Professional Documents
Culture Documents
Ichiro Yokota
Preface
In the end of 19 century, W. Killing and E. Cartan classified the complex simple
Lie algebras, called An , Bn , Cn , Dn (classical type) and G2 , F4 , E6 , E7 , E8 (exceptional
type). These simple Lie algebras and the corresponding compact simple Lie groups
have offered many subjects in mathematicians. Especially, exceptional Lie groups are
very wonderful and interesting miracle in Lie group theory.
Now, in the present book, we describe simply connected compact exceptional
simple Lie groups G2 , F4 , E6 , E7 , E8 , in very elementary way. The contents are given
as follows. We first construct all simply connected compact exceptional Lie groups
G concretely. Next, we find all involutive automorphisms of G, and determine the
group structures of the fixed points subgroup G by . Note that they correspond to
classification of all irreducible compact symmetric spaces G/G of exceptional type,
and that they also correspond to classification of all non-compact exceptional simple
Lie groups. Finally, we determined the group structures of the maximal subgroups of
maximal rank. At any rate, we would like this book to be used in mathematics and
physics.
The author thanks K. Abe, K. Mituishi, T. Miyasaka, T. Miyashita, T. Sato, O.
Shukuzawa, K. Takeuchi and O. Yasukura for their advices and encouragements.
Ichiro Yokota
Contents
I.
II.
21
22
23
26
28
31
33
36
2.4. Simplicity of f4 C
2.5. Killing form of f4
1
2
3
8
9
11
13
14
16
19
21
41
2.6. Roots of f4
2.7. Subgroup Spin(9) of F4
2.8. Connectedness of F4
2.9. Involution and subgroup Spin(9) of F4
2.10. Center z(F4 ) of F4
2.11. Involution and subgroup (Sp(1) Sp(3))/Z 2 of F4
2.12. Automorphism w of order 3 and subgroup (SU (3) SU (3))/Z 3
of F4
C
45
47
51
55
57
58
59
63
65
66
III.
IV.
68
68
70
71
73
75
82
82
85
86
90
95
100
105
105
106
108
108
114
116
118
124
126
129
130
133
142
146
155
155
V.
5.4.
5.5.
5.6.
5.7.
157
157
159
160
161
163
173
176
184
184
186
192
199
Notation
R, C = R Re1 , H = R Re1 Re2 Re3 denote the fields of real, complex,
quaternion numbers, respectively.
For R-vector space V , its complexification {u + iv | u, v V } is denoted by V C .
The complex conjugation in V C is denoted by :
(u + iv) = u iv.
RC is briefly denoted by C.
For K-vector space V (K = R, C, C), IsoK (V ) denotes all K-linear isomorphisms
of V . For a K-linear mapping f of V , Vf denotes {v V | f (v) = v}.
For K-vector spaces V, W (K = R, C), HomK (V, W ) denotes all K-homomorphism f : V W . HomK (V, V ) is briefly denoted by HomK (V ).
Let G be a group and an automorphism of G. Then G denotes {g G | (g) =
g}. For s G, Gs denotes {g G | sgs1 = g}.
For topological spaces X, Y , X Y denotes that X and Y are homeomorphic.
For groups G, G , G
= G denotes that G and G are isomorphic as groups. Isomorphic two groups G, G are often identified: G = G .
M (n, K) denotes all n n matrices with entries in K.
E = diag(1, , 1) M (n, K) is the unit matrix.
For A M (n, K), t A denotes the transposed matrix of A and A denotes the
conjugate transposed matrix of A: A = t A.
O(n) = {A M (n, R) | t AA = E} (orthogonal group),
For a Lie group G, its Lie algebra is denoted by the correspoding small Germann
letter g. For example, su(n) is the Lie algebra of the special unitary group SU (n).
Special notations
,
g,
w.
e2 e3 = e1 ,
e3 e1 = e2 ,
ei ej = ej ei , i 6= j, i 6= 0, j 6= 0,
and the distributive law. Thus C has a multiplication. x1, x R is briefly denoted
by x. In C, the conjugate x, an inner product (x, y), the length |x| and the real part
R(x) are defined respectively by
e1
e7
e2
e4
e3
x0 +
7
X
i=1
xi ei = x0
e6
e5
7
X
i=1
p
|x| = (x, x),
xi ei ,
7
X
i=0
xi ei ,
7
X
i=0
7
X
xi yi ,
yi ei =
i=0
7
X
R x0 +
xi ei = x0 .
i=1
x
by x1 , then we have xx1 = x1 x = 1, and C
|x|2
satisfies all axioms of a field except the associative law x(yz) = (xy)z. Of course
the commutative law xy = yx does not hold. Since the associative law does not
For x C, x 6= 0, we denote
hold, calculations in C are complicated, however the following relations hold. (See
Freudenthal [6] or Yokota [58]). For a, b, x, y C, we have
1
2
3
4
5
x, y C.
i = 1, 2, , 7.
(x, y) =
Gij ei = ej ,
Gij ek = 0, k 6= i, j.
1
ei (ej x),
2
x C.
2F01
2F
23
2F
45
2F67
=
=
=
=
2F02
2F
13
2F
46
2F57
2F03
2F
12
2F
47
2F56
2F05
2F
14
2F
27
2F36
2F = G + G + G G
15
04
15
26
37
2F
=
G
+
G
+
G
+
G
26
04
15
26
37
2F
=
G
+
G
+
G25 + G34
07
07
16
2F = G + G G G
16
07
16
25
34
=
=
=
=
x C,
(Gij ) = Fij ,
i, j = 0, 1, , 7, i 6= j,
= .
x C.
Hence Aut(D4 ). Since maps the R-basis {Gij | 0 i < j 7} to the R-basis
{Fij | 0 i < j 7} of D4 , is an R-linear isomorphism. To show that is an
automorphism of D4 , it is sufficient to show that
[Gij , Gkl ] = [Gij , Gkl ],
which in turn would follow from the relations
[Fij , Fjk ] = Fik ,
[Fij , Fkl ] = 0,
i, j, k are distinct,
i, j, k, l are distinct.
1
1
ei (el (ek x)) el (ek (ei x))
4
4
1
1
= el (ei (ek x)) + el (ei (ek x)) = 0,
4
4
1
1
[Fij , Fjk ]x = (Fij Fjk Fjk Fij )x = ej (ei (ek (ej x))) ek (ej (ej (ei x)))
4
4
1
1
1
1
= ei (ek (ej (ej x))) + ek (ei x) = ei (ek x) + ek (ei x)
4
4
4
4
1
= ek (ei x) = Fik x,
2
1
1
[Fij , Fkl ]x = (Fij Fkl Fkl Fij )x = ej (ei (el (ek x))) el (ek (ej (ei x)))
4
4
1
1
= ej (ek (ei (el x))) ej (el (ek (ei x))) = = 0.
4
4
Hence Aut(D4 ). Finally, since = , we have Aut(D4 ).
Definition. For a C, we define R-linear mappings La , Ra , Ta : C C respectively by
La x = ax,
Ra x = xa,
Ta x = ax + xa = (La + Ra )x,
x C.
(1)
(2)
La , Ra , Ta D4 .
La = Ra , Ra = La ,
(3)
La = Ta ,
(4)
La = Ra ,
Ra = Ra ,
Ra = Ta ,
Ta = Ta .
Ta = La .
Ta = La .
Proof. (1) (La x, y) = (ax, y) = (x, ay) = (x, ay) = (x, La y), x, y C. Hence
La D4 . Similarly, Ra D4 and Ta = La + Ra D4 .
It follows that
Lei x = ei x = 2Fi0 x, x C,
x=1
2ei ,
2,
x = ei
Tei x = ei x + xei =
0,
x = ej , j 6= 0, i.
2 = 1,
3 = 1,
= .
, 2
,
2 3 1
3 1 2
1 3 2
2
x, y C.
D3 = D1 .
x, y C.
(i)
x, y C.
(ii)
[Lb , Lc ],
(iii)
a, b, c C0
P
P
(Lemma 1.3.4), putting D2 = Ra + [Rb , Rc ], D3 = Ta + [Tb , Tc ] and using (i),
(iii), we obtain
(D1 x)y + x(D2 y) = D3 (xy), x, y C.
Next, we shall show that D2 and D3 are determined uniquely for D1 . To prove this,
it is sufficient to show for D1 = 0 that D2 = D3 = 0. Now, suppose
x(D2 y) = D3 (xy),
x, y C.
x, y C.
(iv)
Putting D1 = p, we have 2(p, 1) = (p, 1)+(1, p) = (D1, 1)+(1, D1) = 0, which implies
p C0 . Furthermore, putting y = 1 in (iv), we have xp = Dx, so (iv) becomes
x(yp) = (xy)p,
for all x, y C.
x, y C
x, y C,
x, y C.
implies
x, y C,
x, y C0 .
x, y C.
Hence g2 b3 . The relation in the lemma follows easily from Theorem 1.3.5 and
Theorem 1.3.6.
Theorem 1.4.3. Any element of g2 is expressed by the sum of elements of the
following seven types.
G23 + G45 + G67 ,
G12 + G47 + G56 ,
G14 G27 G36 ,
, , R
++ =0
0<i<j
13 46 + 57 = 0,
12 + 47 + 56 = 0,
15 + 26 37 = 0,
14 27 36 = 0,
17 24 + 35 = 0,
16 + 25 + 34 = 0,
from which the first result follows. Conversely, any of these seven elements D obviously belongs to b3 and the condition D = D (Lemma 1.4.2) is verified from the
table of Lemma 1.3.1.
1.5. Lie subalgebra su(3) of g2
The Cayley algebra C naturally contains the field C of complex numbers as
C = {x0 + x1 e1 | xi R}.
Any element x C is expressed by
x = x0 + x1 e1 + x2 e2 + x3 e3 + x4 e4 + x5 e5 + x6 e6 + x7 e7
(xi R)
that is,
x = a + m1 e 2 + m2 e 4 + m3 e 6 ,
a, mi C.
m1
a + m2 .
m3
X
X
1
mi n i ,
(mi , ni ), hm, ni =
(m n + n m) =
2
i=1
i=1
m2 n 3 n 2 m3
m n = m3 n 1 n 3 m1
m1 n 2 n 1 m2
m1
n1
for m = m2 , n = n2 C 3 . Since these operations correspond to their
m3
n3
respective operations in C, hereafter, we identify C C 3 with C , that is,
(m, n) =
C C 3 = C.
We shall study the following subalgebra (g2 )e1 of g2 :
(g2 )e1 = {D g2 | De1 = 0}.
Theorem 1.5.1.
(g2 )e1
= su(3).
(D)(a + m) = Dm,
of an R-basis of su(3) (where Ekl M (3, R) is the matrix with the (k, l)-entry is 1
and otherwise are 0), we have
(e1 (E11 E22 )) = G23 + G45 ,
(E12 E21 ) = G24 + G35 ,
(E13 E31 ) = G26 + G37 ,
(E23 E32 ) = G46 + G57 ,
Hence (D) g2 (Theorem 1.4.3). Clearly (D)e1 = 0, so that (D) (g2 )e1 .
Obviously : su(3) g2 is a homomorphism as Lie algebras and is injective, so
that we identify su(3) and (su(3). We set
S1 = 2G12 G47 G56 , S2 = 2G13 G46 + G57 ,
S3 = 2G14 + G27 + G36 , S4 = 2G15 + G26 G37 ,
S5 = 2G16 G25 G34 , S6 = 2G17 G24 + G35 ,
and let S be the R-vector subspace of g2 spanned by S1 , , S6 . Then we have the
following decomposition of g2 .
g2 = su(3) S.
Now, we shall show that : su(3) (g2 )e1 is onto. Let B (g2 )e1 . Denote
B=D+
6
X
xi Si ,
D su(3), xi R.
i=1
xi C.
For a while, in the Lie algebra su(3) g2 , we use the following notations.
H1 = G23 + G45 , H2 = G45 + G67
L12 = G24 + G35 , L21 = G25 + G34 , L13 = G26 + G37 ,
L21 = G27 + G36 , L23 = G46 + G57 , L32 = G47 + G56 .
Lemma 1.6.1. The Lie brackets [D, S], D su(3), S S are given as follows.
H1
H2
L12
L21
L13
L31
L23
L32
S1
S2
0
S3
S4
S5
S6
0
0
S2
S3
S1 S4
0
S4
S4
S1
S3
S2
S6
0
S5
0
0 S5
0
S6
S4
S3
S3
S2
S1
0
0
S6
S5
S5
S6
0
0
S6
S5
0
0
0
0
S1
S2
S2 S1
S3
S4
S4 S3
(i) and
x1 S5 x2 S6 W
(ii)
(Theorem 1.5.1).
[D , S] S
We show that either su(3) a 6= {0} or S a 6= {0}. Assume that su(3) a = {0}
and S a = {0}. The mapping p|a : a su(3) is injective because S a = {0}.
Since p(a) is a non-zero ideal of su(3) and su(3) is simple, we have p(a) = su(3).
Hence dim a = dim p(a) = dim su(3) = 8. On the other hand, since su(3) a = {0},
q|a : a S is also injective, we have dim a dim S = 6. This leads to a contradiction.
We now consider the following two cases.
(1) Case su(3) a 6= {0}. From the simplicity of su(3), we have su(3) a = su(3),
hence a su(3). On the other hand, we have
a [a, S] [su(3), S] = S (Lemma 1.6.1).
Hence a su(3) S = g2 .
Di g2 C .
1
(1 + ie1 ),
2
a + m C C (C 3 )C = CC ,
0
n 1 0
o
h = 0 2 0 i C, 1 + 2 + 3 = 0
0
0 3
of sl(3, C), and Ekl is a root vector associated with the root k l .
Theorem 1.8.1. The rank of the Lie algebra g2 C is 2. The roots of g2 C relative
to some Cartan subalgebra of g2 C are given by
(1 2 ),
(1 3 ),
(2 3 ),
1 ,
with 1 + 2 + 3 = 0.
Proof.
The remainder roots and associated root vectors are found as follows.
1
2 = 2
i
2
3
i
2
1 3 = 21 + 32 , 2 3 = 1 + 32 ,
2 = 2 ,
3 = 1 + 22 .
3
X
k k ,
k=1
for H = i1 G23 i2 G45 i3 G67 , H = i1 G23 i2 G45 i3 G67 hR (Theorem 1.7.2). Now, the canonical element Hi hR associated with i (B2 (H , H) =
(H), H hR ) are determined as follows.
1
1
H1 = iG23 + iG45 ,
8
8
Hence we have
H2 =
1
1
1
iG23 iG45 + iG67 .
24
12
24
11
1
(1 + 1) = ,
88
4
1
1 1
(1 + 4 + 1) =
,
(2 , 2 ) = B2 (H2 , H2 ) = 8
24 24
12
1 1
1
(1 , 2 ) = B2 (H1 , H2 ) = 8
(1 2) = ,
8 24
8
1
1
(, ) = , (, 1 ) = , (, 2 ) = 0.
