a set of simple roots S =
1
, . . . ,
n

let s
i
= s

i
. For w W let length(w) be the
minimum k so that
w = s
i
1
s
i
2
. . . s
i
k
Theorem 7.2 Let
1
, . . . ,
n
be a set of sim-
ple roots.
W is generated by s
i
= s

i
[ 1 i n.
There is a unique longest element w
0
of
the Weyl group
Note: In types B
n
, C
n
, D
2n
, E
7
, E
8
, F
4
and
G
2
, w
0
= I.
20
Weyl group of type B
2
W : 1, s

, s

, s

, s

, s

, s

, s

B2

2+
+

s
s

s s

21
7.1 Hyperplanes
Let R V = R
n
be a root system. Recall for
, R,
< , >= 2(, )/(, )
where (, ) = . This makes sense for any
v V :
Denition 7.3 For R, v V ,
< v, >= 2(v, )/(, )
Note: if there is only one root length (types
A,D,E) we may take (, ) = 2 for all R,
and then
< v, >= (v, )
You may want to think about this case at rst.
Now each R gives a hyperplane
v [ < v, >= 0
22
Hyperplanes of A
3
23
More generally each k Z, R gives a
hyperplane
v [ < v, >= k
This breaks V up into countably many facets.
24
A2
25
B2
26
G2
27
References
[1] J. Humphreys. Introduction to Lie Algebras
and Representation Theory. Springer-Verlag,
Berlin,Heidelberg,New York, 1972.
[2] James E. Humphreys. Reection groups and Cox-
eter groups, volume 29 of Cambridge Studies in Ad-
vanced Mathematics. Cambridge University Press,
Cambridge, 1990.
28