4
8
(1 , 1 ) = B2 (H1 , H1 ) = 8
15
Using them, we can draw the Dynkin diagram and the extended Dynkin diagram of
g2 C .
According to Borel-Siebenthal theory (Borel and Siebenthal [4]), the Lie algebra
g2 has two subalgebras as maximal subalgebras with the maximal rank 2.
(1) One is a subalgebra of type C1 C1 which is obtained as the fixed points by
an involution of g2 .
(2) The other is a subalgebra of type A2 which is obtained as the fixed points by
an automorphism w of order 3 of g2 .
These subalgebras will be realized in the group G2 in Theorem 1.10.1 and Theorem
1.9.4, respectively.
1.9. Automorphism w of order 3 and subgroup SU (3) of G2
We shall study the following subgroup (G2 )e1 of G2 :
(G2 )e1 = { G2 | e1 = e1 }.
Theorem 1.9.1.
(G2 )e1
= SU (3).
(A)(a + m) = a + Am,
We first prove that (A) (G2 )e1 . For = (A), A SU (3) and x = a + m, y =
e (which is the adjoint matrix of
b + n C C 3 = C, using that if A SU (3) then A
1
A) = A = A , we have
(x)(y) = (a + Am)(b + An)
= (ab hAm, Ani) + (aAn + bAm Am An)
e
= (ab hm, A Ani) + (aAn + bAm t A(m
n))
= (ab hm, ni) + A(an + bm m n)
e4 = a2 ,
e6 = a3
and construct a matrix A = (a1 , a2 , a3 ) M (3, C). From (e2 )(e4 ) = (e2 e4 ) =
e6 , we have a1 a2 = a3 , namely, ha1 , a2 i a1 a2 = a3 , hence we have
ha1 , a2 i = 0,
a3 = a1 a2 .
16
G2 /SU (3) S 6 .
Proof. S 6 = {a C | a
= a, |a| = 1} is a 6 dimensional sphere. Since the group
G2 is a subgroup of O(7) = { O(C) | 1 = 1} (Remark of Theorem 1.2.2), G2 acts
on S 6 . We shall show that this action is transitive. To prove this, it is sufficient to
show that any element a S 6 can be transformed to e1 S 6 by some G2 . Now,
for a1 S 6 , choose any element a2 S 6 such that (a1 , a2 ) = 0. Let
a3 = a1 a2 .
6
a6 = a4 a2 ,
a7 = a3 a4 .
i = 0, 1, , 7
x, y C.
i, j = 0, 1, , 7
1
3
where 1 = +
e1 C C. This w is defined as
2
2
a + m C C 3 = C.
w(a + m) = a + 1 m,
Then w G2 and w3 = 1.
Proof. Recall the mapping : SU (3) G2 of Theorem 1.9.1. We first show that
(SU (3)) (G2 )w . Indeed, for A SU (3) and a + m C C 3 = C, we have
w(A)(a + m) = w(a + Am) = a + 1 Am
= a + A1 m = (A)w(a + m).
Hence w(A) = (A)w, so that (A) (G2 )w . Conversely, let (G2 )w . We
consider the R-vector subspace Cw = {x C | wx = x} of C, then Cw = C. Since
satisfies w = w, Cw is invariant under . Since the restriction of to Cw induces
an automorphism of C, we have
e1 = e1
or e1 = e1 .
18
for all m C 3 ,
(xi R)
m, a H.
), a conjugation
(m + ae4 ) = m ae4 .
(G2 )
= (Sp(1) Sp(1))/Z 2 , Z 2 = {(1, 1), (1, 1)}.
m + ae4 H He4 = C.
We first show that (p, q) (G2 ) . For = (p, q), p, q Sp(1) and x = m + ae4 ,
y = n + be4 H He4 = C, we have
(x)(y) = (qmq + (paq)e4 )(qnq + (pbq)e4 )
= ((qmq)(qnq) (pbq)(paq)) + ((paq)(qnq) + (pbq)(qmq))e4
= q(mn ba)q + (p(an + bm)q)e4
mH
(Proposition 108 of Yokota [58]). By putting = (1, q)1 , we have (G2 ) and
|H = 1. Therfore induces a transformation of C = {x C | x = x} = He4 .
By putting e4 = pe4 , where p H, we have |p| = |pe4 | = |e4 | = |e4 | = 1, which
implies p Sp(1). From
(m + ae4 ) = m + (a)(e4 ) = m + a(pe4 ) = m + (pa)e4 = (p, 1)(m + ae4 ),
we have = (p, 1). Therefore we have
= (1, q) = (1, q)(p, 1) = (p, q),
which shows that is onto. Ker = {(1, 1), (1, 1)} = Z 2 is easily obtained. Thus
we have the isomorphism (Sp(1) Sp(1))/Z 2
= (G2 ) .
Remark 1.
(Sp(1) Sp(1))/Z 2
= SO(4).
xH
Remark 2. Since (G2 ) is connected as the fixed points subgroup by an automorphism of the simply connected group G2 , the fact that : Sp(1) Sp(1) (G2 )
is onto can be proved as follows. The elements
2G12 G47 G56 , G47 + G56 ,
2G13 G46 G57 ,
G46 + G57 ,
2G23 G45 G67 , G45 + G67
forms an R-basis of (g2 ) . So dim(g2 ) = 6 = 3 + 3 = dim(sp(1) sp(1)). Hence
is onto.
20
or = .
However
/ z(G2 ) (Theorem 1.10.1). Hence = 1.
1.12. Complex exceptional Lie group G2 C
Definition. The group G2 C is defined to be the automorphism group of the
complex Cayley algebra CC :
G2 C = { IsoC (CC ) | (xy) = (x)(y)}.
Lemma 1.12.1. For G2 C , we have
(x, y) = (x, y),
x, y CC .
x CC .
Now we have
1
1
((x)(y) + (y)(x)) = ((x)(y) + (y)(x))
2
2
1
= ((xy + yx) = ((x, y)) = (x, y).
2
(x, y) =
For HomC (CC ), we denote the complex conjugate transpose of with respect to
the inner product hx, yi by : h x, yi = hx, yi.
Lemma 1.12.2. (1) For G2 C , we have = 1 G2 C .
This algebra C is called the split Cayley algebra, and is isomorphic to (CC ) = {x
CC | x = x} as algebras. Now, the group G2(2) is defined to be the automorphism
group of the split Cayley algebras C :
G2(2) = { IsoR (C ) | (xy) = (x)(y)}.
and which can also be defined by
22
G2(2) = (G2 C ) = { G2 C | = }.
Theorem 1.13.1. The polar decomposition of the Lie groups G2(2) is given by
G2(2) (Sp(1) Sp(1))/Z 2 R8 .
Proof. In the split Cayley algebra C , the inner product (x, y) is defined as
1
(x, y) = (xy + yx). If we define an inner product (x, y) of C by (x, y) , then
2
(x, y) is a positive definite inner product. For G2(2) , the transpose t of with
respect to the inner product (x, y) is t = 1 G2(2) . Since G2(2) is an algebraic
subgroup of IsoR (C ) = GL(8, R), from Chevalleys lemma, we have
G2(2) (G2(2) O(8)) Rd
x Rn .
Da is called the reflection with respect to the hyperplane orthogonal to a. Its determinant is 1: det(Da ) = 1.
Lemma 1.14.1. The group O(n) is generated by reflections, that is, any A O(n)
can be expressed by the product of finite number of reflections:
A = Dam Da2 Da1 ,
ai S n1 .
ai S n1 .
x, y C.
Moreover, 1 , 2 are determined uniquely for 3 up to the sign, that is, for 3 , such
1 , 2 have to be 1 , 2 or 1 , 2 .
Proof. Since 3 SO(8) is expressed by the product of even number of reflections (Lemma 1.14.1), it is sufficient to show the existence of 1 , 2 only for
3 = Db Da , a, b C, |a| = |b| = 1. Now, since
Da x = x 2(x, a)a = x (xa + ax)a = axa,
x C,
x, y C.
x, y C.
x, y C.
From this, we see that p is a real number. Hence p = 1 because |p| = 1. Thus we
have 1 = 2 = 1 or 1 = 2 = 1.
Lemma 1.14.3. For 1 , 2 , 3 O(8), the relation
(1 x)(2 y) = 3 (xy),
x, y C
(2 x)(3 y) = 1 (xy),
x, y C,
implies
(3 x)(1 y) = 2 (xy),
x, y C.
x, y C,
then 1 , 2 , 3 SO(8).
Proof. Suppose 1
/ SO(8). Since the mapping : C C defined by x = x
belongs to O(8), and since det = 1, we have 1 = 1
SO(8). Using the
1
principle of triality (Theorem 1.14.2) on the element 1 (cf. Lemma 1.14.3), there
exist 2 , 3 SO(8) such that
(1 (1 x))(2 (2 y)) = 3 ((1 x)(2 y)) = 3 (3 (xy)),
x, y C.
x, y C.
x, y C.
(i)
x, y C.
(ii)
for all x C.
x, y C.
x, y C.
1
3
e1 C C. Then 1 3 = 1, so 1 G2 by Proposition 1.14.5.
Let 1 = +
2
2
This 1 is nothing but w of Section 1.9: 1 = w, because
1 (a + m) = 1 (a + m)1 = 1 a1 + 1 2 m = a + 1 m = w(a + m),
for a + m C C 3 = C.
1.15. Spinor group Spin(7)
If we use the principle of triality for SO(7), then there exist
e, SO(8)
satisfying
(x)(e
y) = (xy),
x, y C.
Putting x = 1, we have
ey = y, and so
e = . Then (i) becomes
(x)(e
y) =
e(xy),
x, y C.
(i)
(ii)
e3 = {e
B
SO(8) | (x)(e
y) =
e(xy), x, y C for some SO(7)}.
26
e3 is a compact group.
B
Theorem 1.15.1.
e3 /G2 S 7 ,
B
e3 is connected.
In particular, the group B
e3 SO(7) = G2 .
B
e3 is a subgroup
Proof. S 7 = {a C | |a| = 1} is a 7 dimensional sphere. Since B
e3 acts on S 7 . We shall show that B
e3 acts transitively on S 7 .
of SO(8), the group B
7
To prove this, it is sufficient to show that any b0 S can be transformed to 1 S 7
e3 . Now, we choose any element a1 S 7 such that (1, a1 ) = 0. and
by some B
choose any element a2 S 7 such that (1, a2 ) = (a1 , a2 ) = 0. Let a3 S 7 be the
element determined by a3 b0 = a2 (a1 b0 ). More precisely, if we let a3 = (a2 (a1 b0 ))b0 ,
then a3 S 7 and satisfies (1, a3 ) = (a1 , a3 ) = (a2 , a3 ) = 0. Choose any element
a4 S 7 such that (1, a4 ) = (a1 , a4 ) = (a2 , a4 ) = (a3 , a4 ) = 0. Let a5 , a6 , a7 S 7 be
elements determined by
a5 b0 = a1 (a4 b0 ),
a6 b0 = a2 (a4 b0 ),
a7 b0 = a6 (a1 b0 ).
i = 0, 1, , 7
x, y C.
(i)
i, j = 0, 1, , 7.
Again we need to verify many cases, but here we will only show the following two
examples.
(e1 )((e3 )b0 ) = a1 (a3 b0 ) = a1 (a2 (a1 b0 )) = a1 (a1 (a2 b0 )) = a2 b0
= (e2 )b0 = (e1 e3 )b0 ,
Now, if we put
b i = ai b 0 ,
i = 0, 1, , 7,
eei = bi , i = 0, 1, , 7
belongs to O(8). Since
ex = (x)b0 , it follows from (i) that
(x)(e
y) =
e(xy),
x, y C.
(ii)
Since O(7),
e O(8) satisfy (ii), SO(7) ,
e SO(8) (Lemma 1.14.4). Hence
1
e
e B3 and
e1 = b0 , and so
e b0 = 1. This shows the transitivity. The isotropy
e3 at 1 S 7 is G2 . Indeed, if
e3 satisfies
subgroup of B
e B
e1 = 1, then we have
e
=
e, so
e G2 . Conversely, G2 satisfies B3 and 1 = 1. Thus we have
e3 /G2 S 7 .
the homeomorphism B
e3
B
= Spin(7).
Theorem 1.15.2.
x, y C.
(xy) = (e
x)(e
y),
x, y C.
f3 SO(7) by p(e
We define a mapping p : B
) = . It is not difficult to see that p is a
homomorphism. The principle of triality implies that p is onto and Ker p = {1, 1}.
Next, we shall prove that p is continuous. From Lemma 1.14.3, we have
e (for example, e1 = (e
e2 )(e
isomorphism
e3 /{1, 1}
B
= SO(7).
e 4 satisfies (x)(e
Since an element (,
e, e
) of D
y) =
e(xy), x, y C, we see that
e 4 contains Spin(7) as a subgroup under the identification
D
28
e 4.
Spin(7)
e (,
e, e
) D
e 4 /Spin(7) S 7 .
Proposition 1.16.1.
D
e 4 is connected.
In particular, the group D
e4
D
= Spin(8).
e 4 SO(8) by
Proof. We define a mapping p : D
p(1 , 2 , 3 ) = 1 .
Proof. The proof follows easily from z(SO(8)) = {1, 1} and the principle of
triality.
29
G2 = { Spin(8) | = , S3 }
= { Spin(8) | = , = }
= { Spin(8) | = }
= { Spin(7) | = }.
Moreover we have the isomorphism
Spin(8)/{(1, 1, 1), (1, 1, 1)}
= Spin(8)/{(1, 1, 1), (1, 1, 1)},
that is,
SO(8)
= Ss(8).
Proof. The group multiplication between , , is the same as that of Theorem
1.3.5. Next, we shall show
G2 = { Spin(8) | = }.
If (1 , 2 , 3 ) Spin(8) satisfies (1 , 2 , 3 ) = (1 , 2 , 3 ), then we have 1 =
2 = 3 (= ), that is,
(x)(y) = (xy),
x, y C,
(i)
Put a = 1, then |a| = 1, and put x = 1 and y = 1 in (i), then we have a(y) = (y)
and (x)a = (x), respectively. Hence, we get
a(x) = (x)a
for all x C.
1
X = X(, x) = x3
x2
of the form
x3 x2
2 x1 , i R, xi C.
x1 3
1
(XY + Y X).
2
In J, we define the trace tr(X), an inner product (X, Y ) and a trilinear form tr(X, Y, Z)
respectively by
tr(X) = 1 + 2 + 3 , X = X(, x),
(X, Y ) = tr(X Y ), tr(X, Y, Z) = (X, Y Z).
Moreover, in J, we define a multiplication X Y , called the Freudenthal multiplication, by
X Y =
1
(2X Y tr(X)Y tr(Y )X + (tr(X)tr(Y ) (X, Y ))E),
2
(where E is the 3 3 unit matrix) and a trilinear form (X, Y, Z) and the determinant
detX respectively by
(X, Y, Z) = (X, Y Z),
detX =
1
(X, X, X).
3
For X = X(, x), Y = Y (, y) and Z = Z(, z) J, the explicit forms in the terms
of their entries are as follows.
(X, Y ) =
tr(X, Y, Z) =
3
X
i=1
3
X
(i i + 2(xi , yi )),
(i i i + R(xi yi+1 zi+2 + xi zi+1 yi+2 )
i=1
(X, Y, Z) =
3
X
1
i=1
(i (yi , zi ) + i (zi , xi ) + i (xi , yi )) ,
detX = 1 2 3 + 2R(x1 x2 x3 ) 1 x1 x1 2 x2 x2 3 x3 x3 .
Lemma 2.1.1. The followings hold in J.
(1) (i)
(ii)
(2) (i)
(ii)
X Y = Y X,
X Y = Y X.
1
E X = X, E X = (tr(X)E X), E E = E.
2
The inner product (X, Y ) is symmetric and positive definite.
tr(X, Y, Z) = tr(Y, Z, X) = tr(Z, X, Y ) = tr(X, Z, Y ) = tr(Y, X, Z)
= tr(Z, Y, X).
The similar statement is also valid for (X, Y, Z).
(iii)
(iv)
(3) (i)
(ii)
2 3 x1 x1 x1 x2 3 x3
X X = x1 x2 3 x3 3 1 x2 x2
x3 x1 2 x2 x2 x3 1 x1
x3 x1 2 x2
x2 x3 1 x1 ,
1 2 x3 x3
X = X(, x),
1 0 0
0 0 0
0 0 0
E1 = 0 0 0 , E2 = 0 1 0 , E3 = 0 0 0 ,
0 0 0
0 0 0
0 0 1
0 x 0
0 0 0
0 0 x
F1 (x) = 0 0 x , F2 (x) = 0 0 0 , F3 (x) = x 0 0 .
x 0 0
0 0 0
0 x 0
The tables of the Jordan multiplication and the Freudenthal multiplication among
elements above are given as follows.
Ei Ej = 0, i 6= j
Ei Ei = Ei
1
Ei Fj (x) = Fj (x), i 6= j
Ei Fi (x) = 0
2
Fi (x) Fi (y) = (x, y)(Ei+1 + Ei+2 ), Fi (x) Fi+1 (y) = 1 Fi+2 (xy),
2
32
Ei Ei+1 = Ei+2
2
Ei Fj (x) = 0, i 6= j
Ei Ei = 0
1
Ei Fi (x) = Fi (x)
(i)
(ii)
for all X J.
(3) X Y =
(X Y ),
(4) X X = t 1 (X X),
for all X, Y, Z J.
for all X, Y J.
for all X J.
Proof. (1) (2) det (X) = det X implies that (X, X, X) = (X, X, X).
Putting X + Y + Z in place of X and comparing the coefficient of , we obtain
(2).
(2) (1) is evident.
(Y Y ) t 1 (Y Y ) = (Y Y ) (Y Y ) (Lemma 2.2.2.(4))
= (det Y )Y (Lemma 2.1.1.(3)) = (det Y )Y = ((det Y )Y )
= ((Y Y ) (Y Y )), Y J.
34
(2) Case det X = 0. If det (t 1 X) 6= 0, we can use the result of (1). If we put t X
instead of X in det (t 1 X) = detX of (1), then det t 1 (t X)) = det (t X). Hence
0 = det X = det t X) 6= 0 which is a contradiction. Thus we have det (t 1 X) =
det (t X) = 0, so det (t 1 X) = det (t X) = det X is also valid.
Lemma 2.2.4. For IsoR (J), the following five conditions are equivalent.
(1)
(2)
(X Y ) = X Y .
(3)
(4)
(5)
(X Y ) = X Y .
(2) (3) Since we have already shown (2) (1), we can use tr(X) = tr(X)
(Lemma 2.2.1.(2)). Now,
3
1
3det(X) = tr(X, X, X) tr(X)(X, X) + tr(X)3
2
2
1
3
3
= tr(X, X, X) tr(X)(X, X) + tr(X) = 3detX.
2
2
(3) (5) ((X Y ), Z) = (X Y, Z) = (X, Y, Z) = (X, Y, Z) (Lemma
2.2.2) = (X Y, Z) holds for all Z J, so we have X Y = (X Y ).
(5) (4)
1
(tr(X)E X),
2
if we denote E = P , then
X P =
1
1
tr(X)P X,
2
2
X J.
0 = 1 1,
3 = 2 ,
2 = 3 ,
p1 = p1 ,
0 = p2 ,
0 = p3 .
1
1
relation (X, Y, Z) = tr(X, Y, Z) tr(X)(Y, Z) tr(Y )(Z, X)
2
2
1
tr(X)tr(Y )tr(Z) and (X, Y, Z) = (X, Y, Z), we obtain
2
tr(X, Y, Z) = tr(X, Y, Z).
1 x3
e x3 2
x2 x1
x3
2
x1
x2
x1 .
3
Then
e F4 . So we identify G2 with
e F4 : G2 F4 .
2.3. Lie algebra f4 of F4
= { IsoR (J) | t 1 (X Y ) = X Y }.
36
Lemma 2.3.1. The Lie algebra e6(26) of the group E6(26) is given by
e6(26) = { HomR (J) | (X, X, X) = 0}
Proof. It is easy to verify that these conditions in e6(26) are equivalent (see
Lemma 2.2.2). Now, if HomR (J) satisfies ((exp t)X, (exp t)X, (exp t)X) =
(X, X, X) for all t R, then we have (X, X, X) = 0 by putting t = 0 after differentiating with respect to t. Conversely, if HomR (J) satisfies t (X Y ) = X
Y +X Y , then it is easy to verify that = exp t satisfies t 1 (X Y ) = X Y .
Theorem 2.3.2. The Lie algebra f4 of the group F4 is given by
f4 = { HomR (J) | (X Y ) = X Y + X Y }
n
tr(X, Y, Z) + tr(X, Y, Z) + tr(X, Y, Z) = 0 o
= HomR (J)
(X, Y ) + (X, Y ) = 0
= { HomR (J) | (X, X, X) = 0, (X, Y ) + (X, Y ) = 0}
= { HomR (J) | (X, X, X) = 0, E = 0}
= { HomR (J) | (X Y ) = X Y + X Y }.
Lemma 2.3.3.
[J, J] M .
e:JJ
Since [M , J] J, any element A M induces an R-linear mapping A
defined by
e = 1 [A, X], X J.
AX
2
Lemma 2.3.4. For X J, there exists a C0 such that
[X, XX] = aE.
Proof. Let X = xij , xij C, xij = xji . The (i, j)-entry aij of [X, XX] =
X(XX) (XX)X is given by
X
X
aij =
(xik (xkl xlj ) (xik xkl )xlj ) =
{xik , xkl , xlj }, i, j = 1, 2, 3.
k,l
k,l
37
however they are equal, that is, a11 = a22 = a33 ( = a). Hence we have [X, XX] = aE.
Since X, XX J, we have [X, XX] M (Lemma 2.3.3). Therefore, (aE) = aE
and so aE = aE, which imply that a = a.
To prove the following Proposition 2.3.6, in M (3, C), we define a real valued symmetric inner product (X, Y ) by
(X, Y ) =
1
tr(XY + Y X ).
2
1
= R(tr((Y Z)X + X (Z Y ))) = (Y Z, X).
2
e f4 .
Proposition 2.3.6. For A M , tr(A) = 0, we have A
X J.
0 0
A1 (a) = 0 0
0 a
0 0
0
a , A2 (a) = 0 0
0
a 0
a
0 ,
0
0
A3 (a) = a
0
a
0
0
0
0.
0
ei (a)Ei = 0
ei (a)Fi (x) = (a, x)(Ei+1 Ei+2 )
A
A
ei (a)Ei+1 = 1 Fi (a)
ei (a)Fi+1 (x) = 1 Fi+2 (ax)
A
A
2
2
d4 = { f4 | Ei = 0, i = 1, 2, 3}
is isomorphic to the Lie algebra D4 = so(8):
D4 = {D HomR (C) | (Dx, y) + (x, Dy) = 0}
under the correspondence
D4 D1 d4
given by
1
x3
x2
x3
2
x1
0
x2
x1 = D3 x3
3
D2 x2
D3 x3
0
D1 x1
D2 x2
D1 x1 ,
0
0
D3 x3 D2 x2
2 3 x1 x1
= D3 x3
0
D1 x1 , x1 x2 3 x3
0
x3 x1 2 x2
D2 x2 D1 x1
x1 x2 3 x3
3 1 x2 x2
x2 x3 1 x1
x3 x1 2 x2
x2 x3 1 x1
1 2 x3 x3
x, y C.
1
Hence Di D4 , i = 1, 2, 3. Moreover, by applying on F1 (x) F2 (y) = F3 (xy), we
2
see that
(D1 x)y + x(D2 y) = D3 (xy), x, y C.
This shows that is onto. Thus Proposition 2.3.7 is proved.
Theorem 2.3.8. Any element f4 is uniquely expressed by
e
= D + A,
D d4 , A M , diagA = 0,
where diagA = 0 means that all diagonal elements aii of A are 0. In particular, the
dimension of f4 is
dim f4 = 28 + 24 = 52.
Proof. Applying f4 on Ei Ei = Ei and Ei Ej = 0, i 6= j, we have
2Ei Ei = Ei ,
Ei Ej + Ei Ej = 0.
40
0 a3 a2
0 a3
0
0
0 a2
E1 = a3 0
0 , E2 = a3 0 a1 , E3 = 0
0 a1 ,
a2
0
0
0 a1 0
a2 a1 0
0
a3 a2
ai C. If we construct a matrix A = 2 a3 0
a1 using these elements ai ,
a2 a1 0
then A M with diagA = 0, and we have
e i,
Ei = AE
i = 1, 2, 3.
e i = 0, i = 1, 2, 3, hence A = 0 and so D = 0.
Certainly, if we apply it on Ei , then AE
Finally, we have dim f4 = 28 + 24 = 52 from the expression above. Thus the theorem
is proved.
2.4. Simplicity of f4 C
Let JC = {X1 + iX2 | X1 , X2 J} be the complexification of the Jordan algebra
J. In the same manner as in J, in JC we can also define the multiplications X Y ,
X Y , the inner product (X, Y ), the trilinear forms tr(X, Y, Z), (X, Y, Z) and the
determinant detX. They have the same properties as those of J. JC is called the
complex exceptional Jordan algebra. JC has two complex conjugations, namely,
X1 + iX2 = X 1 + iX 2 , (X1 + iX2 ) = X1 iX2 ,
Xi J.
X, Y JC .
= { HomC (JC ) | t (X Y ) = X Y + X Y },
f4 C = { HomC (JC ) | (X Y ) = X Y + X Y }
= { HomC (JC ) | (X Y ) = X Y + X Y }.
These are the complexification of the Lie algebras e6(26) and f4 respectively. Then the
properties of e6(26) and f4 stated in Section 2.3 also hold for e6 C and f4 C . Hereafter
we will describe f4 C , but the statements are also valid for f4 using J instead of JC .
41
e : JC JC by
For A JC , we define a C-linear mapping A
e = A X,
AX
X JC .
e e6 C .
Proposition 2.4.1. (1) For A JC , tr(A) = 0, we have A
e B]
e f4 C .
(2) For A, B JC , we have [A,
e X, X) = (A X, X X) = (A, X (X X))
Proof. (1) (AX,
e e6 C .
Hence A
X JC .
h
i
1
1
A tr(A)E , B tr(B)E
e6 C ,
3
3
e B]E
e = A(
e BE)
e B(
e AE)
e = A B B A = 0.
[A,
C
e B]
e f4 (Theorem 2.3.2).
Hence [A,
(2)
e B]
e =
[A,
ei+1 (a),
ei , Fei+1 (a)] = 1 A
[E
2
[Fei (a), Fei (b)] d4 C ,
ei (a)] = A
ei (Di a),
[D, A
ei (a), A
ei (b)] d4 C ,
[A
ei , Fei (a)] = 0,
[E
ei , Fei+2 (a)] = 1 A
ei+2 (a),
[E
2
1e
[Fei (a), Fei+1 (b)] = A
i+2 (ab),
2
where D = (D1 , D2 , D3 ) d4 C ,
ei+2 (ab),
ei (a), A
ei+1 (b)] = 1 A
[A
2
ei+1 (ba).
ei (a), A
ei+2 (b)] = 1 A
[A
2
P7
Proof. (1) Let 0 6= x = i=0 xi ei , xi C. Suppose xi 6= 0 (moreover we may
assume i 6= 0). Then we have
G
i0
xi e0 x0 ei
x
Gj0 (j6=i)
xi ej
xj 1 G0j
e0 = 1.
42
D d4 C , ai CC ,
e A
e A
e = d4 C A
e .
f4 C = d4 C A
1
2
3
e C . Now, let a be
e C be projections of f C = d4 C A
Let p : f4 C d4 C and q : f4 C A
4
a non-zero ideal of f4 C . Then p(a) is an ideal of d4 C . Indeed, if D p(a), then there
3
X
ei (ai ) a. For any D d4 C , we have
A
exist ai CC , i = 1, 2, 3, such that D +
i=1
3
3
i
h
X
X
ei (Di ai ) (Lemma 2.4.2),
ei (ai ) = [D , D] +
A
A
a D , D +
i=1
i=1
hence [D , D] p(a).
i = 1, 2, 3.
ei (ai )] = A
ei (Dai ) (Lemma 2.4.2) and d4 C CC = CC
The last equation follows from [D, A
C
e A
e C A
e C = f C.
(Lemma 2.4.3.(3)). Hence a d4 C A
1
e C a a.
e1 (a1 ) + A
e2 (a2 ) + A
f3 (a3 ) A
A
e1 (1), A
e1 (e1 )] d4 C a
0 6= [A
e1 (1), A
e1 (e1 )]F2 (1) = 2F2 (e1 )), so the case is
(as for the first inequality, note that [A
e1 (1), then
reduced to the case (1). If b 6= 0, then take the Lie bracket with A
e3 (b) + A
e2 (c) a
A
and then
e3 (1) + A
e2 (c ) a
A
under some actions of d4 C . If c = 0, then this can be reduced to the case (1) as in
the above. The case c 6= 0 can also be reduced to the case (1) by the same argument
as above. Thus we have a = f4 C , which proves the simplicity of f4 C .
Lemma 2.4.5. (1) For f4 C and A, B JC , we have
f B]
f
e B]
e ] = [A,
e + [A,
e B].
[, [A,
X
ei , B
ei ], Ai , Bi JC .
[A
(2) Any element f4 C is expressed as =
i
e B]
e ]X = [A,
e B]X
e [A,
e B]X
e
Proof. (1) [, [A,
= (A (B X) B (A X)) (A (B X) B (A X))
= A (B X) + A (B X) + A (B X) B (A X)
B (A X) B (A X) A (B X) + B (A X)
= A (B X) B (A X) + A (B X) B (A X)
f B]X
f
e + [A,
e B]X,
= [A,
X JC .
P e e
(2) By (1) we see that a = { i [Ai , Bi ] | Ai , Bi JC } is an ideal of f4 C . From the
simplicity of f4 C , we have a = f4 C .
ei (a)Fi+2 (1) = Fi+1 (a), etc., we have Ei Ei+1 , Fi (a) W and hence we have
and A
W = J0 C . Now, in the general case, choose a non-zero element X from W:
0 6= X = (E1 E2 ) + (E2 E3 ) + F1 (x1 ) + F2 (x2 ) + F3 (x3 ) W.
44
Proof.
e B]C,
e D) = ([C,
e D]A,
e
([A,
B).
e B]C,
e D) = (A (B C), D) (B (A C), D)
([A,
= (B C, A D) (A C, B D)
e D]A,
e
= (C (D A), B) (D (C A), B) = ([C,
B).
f4 C , A, B JC .
(1 , 2 )4 =
X
i,j
ei , B
ei ]Cj , Dj ) =
([A
X
i,j
ej , D
e j ]Ai , Bi ),
([C
P e e
P e e
C
for 1 = i [A
i , Bi ], 2 =
j [Cj , Dj ], Ai , Bi , Cj , Dj J (Lemma 2.4.5.(2)). Then,
Lemma 2.5.1 shows that the definition of the inner product (1 , 2 )4 is independent
of the choice of the expressions of 1 , 2 , and that (1 , 2 )4 is symmetric.
45
J.
, i f4 C .
e B],
e 2 = [C,
e D],
e A, B, C, D
Proof. It is sufficient to show the lemma for 1 = [A,
e B]
e ], [C,
e D])
e 4 + ([A,
e B],
e [, [C,
e D]
e ])4
([, [A,
f B]
f [C,
f D]
f 4 (Lemma 2.4.5.(1))
e + [A,
e B],
e D])
e 4 + ([A,
e B],
e [C,
e + [C,
e D])
= ([A,
f B]C,
f
e D) + ([A,
e B]C,
e B]C,
e
e B]C,
e D)
= ([A,
D) + ([A,
D) + ([A,
B4 (1 , 2 ) = 9(1 , 2 )4 = 3tr(1 2 ),
e1 (1), [A
e1 (1), Gi0 ] ] = [A
e1 (1), A
e1 (ei )] = Gi0 , i 6= 0
[A
e1 (1), [A
e1 (1), A
e1 (ei )] ] = [A
e1 (1), Gi0 ] = A
e1 (ei ), i 6= 0
[A
e2 (ei ),
e1 (1), A
e3 (ei )] = 1 A
e1 (1), [A
e1 (1), A
e2 (ei )] ] = 1 [A
[A
2
4
e1 (1), [A
e1 (1), A
e3 (ei )] ] = 1 [A
e3 (ei ),
e1 (1), A
e2 (ei )] = 1 A
[A
2
4
the others = 0.
Hence we have
e1 (1))2 ) = (1) 7 2 +
B4 (, ) = tr((adA
46
1
8 2 = 18.
4
Therefore k = 3.
1
1
21+
8 2 = 6.
2
4
e B]
e = 0 for all B J0 C . Then 0 = (, [A,
e B])
e 4 =
Proof. Assume that [A,
C
C
C
C
e A])
e 4 = (B, A) for any f4 . Since f4 J0 = J0 (Proposition 2.4.6.(2)),
(, [B,
we have (J0 C , A) = 0, so that A = 0.
2.6. Roots of f4 C
Before we obtain the roots of the Lie algebra f4 C , we recall the roots of the Lie
algebra D4 C :
D4 C = {D HomC (CC ) | (Dx, y) + (x, Dy) = 0}.
If we let Hk = iGk4+k for k = 0, 1, 2, 3, then
h = {H =
3
X
k=0
i Hk | k C}
0 k < l 3.
The root vectors associated with these roots are respectively given by
k l
k + l
k + l
k l
1 1
1 1
1
= =
1
2 1
1 1
1 1
1 1
,
1
1
1
1
1 1 1
1 1
1 1
=
1
1 1
2
1 1 1
1
1
1
1
1
1
( 0 + 1 + 2 3 ),
(0 1 + 2 3 ),
2
2
1
1
(0 + 1 + 2 + 3 ),
( 0 1 + 2 + 3 ),
2
2
1
1
(0 + 1 + 2 3 ),
( 0 1 + 2 3 ),
2
2
1
1
( 0 + 1 + 2 + 3 ),
(0 1 + 2 + 3 ).
2
2
Proof. We use the decomposition in Theorem 2.4.4:
Let h =
e C A
e C A
e C.
f4 C = d4 C A
1
2
3
3
n
o
X
H =
k Hk | k C d4 C f4 C . Since h is a Cartan subalgebra of
k=0
0k<l3
e1 (a)] = A
e1 (Ha) (Lemma 2.4.2), where
are also roots of f4 C . Furthermore, from [H, A
H(ek + ie4+k ) =
3
X
j=0
e2 (a)] = A
e2 ((H)a),
an associated root vector. Next, by Lemma 2.4.2 we have [H, A
where
H =
3
X
k Hk
k=0
1
= (0 (H0 + H1 H2 + H3 ) + 1 (H0 + H1 + H2 H3 )
2
+2 (H0 + H1 + H2 + H3 ) + 3 (H0 H1 + H2 + H3 ))
1
1
= (0 1 + 2 3 )H0 + (0 + 1 + 2 3 )H1
2
2
1
1
+ (0 + 1 + 2 + 3 )H2 + (0 1 + 2 + 3 )H3 ,
2
2
e2 (ek +
and so we see that coefficients of H0 , H1 , H2 , H3 are roots of f4 C and that A
i + ie4+k ) is associated root vector for 0 k 3. The roots above with negative sign
e2 (ek ie4+k ) are associated root vectors. Finally, from
are also roots of f4 C and A
e3 (a)] = A
e3 ((H)a) (Lemma 2.4.2), where
[H, A
1
1
(0 + 1 2 + 3 )H0 + (0 + 1 + 2 3 )H1
2
2
1
1
+ (0 + 1 + 2 + 3 )H2 + (0 1 + 2 + 3 )H3 ,
2
2
H =
1
(0 1 2 + 3 )
2
i
3
4
i
3
2
4
= 1 + 2 +
=
2 +
=
= 1 + 22 +
49
3
3
3
33 + 24 ,
0 1
0 2
0 + 3
1 2
1 + 3
2 + 3
=
=
=
=
=
=
0 + 1
0 + 2
0 + 3
1 + 2
1 + 3
2 + 3
=
=
=
=
=
=
1
1 +
1
1
2
2 + 23 + 24
2
+ 2 + 23 + 24
+ 22 + 23 + 24
1 + 22
1 + 2
21 + 32
2
1 + 32
1 + 22
1
( 0 + 1 + 2 + 3 )
2
1
(0 1 2 + 3 )
2
1
( 0 + 1 2 + 3 )
2
1
( 0 1 + 2 + 3 )
2
1
(0 1 + 2 + 3 )
2
1
(0 + 1 2 + 3 )
2
1
( 0 1 2 + 3 )
2
1
(0 + 1 + 2 + 3 )
2
+
+
+
+
+
+
23
23
43 + 24
23
43 + 24
43 + 44 ,
= 1 + 22 + 33 + 4
=
= 1 + 22 + 23 + 4
= 1 +
2 + 23 + 4
3 + 4
2 + 3 + 4
1 + 2 + 3 + 4
2 + 23 + 4 .
3
X
k=0
k k ,
H=
3
X
k=0
k Hk , H =
3
X
k=0
k Hk hR .
Indeed, since
HEi = 0, i = 1, 2, 3,
HF1 (x) = F1 (Hx), HF2 (x) = F2 ((H)x), HF3 (x) = F3 ((H)x),
we have
50
B4 (H, H ) =
3
X
k l B4 (Hk , Hl )
k,l=0
X
= 3 (k l (tr(Hk Hl ) + tr((Hk )(Hl )) + tr((Hk )(Hl ))))
k,l
3
X
X
= 3 k l (2 + 2 + 2)kl = 18
k k .
k=0
k,l
1
(H1 H2 ),
18
1
=
(H0 H1 H2 + H3 ).
36
H1 =
H2 =
H3
H4
Hence we have
1 1
1
(1 + 1) =
18 18
9
and the other inner products are similarly calculated. Hence, the inner products
induced by the Killing form B4 between 1 , 2 , 3 , 4 and are given by
(1 , 1 ) = B4 (H1 , H1 ) = 18
(1 , 1 ) = (2 , 2 ) =
1
,
9
1
,
18
1
(3 , 4 ) = ,
36
(3 , 3 ) = (4 , 4 ) =
1
1
, (2 , 3 ) = ,
18
18
(1 , 3 ) = (1 , 4 ) = (2 , 4 ) = 0,
1
1
(, ) = , (, 1 ) = , (, i ) = 0, i = 2, 3, 4,
9
18
using them, we can draw the Dynkin diagram and the extended Dynkin diagram of
f4 C .
(1 , 2 ) =
Theorem 2.7.1.
{ F4 | Ei = Ei , i = 1, 2, 3}
= Spin(8).
Proof. We recall the group Spin(8) of Theorem 1.16.2 and define a mapping
: Spin(8) D4 = { F4 | Ei = Ei , i = 1, 2, 3} by
1
3 x3 2 x2
1 x3 x2
(1 , 2 , 3 ) x3 2 x1 = 3 x3
2
1 x1 .
x2 x1 3
3
2 x2 1 x1
We first prove that = (1 , 2 , 3 ) D4 . Indeed, the fact that det (X) = det X
can be seen by observing that
R((1 x1 )(2 x2 )(3 x3 )) = R((3 (x1 x2 ))3 x3 )
= (3 (x1 x2 ), 3 x3 ) = (x1 x2 , x3 ) = R(x1 x2 x3 ),
which together with E = E, shows that F4 and Ei = Ei , i = 1, 2, 3. Therefore
D4 . Certainly is a homomorphism. We shall show that is onto. Let D4 .
We put
Ji = {Fi (x) | x C} = {X J | 2Ei+1 X = 2Ei+2 X = X},
i = 1, 2, 3.
x, y C.
1
Hence i O(8), i = 1, 2, 3. Moreover, by applying on F1 (x) F2 (y) = F3 (xy),
2
we see that
(1 x)(2 y) = 3 (xy),
x, y C.
o
n 0 0 0
J01 = {X J | E1 X = 0, tr(X) = 0} = 0 x R, x C ,
0 x
52
J23
n 0
= {Y J | 2E1 Y = Y } = y 3
y2
y3
0
0
y2
o
0 y2 , y3 C .
0
Lemma 2.7.2. Suppose that we are given an element A J01 such that (A, A) =
2. Choose any element X0 J01 such that (X0 , X0 ) = 2, (A, X0 ) = 0. Choose any
element Y0 J23 such that (Y0 , Y0 ) = 2, 2A Y0 = Y0 . Let
Z0 = 2X0 Y0 .
Choose any element X1 J01 such that (X1 , X1 ) = 2, (A, X1 ) = (X0 , X1 ) = 0.
Choose any element X2 J01 such that (X2 , X2 ) = 2, (A, X2 ) = (X0 , X2 ) =
(X1 , X2 ) = 0. Let
Y1 = 2Z0 X1 ,
Z2 = 2X2 Y0 ,
X3 = 2Y1 Z2 .
Finally, choose any element X4 J01 such that (X4 , X4 ) = 2, (A, X4 ) = (X0 , X4 ) =
(X1 , X4 ) = (X2 , X4 ) = (X3 , X4 ) = 0. Let
Z4 = 2X4 Y0 , Y2 = 2Z0 X2 , Y3 = 2Z0 X3 ,
X5 = 2Y1 Z4 , X6 = 2Y2 Z4 ,
X7 = 2Y3 Z4
and moreover let
Yi = 2Z0 Xi ,
Zi = 2Xi Y0 ,
i = 4, 5, 6, 7,
i = 1, 3, 5, 6, 7.
(E2 E3 ) = A,
F3 (ei ) = Zi , i = 0, 1, , 7
we have (E2 E3 ) = A, and which shows the transitivity. We determine the isotropy
subgroup of (F4 )E1 at E2 E3 S 8 . Let (F4 )E1 satisfy (E2 E3 ) = E2 E3 .
Since (F4 )E1 satisfies E1 = E1 and E = E, it also satisfies (E2 + E3 ) = E2 +
E3 . Therefore we have E2 = E2 and E3 = E3 , so that Spin(8). Conversely,
Spin(8) satisfies (E2 E3 ) = E2 E3 . Thus we have the homeomorphism
(F4 )E1 /Spin(8) S 8 .
Remark. If we know the dimension of the group (F4 )E1 , without using Lemma
2.7.2, Proposition 2.7.3 can be simply proved as follows. The group (F4 )E1 acts on S 8 .
The isotropy subgroup of (F4 )E1 at E2 E3 is Spin(8) and dim((F4 )E1 /Spin(8)) =
dim(F4 )E1 dim Spin(8) = 36 28 = dim S 8 . Therefore, we have (F4 )E1 /Spin(8)
S8.
Theorem 2.7.4.
(F4 )E1
= Spin(9).
Spin(8)
p
SO(8)
(F4 )E1
p
SO(9)
S8
=
S8
we see that p : (F4 )E1 SO(9) is onto by the five lemma. Ker p = {1, }, where =
(1, 1, 1) (which corresponds to defined in the following Section 2.9). Indeed,
let Ker p, then satisfies X = X for all X J01 . From (E2 E3 ) = E2 E3
follows that Spin(8). Let = (1 , 2 , 3 ) Spin(8). Since F1 (x) = F1 (x),
that is, F1 (1 x) = F1 (x), so 1 x = x for all x C, hence we have 1 = 1. From the
principle of triality, we have = (1, 1, 1) = 1 or = (1, 1, 1) = . Hence we have
Ker p = {1, }, and so we have the isomorphism
(F4 )E1 /{1, }
= SO(9).
Therefore (F4 )E1 is isomorphic to the group Spin(9) as the universal covering group
of SO(9).
54
Spin(9)/Spin(7) S 15 .
Theorem 2.7.5.
(Spin (7) and Spin(7) are conjugate in the group O(7). Indeed, a mapping f :
Spin (7) Spin(7),
f ( ) = 1 , (x = x, x C)
gives the conjugation. In particular, Spin (7)
= Spin(7))
2.8. Connectedness of F4
We denote by (F4 )0 the connected component of F4 containing the identity 1.
Lemma 2.8.1. For
Y (, y), where
y1
y2
y3
= 1
2 3
(a, x1 )
2 + 3
+
cos 2|a| +
sin 2|a|
=
2
2
|a|
2 3
(a, x1 )
2 + 3
cos 2|a|
sin 2|a|,
=
2
2
|a|
(2 3 )a
2(a, x1 )a
= x1
sin2 |a|
sin 2|a|
2|a|
|a|2
x3 a
sin |a|
= x2 cos |a|
|a|
ax2
sin |a|
= x3 cos |a| +
|a|
sin |a|
if a = 0, then
means 1 , then 1 (a) (F4 )0 .
|a|
0 0 0
e1 (a) f4 (Proposition 2.3.6) and
Proof. For A1 (a) = 0 0 a , we have A
0 a 0
55
1
X = 0
0
0
0 ,
3
i R.
which is the maximal value and attains at some t > 0. This contradicts the maximum
of 1 2 + 2 2 + 3 2 . Hence x1 = 0. x2 = x3 = 0 can be similarly proved by constructing
2 (a), 3 (a) (F4 )0 analogous to 1 (a) of Lemma 2.8.1. Hence X0 is of diagonal form.
We now give the proof of
the latter halfof the proposition. If X J is transformed
1 0 0
to a diagonal form X = 0 2 0 by (F4 )0 , then
0 0 3
P3
tr(X) = tr(X) (Lemma 2.2.1.(2)) = i=1 i ,
P3
(X, X) = (X, X) (Lemma 2.2.4) = i=1 i 2 ,
P3
tr(X, X, X) = tr(X, X, X) (Lemma 2.2.4) = i=1 i 3 .
56
Hence, 1 , 2 , 3 are uniquely determined (up to order) as the solutions of the following
simultaneous equation:
1 + 2 + 3 = tr(X)
1 2 + 2 2 + 3 2 = (X, X)
3
1 + 2 3 + 3 3 = tr(X, X, X).
F4 /Spin(9) CP2 .
1 0 0
X = 0 2 0 , i R
0 0 3
(Proposition 2.8.2). From the
is,
2
1
0
0
0
1
0 = 0
3 2
0
0
2
0
0
0 .
3
0 1 0
: J J, X = T XT 1, where T = 1 0 0 SO(3), then (F4 )0 and
0 0 1
E2 = E1 . Hence X = E1 . In the case X = E3 , the situation is similar to the
above. Therefore the transitivity is proved. Since we have CP2 = (F4 )0 E1 , CP2 is
connected. Now, the group F4 acts transitively on CP2 and the isotropy subgroup
of F4 at E1 CP2 is Spin(9) (Theorem 2.7.4). Thus we have the homeomorphism
F4 /Spin(9) CP2 . Finally, the connectedness of F4 follows from the connectedness
of CP2 and Spin(9).
2.9. Involution and subgroup Spin(9) of F4
57
1 x3 x2
1
x3 x2
x3 2 x1 = x3
2
x1 .
x2 x1 3
x2 x1
3
0
n 1 0
J = {X J | X = X} = 0 2 x1
0 x1 3
o
i R, x1 C
= J(2, C) E1 ,
J = {X J | X = X} =
= {X J | 2E1 X = X}.
0
x3
x2
x3
0
0
x2
o
0 xi C
0
X1 J , X2 J
X J.
Thus = , and so (F4 ) . Therefore we have shown that (F4 ) = (F4 )E1
=
Spin(9) (Theorem 2.7.4).
2.10. Center z(F4 ) of F4
Theorem 2.10.1. The center z(F4 ) of the group F4 is trivial:
z(F4 ) = {1}.
Proof. Let z(F4 ). From the commutativity with : = , we have
Spin(9) (Theorem 2.9.1) and so z(Spin(9)). Since the center z(Spin(9)) of
Spin(9) is the group of order 2, we have
=1
or
= .
However,
/ z(F4 ) (Theorem 2.9.1), which implies that = 1.
According to a general theory of compact Lie groups, it is known that the center
of the simply connected simple Lie group of type F4 is trivial. Hence F4 has to be
simply connected. Thus we have the following Theorem.
Theorem 2.10.2. F4 = { IsoR (J) | (XY ) = XY } is a simply connected
compact Lie group of type F4 .
Remark. If we know that the space CP2 is simply connected, the simply connectedness of the group F4 follows from F4 /Spin(9) CP2 of Theorem 2.8.3.
2.11. Involution and subgroup (Sp(1) Sp(3))/Z 2 of F4
We define an involutive R-linear transformation of J by
1 x3 x2
1 x3 x2
x3 2 x1 = x3 2 x1 .
x2 x1 3
x2 x1 3
59
1
X = x3
x2
X J is expressed by
x3 x2
1 m3
2 x1 = m3 2
x1 3
m2 m1
0
m2
m1 + a3 e4
3
a2 e 4
a3 e 4
0
a1 e4
a2 e4
a1 e 4 ,
0
1 m3 m2
m3 2 m1 + (a1 , a2 , a3 ).
m2 m1 3
(In J(3, H) the multiplication M N and the inner product (M, N ) are analogously
1
defined as in J, and in H 3 the inner product (a, b) is defined naturally by (a b +
2
b a)). Since these operations correspond to their respective operations in J, hereafter,
we identify J(3, H) H 3 with J, that is,
J(3, H) H 3 = J.
The group F4,H is defined to be the automorphism group of the Jordan algebra
JH = J(3, H) (in which multiplications and are analogously defined as in J):
F4,H = { IsoR (JH ) | (M N ) = M N }
= { IsoR (JH ) | (M N ) = M N }.
Proposition 2.11.1.
F4,H
= Sp(3)/Z 2 , Z 2 = {E, E}.
M JH .
projective
ai1
Ei = Ai Ei Ai . Let ai = ai2 be the i-th column vector of Ai , then we have
ai3
i = 1, 2, 3.
m, n H.
(i)
3 m = (1 m)q.
m, n H,
qr 0 0
Construct a matrix B = 0 pr 0 , then B Sp(3) and we have
0 0 r
M = BM B ,
M JH ,
AB Sp(3).
Therefore is onto. Ker = {E, E} can be easily obtained. Thus we have the
isomorphism Sp(3)/Z 2
= F4,H .
Theorem 2.11.2.
(F4 )
= (Sp(1) Sp(3))/Z 2 , Z 2 = {(1, E), (1, E)}.
61
M + a JH H 3 = J.
m + ae4 H He4 = C,
F4
1 x3 x2
1 x3
w x3 2 x1 = x3 2
x2 x1 3
x2 x1
x2
x1 .
3
1
x3
x2
element
x3
2
x1
x2
x1 ,
3
1
a3
a2
i R, xi = ai + mi C C 3 = C
a3
2
a1
a2
a1 + m1 , m2 , m3
3
m2 n3 m3 n1 m1 n2
M (3, C),
+
+
+
M N =
n2 m3 n3 m1 n1 m2
e
M A N A = (M N ) tA,
where M = mij , N = nij M (3, C).
In J(3, C) M (3, C), we define the multiplication , the inner product ( , ) and
the R-linear transformation w respectively by
1
1
(X + M ) (Y + N ) = (X Y (M N + N M )) (M Y + N X + M N ),
2
2
(X + M, Y + N ) = (X, Y ) + 2(M, N ),
1
3
e1 ,
w(X + M ) = X + 1 M,
1 = +
2
2
63
(in J(3, C), multiplications and are defined as in J). Since these operations
correspond to their respective operations in J, we identify J(3, C) M (3, C) with J,
that is,
J(3, C) M (3, C) = J.
The group F4,C is defined to be the automorphism group of the Jordan algebra
JC = J(3, C) :
F4,C = { IsoR (JC ) | (X Y ) = X Y }
= { IsoR (JC ) | (X Y ) = X Y }.
A SU (3),
and the group SU (3) Z 2 be the semi-direct product of groups SU (3) and Z 2 under
this action.
Proposition
2.12.1. F4,C
= (SU (3)/Z 3 ) Z 2 , Z 3 = {E, 1 E, 1 2 E}, 1 =
1
3
+
e1 .
2
2
Proof. We define a mapping : SU (3) Z 2 F4,C by
(A, 1)X = AXA ,
(A, )X = AXA ,
X JC .
or 1 x = px, 2 x = xq, 3 x = q x p.
qr 0
= qp and construct a matrix B = 0 pr
0 0
or X = BXB ,
0
0 .
r
X JC .
X + M J(3, C) M (3, C) = J.
or X = AXA ,
X JC
(Proposition 2.12.1). In the former case, let = (E, A)1 , then |JC = 1, and so
G2 . Moreover (G2 )e1 = (G2 )w = SU (3) (Theorem 1.9.4), and hence, there
exists P SU (3) such that
(X + M ) = X + P M = (P, E)(X + M ),
X + M JC M (3, C) = J.
Therefore we have
= (E, A) = (E, A)(P, E) = (P, A).
In the latter case, consider the mapping 1 : J J given by 1 (X + M ) = X + M ,
X + M J and remember that 1 G2 F4 . Let = 1 (E, A)1 . Then F4
and |JC = 1, which shows that (G2 )e1 = (G2 )w (Theorem 1.9.4) (F4 )w . Since
, (E, A) (F4 )w , we have 1 (F4 )w , so that 1 (G2 )w which is a contradiction
(Theorem 1.9.4). Consequently the proof of is onto is completed. The fact that
Ker = {(E, E), (1 E, 1 E), (1 2 E, 1 2 E)} = Z 3 can be easily obtained. Thus we
have the isomorphism (SU (3) SU (3))/Z 3
= (F4 )w .
Remark 1. Since (F4 )w is connected as the fixed points subgroup of F4 by an
automorphism of order 3 of the simply connected group F4 (Rasevskii [32]), that
: SU (3) SU (3) (F4 )w is onto is proved as follows. The elements
G01 , G23 , G45 , G67 , G26 + G37 , G27 + G36 ,
G24 + G35 , G25 + G34 , G46 + G57 , G47 + G56 ,
e1 (1), A
e2 (1), A
e3 (1), A
e1 (e1 ), A
e2 (e1 ), A
e3 (e1 )
A
65
= { IsoC (JC ) | (X Y ) = X Y }.
66
1
(XY + Y X).
2
F4(20) = (F4 C ) .
Theorem 2.14.1. The polar decompositions of the Lie groups F4(4) and F4(20)
are respectively given by
F4(4) (Sp(1) Sp(3))/Z 2 R28 ,
z(F4(20) ) = {1}.
67
E6 C
(Lemma 2.2.3), then the definition of the group E6 can be also given by
E6 = { E6 C | () = } = (E6 C ) .
Theorem 3.1.1. E6 is a compact Lie group.
Proof. E6 is a compact Lie group as a closed subgroup of the unitary group
U (27) = U (JC ) = { IsoC (JC ) | hX, Y i = hX, Y i}.
3.2. Lie algebra e6 of E6
Before investigating the Lie algebra e6 of the group E6 , we will study the Lie
algebra e6 C of the group E6 C .
Theorem 3.2.1. (1) The Lie algebra e6 C of the Lie group E6 C is given by
e6 C = { HomC (JC ) | (X, X, X) = 0}.
(2) Any element e6 C is uniquely expressed by
= + Te,
f 4 C , T J0 C .
implies
= 0, T = 0.
where i = i + Tei , i f4 C , Ti J0 C .
f for f C , T J0 C . Now,
Proof. It is sufficient to show that [, Te] = T
4
[, Te]X = (T X) T X = T X + T X T X
f X, X JC .
= T X = T
() = t = t ( + Te) = Te.
Therefore t = Te e6 C .
Theorem 3.2.4. (1) The Lie algebra e6 of the Lie group E6 is given by
e6 = { HomC (JC ) | (X, X, X) = 0, hX, Y i + hX, Y i = 0}.
(2) Any element e6 is uniquely expressed by
= + iTe,
f4 , T J0 .
g
T = + f
T (Lemma 3.2.3), that is, = and T = T . Hence f4 ,
and T is of the form T = iT, T J0 .
Proposition 3.2.5. The complexification of the Lie algebra e6 is e6 C .
Proof. For e6 C , the conjugate transposed mapping of with respect to
the inner product hX, Y i of JC is = t e6 C , and for e6 C , belongs to e6
if and only if = . Now, any element e6 C can be uniquely expressed as
=
+
+i
,
2
2i
+
,
e6 .
2
2i
e6 C = f4 C e
J0 .
C
C
J0 . Now, let a be
J0 be projections of e6 C = f4 C e
Let p : e6 C f4 C and q : e6 C e
a non-zero ideal of e6 C . Then p(a) is an ideal of f4 C . Indeed, if p(a), then there
exists T J0 C such that + Te a. For any 1 f4 C , we have
hence [1 , ] a.
a [1 , + Te] = [1 , ] + g
1 T (Theorem 3.2.2),
C
We shall show that either f4 C a 6= {0} or e
J0 a 6= {0}. Assume that f4 C a = {0}
C
C
J a = {0}.
and e
J a = {0}. Then the mapping p|a : a f4 C is injective because e
0
J0 = e 6 C .
Consequently a f4 C e
C
e C a a.
e (A J0 C ) be a non-zero element of J
(2) Case e
J0 a 6= {0}. Let A
0
e B]
e 6= 0 (Lemma 2.5.4), then 0 6= [A,
e B]
e f4 C a.
Choose B J0 C such that [A,
Hence this case is reduced to the case (1).
70
Therefore we have a = e6 C .
Proposition 3.3.2. (1) JC is a simple Jordan algebra.
(2) JC is an e6 C -irreducible C-module.
o
nX
i Ai | i e6 C , Ai JC = JC .
(3) e6 C JC =
i
1
1
(B, X)A + (A, B)X 2B (A X),
2
6
X JC .
A (B X) + B (X A) + X (A B) tr(A)B X tr(B)A X
1
tr(X)A B + (tr(A)tr(B)X + tr(B)tr(X)A + tr(X)tr(A)B)
2
1
((A, B)X + (B, X)A + (X, A)B) = (A, B, X)E.
2
Therefore, using the above, we have
1
1
1
(B, X)A + (A, B)X 2B (A X) + (A, B)X
2
6
3
1
1
= (B, X)A + (A, B)X 2B (A X) + tr(B)A X + tr(A X)B
2
2
(tr(B)tr(A X) (B, A X))E
1
1
= (B, X)A + (A, B)X 2B (A X) + tr(A)B X + tr(X)B A
2
2
1
tr(A)tr(X)B + (A, X)B + tr(B)A X tr(B)tr(A)X
2
1
1
1
tr(B)tr(X)A + tr(B)tr(A)tr(X)E tr(B)(A, X)E
2
2
2
1
1
+ (tr(A)tr(X) (A, X))B tr(B)(tr(A)tr(X) (A, X))E + (A, B, X)E
2
2
= tr(A)B X + tr(B)X A + tr(X)A B
1
(tr(A)tr(B)X + tr(B)tr(X)A + tr(A)tr(X)B)
2
1
+ ((A, B)X + (B, X)A + (A, X)B) + (A, B, X)E 2B (A X)
2
= A (B X) + B (A X) + X (A B) 2B (A X)
= A (B X) B (A X) + (A B) X
e B]X
e + (A B) X.
= [A,
x2
1 x3 x2
41
x3
1
(E1 E1 ) x3 2 x1 = x3 22 2x1 .
6
x2 x1 3
x2 2x1 23
X, Y JC .
(A B) = B A.
72
X, Y JC .
[, A B] = A B + A B.
X
(Ai Bi ), Ai , Bi JC .
(2) Any element e6 C is expressed by =
i
, i e6 C .
73
g1 g
e
= (([, 1 ] + [Te, Te1 ]) + (T
1 T ), 2 + T2 )6 (Theorem 3.2.2)
= ([, 1 ], 2 )4 + ([Te, Te1 ], 2 )4 + (T1 1 T, T2 )
= (1 , [, 2 ])6 .
(2) For = + Te, f4 C , T J0 C , we have
e B]
e + A B 1 (A, B)E
(, A B)6 = + Te, [A,
3
6
1
e
e
= (, [A, B])4 + T, A B (A, B)E
3
= (A, B) + (TeA, B) = (( + Te)A, B) = (A, B).
Lemma 3.5.2. In e6 C , we have
[(Ei Ei+1 ) , D] = 0, D d4 C ,
ei (a)] = 1 Fei (a),
[(Ei Ei+1 ) , A
2
1
e
[(Ei Ei+1 ) , Ai+1 (a)] = Fei+1 (a),
2
ei+2 (a)] = Fei+2 (a),
[(Ei Ei+1 ) , A
h
i
e2 (ei )] ] = , 1 Fe2 (ei ) = 1 A
e2 (ei ),
[, [, A
2
4
e3 (ei )] ] = [, Fe3 (ei )] = A
e3 (ei ),
[, [, A
i
h
1e
1e
[, [, Fe1 (ei )] ] = , A
1 (ei ) = F1 (ei ),
2
4
h
i 1
1e
e
[, [, F2 (ei )] ] = , A2 (ei ) = Fe2 (ei ),
2
4
e3 (ei )] = Fe3 (ei ),
[, [, Fe3 (ei )] ] = [, A
the others = 0.
Hence
B6 (, ) = tr((ad)2 ) =
1
4 + 1 2 8 = 24.
4
Therefore k = 12. Next, we will calculate tr() as follows.
1
1
Fe1 (ei ) = Fe1 (ei ) = Fe1 (ei ),
2
4
1 e
1e
e
E2 = E2 = E2 , F2 (ei ) = F2 (ei ) = F2 (ei ),
2
4
E3 = 0 = 0,
Fe3 (ei ) = 0 = 0.
E1 = E1 = E1 ,
Hence
tr() = 1 2 +
1
8 2 = 6.
4
Therefore k = 4.
Lemma 3.5.4. The followings hold in e6 C .
(1) A (A A) = 0,
A JC ,
A (B C) + B (C A) + C (A B) = 0,
A, B, C JC .
= 2(, (A A) A)6 .
Hence (, (A A) A)6 = 0 for all e6 C , so that (A A) A = 0, that is,
A (A A) = 0 (Lemma 3.4.3.(2)). If we put A + B + C in the place of A, then
the result follows from the coefficient of .
(2) Assume that A B = 0, that is B A = 0 (Lemma 3.4.3.(2)) for all B JC .
Then for any e6 C , 0 = (, B A)6 = (B, A) (Lemma 3.5.1.(2)). Since e6 C JC =
JC (Proposition 3.3.2.(3)), we have (JC , A) = 0, so that A = 0.
3.6. Roots of e6 C
Let
Mr = {X M (3, C) | all diagonal elements of X are real},
75
0
3 x12 2 x13
1 x12 x13
x21 2 x23 = 3 x21
0
1 x23 .
x31 x32 3
0
2 x31 1 x32
Observe that this mapping is an extension of : JC JC .
Lemma 3.6.1. For d4 C , we have
(X Y ) = X Y + X Y,
X, Y (Mr )C .
31 = 2 ,
13 = 2 ,
12 = 3 ,
21 = 3 .
where T = t T .
T + T
T T
, T2 =
. Then Te1 e6 C
2
2
(Proposition 2.3.6) e6 C . Therefore Te = Te1 + Te2
Proof. We decompose T = T1 + T2 , T1 =
f
e
Proof. (1) (R)X
= R X = (R X) R X (Lemma 3.6.1) = (RX)
C
f
e
e
e
R(X) = [, R]X, X J . Hence, R = [, R].
e Te]X = H
e TeX TeHX
e
(2) [H,
1 e
(H(T X + XT ) Te(HX + XH ))
2
1
= (HT X + XT H + HXT + T XH T HX XH T T XH HXT )
4
(since H M (3, C), products of matrices above are associative)
1
1
= ([H, T ]X + X[H, T ]) = [H, T ] X, X JC .
4
2
=
Theorem 3.6.4. The rank of the Lie algebra e6 C is 6. The roots of e6 C relative
to some Cartan subalgebra are given by
(k l ), (k + l ), 0 k < l 3,
1
k (2 3 ),
0 k 3,
2
1
1
(0 1 + 2 3 ) (3 1 ),
2
2
1
1
( 0 + 1 + 2 3 ) (3 1 ),
2
2
1
1
(0 + 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 3 ) (1 2 ),
2
2
1
1
(0 + 1 + 2 3 ) (1 2 ),
2
2
1
1
( 0 + 1 + 2 + 3 ) (1 2 ),
2
2
1
1
(0 1 + 2 + 3 ) (1 2 ),
2
2
77
with 1 + 2 + 3 = 0.
Proof. We use the decomposition of Theorem 3.3.1.(2):
C
e6 C = f4 C e
J0 .
Let
h=
e e6 C
h = h + H
3
3
X
X
h =
k Hk = k iGk4+k , k C )
k=0
k=0
,
3
X
E
,
C,
=
0
H
=
j j
j
1
2
3
j=1
e ] = hg
e
[h , H
H (Lemma 3.6.3.(1)) = 0 = 0,
e H
e ] = 1 [H, H ] (Lemma 3.6.3.(2)) = 0.
[H,
2
We denote
aEkl
0
example, F23 (a) = 0
0
0
0
0
M
(3, CC ) by Fkl (a): Fkl (a) = aEkl , a CC , k 6= l. For
0
a . Then, we have
0
e Fe23 (a)]
[h, Fe23 (a)] = [h , Fe23 (a)] + [H,
1
= (h F23 (a)) + [H, F23 (a)] (Lemma 3.6.3)
2
1
= Fe23 (h a) + [H, F23 (a)]
2
1
= k + (2 3 ) Fe23 (a),
2
78
1
root of e6 C and Fe23 (ek + ie4+k ) is its root vector. Similarly k + (2 3 ) is a
2
root of e6 C and Fe23 (ek ie4+k ) is its root vector. Again, from
[h, Fe32 (a)] =
1
k + (3 2 ) Fe32 (a),
2
a = ek + ie4+k ,
1
1
we see that k + (3 2 ) is a root of e6 C . Similarly k + (3 2 ) is a root of
2
2
e6 C and Fe32 (ek ie4+k ) is its root vector. Next, using the relation
1 e e
1
[h, Fe31 (a)] = (h F31 (a)) + [H,
F31 (a)] = Fe31 ((h )a) + (3 1 )Fe31 (a),
2
2
1 e e
1
e
e
[h, F12 (a)] = (h F12 (a)) + [H, F12 (a)] = F12 ((h )a) + (1 2 )Fe12 (a),
2
2
where
1
1
(0 1 + 2 3 )H0 + (0 + 1 + 2 3 )H1
2
2
1
1
+ (0 + 1 + 2 + 3 )H2 + (0 1 + 2 + 3 )H3 ,
2
2
1
1
h = (0 + 1 2 + 3 )H0 + (0 + 1 + 2 3 )H1
2
2
1
1
+ (0 + 1 + 2 + 3 )H2 + (0 1 + 2 + 3 )H3
2
2
h =
3
3
2 4
79
=
=
=
=
=
=
1
1
1
0
0
0
0
1
1
1
1
0
0
0
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 + 1
0 + 2
0 + 3
1 + 2
1 + 3
2 + 3
1
0 + (2 3 ) = 1 1 1
2
1
1 + (2 3 ) = 0 1 1
2
1
2 + (2 3 ) = 0 0 1
2
1
3 + (2 3 ) = 0 0 0
2
1
0 (2 3 ) = 1 1 1
2
1
1 (2 3 ) = 0 1 1
2
1
2 (2 3 ) = 0 0 1
2
1
3 (2 3 ) = 0 0 0
2
1
1
(0 1 + 2 3 ) + (3 1 ) =
2
2
1
1
( 0 + 1 + 2 3 ) + (3 1 ) =
2
2
1
1
(0 + 1 + 2 + 3 ) + (3 1 ) =
2
2
1
1
( 0 1 + 2 + 3 ) + (3 1 ) =
2
2
1
1
( 0 + 1 2 + 3 ) + (3 1 ) =
2
2
1
1
(0 1 2 + 3 ) + (3 1 ) =
2
2
1
1
( 0 1 2 3 ) + (3 1 ) =
2
2
1
1
(0 + 1 2 3 ) + (3 1 ) =
2
2
1
1
( 0 1 + 2 3 ) (1 2 ) =
2
2
1
1
( 0 1 2 + 3 ) (1 2 ) =
2
2
1
1
( 0 + 1 + 2 + 3 ) + (1 2 ) =
2
2
80
=
=
=
=
=
=
1
1
1
0
0
0
2
1
1
1
1
0
0 0
0 0
0 0
0 0
1 1
1 1
1 1
1 1
2
2
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0 0
1 0
1 0
1 2
3 1
2 1
0 1
2 1
2 1
1 1
2 1
2 1
1 2
2 1
2 1
0 0
0 0
1 0
1 1
1 0
1 0
0 1
1 0
1 0
1 1
2 1
1 0
1 1
1 1
1 0
0 0
0 0
0 1
1
1
1
1
1
1
1
1
( 0 + 1 2 3 ) (1 2 ) = 1 2
2
2
1
1
(0 + 1 2 + 3 ) (1 2 ) = 0 1
2
2
1
1
(0 + 1 + 2 3 ) (1 2 ) = 0 1
2
2
1
1
( 0 + 1 + 2 + 3 ) (1 2 ) = 1 2
2
2
1
1
(0 1 + 2 + 3 ) (1 2 ) = 0 0
2
2
Hence = {1 , 2 , , 6 } is a fundamental root system
of h is
hR =
3
nX
k Hk +
3
X
j=1
k=0
j Ej
1 1
1 1
1 1
2 2
1 1
0.
o
| k , j R, 1 + 2 + 3 = 0 .
k=0
for h =
3
X
k Hk +
3
X
j Ej
j=1
k=0
3
X
, h =
k Hk +
k=0
3
X
j Ej
j=1
hR . Indeed,
3
3
3
3
X
X
X
4 X
B4
k Hk ,
k Hk + 12
k Ek ,
j Ej
3
j=1
j=1
k=0
4
18
3
3
X
k=0
k k + 12
k=0
3
X
j j (Theorem 2.6.2)
j=1
3
3
X
X
j j .
= 12 2
k k +
k=0
j=1
(1 , 1 ) = B6 (H1 , H1 ) = 24
81
1
1 1
2=
,
24 24
12
and the other inner products are similarly calculated. Hence, the inner product
induced by the Killing form B6 between 1 , 2 , , 6 and are given by
(i , i ) =
1
,
12
i = 1, 2, 3, 4, 5, 6,
(1 , 2 ) = (2 , 3 ) = (3 , 4 ) = (3 , 5 ) = (5 , 6 ) =
otherwise,
1
1
, (, 4 ) = , (, i ) = 0,
(, ) =
12
24
1
,
24
(i , j ) = 0,
i = 1, 2, 3, 5, 6,
using them, we can draw the Dynkin diagram and the extended Dynkin diagram of
e6 C .
According to Borel-Siebenthal theory, the Lie algebra e6 has three subalgebras as
maximal subalgebras with the maximal rank 6.
(1) The first one is a subalgebra of type T D5 which is obtained as the fixed
points of an involution of e6 .
(2) The second one is a subalgebra of type C1 A5 which is obtained as the fixed
points of an involution of e6 .
(3) The third one is a subalgebra of type A2 A2 A2 which is obtained as the
fixed points of an automorphism w of order 3 of e6 .
The Lie algebra e6 has furthermore two outer involutions , . The subalgebra
obtained as the fixed points of is type F4 and the subalgebra obtained as the fixed
points of is type C4 .
These subalgebras will be realized as subgroups of the group E6 in Theorems
3.10.7, 3.11.4, 3.13.5, 3.7.1 and 3.12.2, respectively.
3.7. Involution and subgroup F4 of E6
We shall study the following subgroup (E6 ) of E6 :
(E6 ) = { E6 | = }
= { E6 | () = } = (E6 ) .
(E6 ) = (E6 )E
= F4 .
it
1 x3 x2
e 1
x3
eit/2 x2
12 (t) x3 2 x1 = x3
eit 2
eit/2 x1 ,
it/2
it/2
x1
3
x2 x1 3
e x2 e
1 = 1
2 + 3
(a, x1 )
2 3
+
cos |a| + i
sin |a|
2 =
2
2
|a|
(2 + 3 )a
|a|
2(a, x1 )a
y = x1 + i
sin2
sin |a|
1
2|a|
|a|2
2
x3 a
|a|
|a|
+i
sin
y2 = x2 cos
2
|a|
2
ax
|a|
|a|
2
y3 = x3 cos
+i
sin
2
|a|
2
sin |a|
means 1 , then 1 (a) (E6 )0 .
if a = 0, then
|a|
1
X = 0
0
0
2
0
0
0 ,
3
i C.
p, q C.
It is sufficient to prove in the case that 2 , 3 are real numbers (otherwise we can
apply some 12 (t1 ) and 13 (t2 ) of Lemma 3.8.1.(1)).
q
t, t > 0 and construct 1 (a(t)) (E6 )0 of
(1) Case q 6= 0. Let a(t) =
|q|
q
(a(t), x1 )
Lemma 3.8.1.(2). Since |a(t)| = t and i
= i |q|, where =
,p ,
|a(t)|
|q|
for Y ((t), y(t)) = 1 (a(t))X0 X, we have
|1 (t)|2 + |2 (t)|2 + |3 (t)|2
2
2 + 3
2
3
+
cos t |q| sin t + i sin t
= |1 |2 +
2
2
2
2 + 3
2
3
+
cos t |q| sin t + i sin t
+
2
2
+
2
2
2
3
2
3
+2
cos t |q| sin t + 2 2 sin2 t
= |1 |2 + 2
2
2
2
+ 2
2
3
2
3
2
= |1 | + 2
+2
+ |q|2 sin2 (t + t0 ) + 2 2 sin2 t
2
2
|1 |2 + |2 |2 + |3 |2 + 2|q|2 + 2 2 cos2 t0 (for some t0 R)
which is the maximal value and attains at some t > 0. This contradicts the maximum
of |1 |2 + |2 |2 + |3 |2 . Hence q = 0.
p
(2) Case p 6= 0. Let a(t) =
t, t > 0 and construct 1 (a(t)) F4 (E6 )0 of
|p|
Lemma 2.8.1. For Y ((t), y(t)) = 1 (a(t))X0 X, the maximal value of |1 (t)|2 +
|2 (t)|2 + |3 (t)|2 is |1 |2 + |2 |2 + |3 |2 + 2|p|2 (the calculation is the same as Proposition 2.8.2) which contradicts the maximum of |1 |2 + |2 |2 + |3 |2 . Hence p = 0.
Consequently, we have x1 = 0. x2 = x3 = 0 can be similarly proved constructing
2 (a), 3 (a) (E6 )0 analogous to 1 (a) of Lemma 3.8.1.(2). Hence X0 is of diagonal
form.
84
The space EIV , called the symmetric space of type EIV , is defined by
EIV = {X JC | detX = 1, hX, Xi = 3}.
Theorem 3.8.3.
In particular, E6 is connected.
E6 /F4 EIV.
1 0 0
X = 0 2 0 , 1 C, 2 0, 3 0
0 0 3
(Proposition 3.8.2). From the condition X EIV , we have
1
3
2
z(E6 ) = {1, 1, 1}, = +
i C.
2
2
Proof. Let z(E6 ). From the commutativity with F4 E6 , we have
E = E = E. Let denote E = Y = Y (, y) JC , then we have
Y = Y,
for all F4 .
85
X JC ,
1 0
where T = 0 1
0 0
have y1 = y2 = y3 = 0
0
1 0 0
0
0 , 0 1 0 and 1
1
0 0 1
0
and 1 = 2 = 3 (= ), that is,
E = Y = E,
0 1
0 0 SO(3). Then we
1 0
C,
= {X JC | 4E1 (E1 X) = X} E1 C ,
(JC ) = {X JC | X = X}
= {X JC | E1 X = 0, hE1 , Xi = 0},
( ) = 1.
Indeed, we have
E2 = (F2 (1) F2 (1)) = F2 (1) F2 (1)
6= 0.
6= 0.
Finally, from
1 = hE1 , E1 i = hE1 , E1 i = hE1 , E1 i = ( )hE1 , E1 i = ( ),
we have ( ) = 1.
In order to investigate the group (E6 ) , we consider the following subgroup (E6 )E1
of E6 :
(E6 )E1 = { E6 | E1 = E1 }.
Lemma 3.10.2. (E6 )E1 is a subgroup of (E6 ) : (E6 )E1 (E6 ) .
Proof. Since (JC ) = {X JC | 4E1 (E1 X) = X} E1 C and (JC ) =
{X JC | E1 X = 0, hE1 , Xi = 0}, these spaces are seen to be invariant under the
action of (E6 )E1 . Hence for (E6 )E1 , we have = (the proof is the same as
that of Theorem 2.9.1). Thus we have (E6 ) .
87
n 0
V 10 = {X JC | 2E1 X = X} = 0
0
Proposition 3.10.3.
by
0
0
o
x C, x C .
(E6 )E1
= Spin(10).
Spin(9)
p
SO(9)
(E6 )E1
p
SO(10)
S9
=
S9
we see that p : (E6 )E1 SO(10) is onto by the five lemma. We also have Ker p =
{1, }. Indeed, Ker p leaves E2 E3 and i(E2 + E3 ) invariant, so that Ei = Ei
for i = 1, 2, 3. Hence we have Spin(8) and so Ker p . Therefore = 1 or
by Theorem 2.7.4. Hence we have the isomorphism
(E6 )E1 /{1, }
= SO(10).
Therefore the group (E6 )E1 is isomorphic to the group Spin(10) as the universal
covering group of SO(10).
For C, 6= 0, we define a
1 x3
() x3 2
x2 x1
C-linear mapping () : JC JC by
4
x2
1
x3
x2
x1 = x3 2 2 2 x1 .
3
x2 2 x1 2 3
(where 12 (t), 13 (t) are mappings defined in Lemma 3.8.1) which is isomorphic to
the usual unitary group U (1) = { C | ( ) = 1}.
Note that U (1) is a subgroup of (E6 ) . From now on, we identify these two groups
U (1).
Lemma 3.10.6. Two subgroups U (1) and Spin(10) of (E6 ) are elementwise
commutative.
Proof. We consider the decomposition JC = E1 C J(2, C)C (JC ) , where
E1 C = {E1 | C},
()|J(2, C)C = 2 1,
()|(JC ) = 1.
J 0 0
J = 0 J 0 M (6, C),
0 0 J
90
J=
0 1
1 0
M, N M (3, H), a H 3 .
The R-linear mappings k : M (3, H) M (6, C) and k : H 3 M (2, 6, C) are extended to C-linear mappings k : M (3, H)C M (6, C) and k : (H 3 )C M (2, 6, C)
respectively by
k(M1 + iM2 ) = k(M1 ) + ik(M2 ),
k(a1 + ia2 ) = k(a1 ) + ik(a2 ),
M1 , M2 M (3, H),
a1 , a2 H 3 .
It is not difficult to see that they satisfy the four properties (1) (4) above.
We define a C-vector space S(6, C) by
S(6, C) = {S M (6, C) | t S = S}
and a C-linear mapping kJ : J(3, H)C S(6, C) by
kJ (M ) = k(M )J.
Then kJ is well-defined. Indeed, for M = M1 + iM2 J(3, H)C , we have
t
Finally, we define Hermitian inner products hS, T i in S(6, C) and hP, Qi in M (2, 6,
C) respectively by
hS, T i = tr(( t S)T ),
M, N J(3, H)C ,
M J(3, H)C .
a, b (H 3 )C .
then
a
b
b
a
+i
91
c
d
d
c
= 0.
0
s12
s13
s14
s15 s16
0
s23
s24
s25 s26
s12
0
s34
s35 s36
s13 s23
det
0
s45 s46
s14 s24 s34
0
1
n3
m3
n2
(m2 )
0
(m3 ) (n3 )
m2
(n2 )
1
(m3 )
0
2
n1
m1
n3
det(kJ (M )) = det
(n3 )
2
0
(m1 ) (n1 )
m3
n2
m2
n1
(m1 )
0
3
(m2 ) (n2 )
m1
(n1 )
3
0
1
m3 + n 3 e 2
(m2 + n2 e2 )
det (m3 + n3 e2 )
2
m1 + n 1 e 2
m2 + n 2 e 2
(m1 + n1 e2 )
3
= 1 2 3 + (m1 + n1 e2 )(m2 + n2 e2 )(m3 + n3 e2 )
3
X
i (mi + ni e2 ) (mi + ni e2 )
+ ((m1 + n1 e2 )(m2 + n2 e2 )(m3 + n3 e2 ))
i=1
Where f4,H = { e6,H | E = 0} is the Lie algebra of the group F4,H and is
isomorphic to the Lie algebra sp(3) by the mapping : sp(3) f4,H given by
(C)M = CM + M C = [C, M ],
M JH
(Proposition 2.11.1) and we see that dim e6,H = 21 + 15 = 35. As in Theorem 3.8.3,
we have a homeomorphism
E6,H /F4,H EIVH = {X (JH )C | detM = 1, hM, M i = 3}
and so we see that the group E6,H is connected.
Proposition 3.11.3.
E6,H
= SU (6)/Z 2 , Z 2 = {E, E}.
M (JH )C .
We first have to prove that (A) E6,H . Indeed, we have (det((A)M ))2 =
det(kJ ((A) M )) (Lemma 3.11.1) = det(A(kJ (M ))t A) = det(kJ (M )) = (detM )2
(Lemma 3.11.1), and so det((A)M ) = detM. On the other hand, since E6,H is
connected (Lemma 3.11.2), the sign of det((A)M ) is constant, that is, independent
of A (assuming that detM 6= 0). Therefore
det ((A)M ) = det M.
93
(E6 )
= (Sp(1) SU (6))/Z 2 , Z 2 = {(1, E), (1, E)}.
Proof. we have
2h t (p, A) (M + a), N + bi
M + a, N + b JC
= 2hM + a, (p, A)(N + b)i
= 2hM + a, kJ 1 (A(kJ (N )) tA) + pbk 1 ( tA)i
= 2hM, kJ 1 (A(kJ (N )) tA)i + 4ha, pbk 1 ( tA)i
= hkJ M, A(kJ (N )) t Ai + 2hk(pa), (kb) t Ai
= h t A(kJ (M )) A, kJ (N )i + 2hk(pa)A, kbi
and so we have t (p, A) = (p, t A), which implies that t (p, A)1 = (p, A).
Claim 2.
(p, A) (E6 ) .
X, Y JC .
k 1 (M ) = k 1 (JM J),
we have
(a) (b) = (pak 1 ( t A)) (pbk 1 ( t A))
= k 1 (A)a bk 1 ( t A),
(a b) = (kJ 1 (A(kJ (a b)) t A))
= k 1 (J (A(k( (a b)))J t A))
= k 1 (J AJk(a b) t A)
= { E6 | () = } = (E6 ) .
For this end, we consider R-vector subspaces (JC ) and (JC ) of JC , which are
eigenspaces of , respectively by
(JC ) = {X JC | X = X}
(
1 m3 m2
0
= m3 2 m1 + i a3 e4
a2 e 4
m2 m1 3
95
a3 e 4
0
a1 e4
a2 e4
a1 e 4
0
)
i R
m i , ai H
(JC ) = {X JC | X
(
1 m3
= i m3 2
m2 m1
= X}
m2
0
m1 + a3 e4
3
a2 e 4
a3 e 4
0
a1 e4
a2 e4
a1 e 4
0
= iJH H 3 ,
)
i R
m i , ai H
The spaces (JC ) and (JC ) are invariant under the action of (E6 ) and we have
the decomposition of JC :
JC = (JC ) (JC ) = (JC ) i(JC ) .
In particular, JC is the complexification of (JC ) : JC = ((JC ) )C .
In the R-vector space
J(4, H) = {P M (4, H) | P = P },
we define the Jordan multiplication P Q and an inner product (P, Q) respectively
by
1
P Q = (P Q + QP ), (P, Q) = tr(P Q).
2
The group Sp(4) acts on J(4, H) by the mapping : Sp(4) J(4, H) J(4, H),
(A, P ) = AP A . Then we have
A(P Q)A = AP A AQA ,
(AP A , AQA ) = (P, Q),
Finally, in the complexification J(4, H)C of J(4, H), we extend naturally the Jordan
multiplication P Q and the inner product (P, Q) and further define a Hermitian
inner product hP, Qi by
hP, Qi = ( P, Q).
The action of Sp(4) on J(4, H) is also naturally extended to J(4, H)C :
A(X1 + iX2 )A = AX1 A + iAX2 A ,
Then we have
hAP A , AQA i = hP, Qi,
96
P, Q J(4, H)C .
We denote by J(4, H)0 the space {P J(4, H) | tr(P ) = 0} and by J(4, H)0
complexification.
its
tr(M )
ia
, M + a (JH )C (H 3 )C = JC .
g(M + a) = 2
1
ia
M tr(M )E
2
M + ia JH iH 3 = (JC ) .
Note that the mapping g in the above definition is the complexification of this
restriction.
Lemma 3.12.1. The mapping g : JC J(4, H)0 C is a C-linear isomorphism
and satisfies
1
gX gY = g((X Y )) + (X, Y )E,
4
(gX, gY ) = (X, Y ),
X, Y JC .
X, Y JC .
1
i
tr(M N ) (a, b)
(aN + bM )
2
2
=2 i
1
1
(aN + bM )
M N (a b + b a) (tr(M N ) (a, b))E
2
2
2
=
1
1
(M, N ) (a, b) E
= g(M + a) g(N + b)
4
2
1
= g(M + a) g(N + b) ((M + a), (N + b))E
4
1
= gX gY (X, Y )E.
4
97
1
Thus the first equality gX gY = g((X Y )) + (X, Y )E is proved. Taking the
4
traces of the both sides, we have
(gX, gY ) = (X, Y ),
X, Y JC .
(i)
Xi , Yi (JC )
(E6 )
= Sp(4)/Z 2 , Z 2 = {E, E}.
X JC .
We first have to prove that (A) (E6 ) . Let Z = (A)X and use Lemma 3.12.1,
then we have
and
Certainly,
1
(g(E))2 = g(E) g(E) = g((E E)) + (E, E)E (Lemma 3.12.1)
4
1
1
= g( E) + h E, EiE = g( E) + h E, EiE
4
4
1
3
1
= g(E) + hE, EiE = g(E) + hE, EiE = g(E) + E.
4
4
4
Now, let
1
(2g(E) + E),
4
then we have P J(4, H) and P 2 = P , tr(P ) = 1, that is, P HP 3 . Indeed,
P =
1
1
(4(g(E))2 + 4g(E) + E) = (2g(E) + E) = P,
16
4
1
1
tr(P ) = tr(2g(E)) + tr(E) = 0 + 1 = 1.
4
4
P2 =
1
tr(M )
ia
p
p
0
1
2
(
= g
1
0
0 D
ia
M tr(M )E
2
1
tr(M )
ipaD
= g 1 2
1
iDa p DM D tr(M )E
2
= DM D + paD = (M + a).
99
0
D
Hence we have
= (A) = (A)(B) = (AB),
AB Sp(4),
so that is onto. Ker = {E, E} = Z 2 can be easily obtained. Thus we have the
isomorphism Sp(4)/Z 2
= (E6 ) .
3.13. Automorphism w of order 3 and subgroup (SU (3)SU (3)SU (3))/Z 3
of E6
Let the C-linear mapping w : JC JC be the complexification of w G2 F4
of Section 2.12. Then w E6 and w3 = 1.
We shall study the following subgroup (E6 )w of E6 :
(E6 )w = { E6 | w = w}.
As in Section 2.12, we identify
J(3, C)C M (3, C)C = JC .
For convenience, we denote J(3, C) and {X J(3, C) | tr(X) = 0} by JC and (JC )0 ,
respectively.
The group E6,C is defined to be obtained by replacing JC with (JC )C in the
definition of the group E6 :
E6,C = { IsoC ((JC )C ) | det (X) = det X, hX, Y i = hX, Y i}.
As in the case E6 , the group E6,C contains a subgroup
F4,C = { E6,C | E = E},
which is also defined by the group F4,C = { IsoR (JC ) | (X Y ) = X Y },
moreover, it is isomorphic to the group (SU (3)/Z 3 ) Z 2 (Proposition 2.12.1). The
Lie algebra e6,C of the group E6,C is
e6,C = { HomC ((JC )C ) | (X, X, X) = 0, hX, Y i + hX, Y i = 0}
= { + iTe | f4,C , T (JC )0 }
(Theorem 3.2.4), where f4,C = { e6,C | E = 0} is the Lie algebra of the group
F4,C . In particular, the dimension of e6,C is
dim(e6,C ) = 8 + 8 = 16.
As in Theorem 3.8.3, we see that the space
EIVC = {X (JC )C | detX = 1, hX, Xi = 3}
100
a+b
ab
+i
e1 = a + b,
2
2
AB
A+B
+i
e1 = A + B,
2
2
1 + ie1
.
2
e6,C
= su(3) su(3).
X (JC )C
gives an isomorphism as Lie algebras. This is the direct consequence of the following
Proposition 3.13.4, so we will omit its proof here.
We define an action of the group Z 2 = {1, } on SU (3) SU (3) by
(A, B) = (B, A),
and let (SU (3) SU (3)) Z 2 be the semi-direct product of the groups SU (3) SU (3)
and Z 2 under this action.
101
1
3
2
2
1 E), (1 E, 1 E)}, 1 = +
e1 .
2
2
Proof. We define a mapping : (SU (3) SU (3)) Z 2 E6,C by
((A, B), 1)X = h(A, B)Xh(A, B) ,
X (JC )C .
First we have to show that = ((A, B), 1) E6,C . Using det(h(A, B)) = h(detA,
detB) (Lemma 3.13.2.(4)) = h(1, 1) = 1 and h(A, B) h(A, B) = h(A , B ) h(A, B)
(Lemma 3.13.2.(3)) = h(A A, B B) = h(E, E) = E, we have
det (X) = (det (h(A, B)))(det X)(det (h(A, B) )) = det X,
hX, Y i = hh(A, B)Xh(A, B) , h(A, B)Y h(A, B) i
= ( h(A, B) X h(A, B) , h(A, B)Y h(A, B) )
= ( X, Y ) = hX, Y i.
Hence E6,C . Since ((E, E), ) = G2,C (= Aut(C)) F4,C E6,C , we also
have ((A, B), ) = ((A, B), 1)((E, E), ) E6,C . Next, we shall show that is a
homomorphism. Indeed, for instance,
((A, B), )((C, D), 1)X
= ((A, B), )(h(C, D)Xh(C, D) )
= h(A, B)h(C, D)Xh(C, D) h(A, B)
= h(A, B)h(D, C)Xh(D, C) h(A, B) (Lemma 3.13.3.(3))
= h(AD, BC)Xh(AD, BC) = ((AD, BC), )X
= ((A, B)(C, D), )X,
X (JC )C .
That Ker = {(E, E), (1 E, 1 E), (1 2 E, 1 2 E)} 1 = Z 3 1 can be easily obtained. In particular, Ker is discrete. Hence induces an injective homomorphism
: su(3) su(3) e6,C .
In particular, dim(su(3) su(3)) = dim(e6,C ), so is an isomorphism ( coincides
with C of Lemma 3.13.3). Hence, induces the surjection : SU (3) SU (3)
(E6,C )0 (which denotes the connected component of E6,C containing the identity 1).
However = ((E, E), ) 6 (E6,C )0 . Certainly, for any A, B SU (3),
h(A, B)Xh(A, B) = X,
X (JC )C
does not hold. Therefore E6,C has just two connected components (Lemma 3.13.1).
Consequently, : (SU (3) SU (3)) Z 2 E6,C is onto and we have the isomorphism
((SU (3) SU (3))/Z 3 ) Z 2
= E6,C .
102
1
1
(X + M ) (Y + N ) = (X Y (M N + N M )) (M Y + N X + M N ),
2
2
hX + M, Y + N i = hX, Y i + 2hM, N i
hX, Y i = hX, Y i
Furthermore, we have
hX, M i = 0 = hX, M i,
hM, N i = ( M, N ) = (P M h(A, B) , P N h(A, B) )
= ( M, N ) = hM, N i,
and so E6 . Clearly, w = w, so that (E6 )w . Evidently is a homomorphism. We shall now show that is onto. Let (E6 )w . The restriction of
to (JC )w = {X JC | wX = X} = (JC )C belongs to E6,C : E6,C . Hence, there
exist A, B SU (3) such that
X = h(A, B)Xh(A, B)
or X = h(A, B)Xh(A, B) ,
X (JC )C
(Proposition 3.13.4). In the former case, let = (E, A, B)1 , then |(JC )C = 1.
Hence G2 , moreover (G2 )e1 = (G2 )w = SU (3) (Theorem 1.9.4). Thus there
exists P SU (3) such that
(X + M ) = X + P M = (P, E, E)(X + M ),
103
Therefore we have
= (E, A, B) = (E, A, B)(P, E, E) = (P, A, B).
In the latter case, consider the mapping 1 : JC JC , 1 (X + M ) = X + M , X +
M (JC )C M (3, C)C = JC and remember that 1 G2 F4 E6 . Let
= 1 (E, A, B)1 , then E6 and |(JC )C = 1. Hence (G2 )e1 = (G2 )w
(Theorem 1.9.4) (E6 )w . Since and (E, A, B) (E6 )w , we have 1 (E6 )w , so
that 1 (G2 )w which is a contradiction (Theorem 1.9.4). Therefor we see that
is onto. That Ker = {(E, E, E), (1 E, 1 E, 1 E), (1 2 E, 1 2 E, 1 2 E)} = Z 3 can
be easily obtained. Thus we have the isomorphism (SU (3) SU (3) SU (3))/Z 3
=
w
(E6 ) .
Remark 1. Since (E6 )w is connected as the fixed points subgroup of E6 by
an automorphism w of order 3 of the simply connected group E6 , the fact that :
SU (3) SU (3) SU (3) (E6 )w is onto can be proved as follows. The elements
G01 , G23 ,
G45 , G67 , G26 + G37 , G27 + G36 ,
G24 + G35 , G25 + G34 , G46 + G57 , G47 + G56 ,
e1 (1), A
e2 (1), A
e3 (1), A
e1 (e1 ), A
e2 (e1 ), A
e3 (e1 ),
A
(E1 E2 ) ,
(E2 E3 ) ,
Fe1 (1),
Fe1 (e1 ),
d = 78.
Since E6 is simply connected (Theorem 3.9.2), E6 C is also simply connected. The Lie
algebra of the group E6 C is e6 C , so E6 C is a complex simple Lie group of type E6 .
3.15. Non-compact exceptional Lie groups E6(6) , E6(2) , E6(14) and E6(26)
of type E6
by
hX, Y i = hX, Y i.
E6(6)
= (E6 C ) .
= (E6 C ) , E6(26)
= (E6 C ) , E6(14)
= (E6 C ) , E6(2)
Theorem 3.15.1. The polar decompositions of the Lie groups E6(6) , E6(2) , E6(14)
and E6(26) are respectively given by
E6(6) Sp(4)/Z 2 R42 ,
E6(26) F4 R26 .
Proof. These are the facts corresponding to Theorems 3.12.2, 3.11.4, 3.10.7 and
3.7.1.
Theorem 3.15.2. The centers of the groups E6(6) , E6(2) , E6(14) and E6(26) are
given by
z(E6(6) ) = {1}, z(E6(2) ) = Z 3 , z(E6(14) ) = Z 3 , z(E6(26) ) = {1}.
105
1
2B
0 A
X
X
3
Y
Y
1
t + B 0
(, A, B, ) = 2A
3
0
A
B
0
0
1
X X + 2B Y + A
3
2A X t Y + 1 Y + B
=
.
3
(A, Y ) +
(B, X)
= (X W + Z Y )
A = (2Y W Z X)
4
P Q = (, A, B, ),
1
B =
(2X Z W Y )
(1) P Q = Q P .
3
(2) (P Q)P (P P )Q + {P, Q}P = 0.
8
1
1
1
(3) (P R)Q (Q R)P + {Q, R}P {P, R}Q {P, Q}R = 0.
8
8
4
Proof. (1) is evident.
(2) For P = (X, Y, , ), Q = (Z, W, , ) PC , we have
(P Q)P
1
1
1
= (X W + Z Y ), (2Y W Z X), (2X Z W Y ),
2
4
4
1
((X, W ) + (Z, Y ) 3( + )) (X, Y, , )
8
= (using the formula X Y of Lemma 3.4.1), etc.)
1
1
1
= X Y, (Y Y X), (X X Y ), ((X, Y ) 3) (Z, W, , ),
2
2
4
3
((X, W ) (Z, Y ) + )(X, Y, , )
8
3
= (P P )Q {P, Q}P.
8
(3) In the equality (2), put P + R in the place of P , then we have
3
3
2(P R)Q (P Q)R (R Q)P + {Q, R}P {P, Q}R = 0.
8
8
Exchanging P with Q, we see that
3
3
2(Q R)P (Q P )R (R P )Q + {P, R}Q {Q, P }R = 0.
8
8
Taking ((i)(ii))3, we have
1
1
1
(P R)Q (Q R)P + {Q, R}P {P, R}Q {P, Q}R = 0.
8
8
4
We define a space MC , called the complex Freudenthal manifold, by
MC = {P PC | P P = 0, P 6= 0}
(
)
P = (X, Y, , ) PC X Y = 0, (X, Y ) = 3,
=
.
X X = Y, Y Y = X
P 6= 0
X
(Y Y )
0
0
1 (X X)
, 1 =
,
, 1. =
1
2 det X
1
0
1
det Y
2
107
(i)
(ii)
belong to MC .
Proof. This is clear from X (X X) = 0 (Lemma 3.5.4.(1)) and (X X)
(X X) = (det X)X (Lemma 2.1.1.(3)).
4.2. Compact exceptional Lie group E7
Definition. We define the groups E7 C and E7 respectively by
E7 C = { IsoC (PC ) | (P Q)1 = P Q},
E7 is a subgroup of E7 C .
Theorem 4.2.1. E7 is a compact Lie group.
Proof. E7 is a compact Lie group as a closed subgroup of the unitary group
U (56) = U (PC ) = { IsoC (PC ) | hP, Qi = hP, Qi}.
Proposition 4.2.2. For E7 C (and so, for E7 ), we have
(1) MC = MC .
P, Q PC .
108
= [1 , 2 ] + 2A1 B2 2A2 B1
2
2
A = 1 + 1 A2 2 + 2 A1
3
3 2
2
B = 1 + 1 B2 + t 2 + 2 B1
3
3
= (A1 , B2 ) (B1 , A2 ).
{P, Q} + {P, Q} = 0, P, Q P .
(i)
(ii)
g
k
=
c
b
l
h
a
d
C
B
A
D
,
g, h, k, l HomC (JC ),
a, b, c, d HomC (JC , C),
A, B, C, D JC ,
, , , C.
P, Q PC .
e7 C fr fr 1
g
rk
= 2
r c
r1 b
r1 l r2 C
h
r1 B
ra
r2 d r3 k
rA
r2 D
3
r
i e7 C ,
= 3 + 2 + 1 + 0 + 1 + 2 + 3 ,
where
3
0
0
=
0
0
0
0
=
0
0
0
0
=
0
b
0
0
0
0
, 3 =
0
0
0
0
0
0
0
0
, 2 =
0
c
0
0
0
0
0
k
, 1 =
0
0
0
0
g 0 0 0
0 h 0 0
0 =
.
0 0 0
0 0 0
0 0
0 0
0 0
0
0 C
0 0
0 0
d 0
l 0
0 B
0 0
0 0
0
0
0
0
0
0
0
0
0
0
a
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
,
0
D
,
0
0
A
0
,
0
0
Y JC .
X JC .
(Since b is continuous, the above is also valid for X JC such that det X = 0).
Similarly, using 1 , we have
110
k(X) = 2A X,
a(Y ) = (A, Y ),
X, Y JC .
(i)
(ii)
(iii)
1
Putting = g ( + 2)1, then, using (i) and (iii), we have
3
3(X, X, X) = 3(g(X), X, X) ( + 2)(X, X, X)
= (X X + h(X X), X) + (g(X), X X) 3( + 2)det X
= 3det X + 3( + )det X 3( + 2)det X = 0.
Therefore
e6 C .
1
Similarly = h (2 + )1 e6 C . Furthermore, from (ii), we have
3
1
2 (X X) + (2 + )(X X) (X X)
3
1
= (detX) X + X + ( + 2)X ,
3
1
2B
0 A
2A
B
0
=
0
A
0
B
0
0
C
= (, A, B, ), e6 , A, B JC , C.
Conversely, we shall show that = (, A, B, ), e6 C , A, B JC , C belongs
to e7 C , that is, exp t E C for all t C. For this purpose, we prove the following
proposition.
111
[, P P ]
h
i
1
1
1
= (, A, B, ), X Y, (Y Y X), (X X Y ), ((X, Y ) 3)
2
2
4
= (using (X Y ) = X Y + X Y (Lemma 3.4.3.(1)), [, A B]
= A B + A B (Lemma 3.4.4.(1)), the formula about A B (Lemma
3.4.1) etc.)
= 2P P.
We will now return to the proof of Theorem 4.3.1. For = (, A, B, ), e6 C ,
A, B JC , C and t C, we have
(exp t)(P Q)(exp t)1
= (exp t(ad))(P Q) ((ad)1 = [, 1 ], 1 e7 C )
X
1
(t(ad))n (P Q)
=
n!
n=0
n X
X
t
n! k
=
P l Q (Proposition 4.3.2)
n!
k! l!
n=0
k+l=n
X
X
tk tl k
=
P l Q
k! l!
n=0
k+l=n
X
X
1
1
=
(t)k P
(t)l Q
k!
l!
k=0
l=0
hP, Qi = { P, Q}.
t 1
= 1 .
if and only if
= .
2A
(, A, A, ) =
0
A
2 A
1
+
3
A
0
0
A
0
.
[(1 , A1 , A1 , 1 ), (2 , A2 , A2 , 2 )] = (, A, A, ),
where
= [1 , 2 ] 2A1 A2 + 2A2 A1
2
2
A = 1 + 1 A2 2 + 2 A1
3
3
= hA1 , A2 i hA2 , A1 i.
dim(e7 ) = 78 + 54 + 1 = 133.
Proof. For e7 C ,
e7
if and only if 1 = .
113
Therefore the theorem follows from Theorem 4.3.1, Lemma 4.3.3.(5) and (, A,
B, ) = ( , A, B, ).
Proposition 4.3.5. The complexification of the Lie algebra e7 is e7 C .
Proof. For e7 C , the conjugate transposed mapping of with respect to
the inner product hP, Qi of PC is = e7 C , and for e7 C , belongs to
e7 if and only if = . Now, any element e7 C is represented as
=
+
+i
,
2
2i
+
,
e7 .
2
2i
(a)
(b)
(c)
Taking ((a)(b))2, ((a)(c))8, we have (X, 0, , 2) W , (0, 0, , ) W , respectively. Consequently (X, 0, 0, 3) W . Next, if we choose X1 JC such that
(X1 , X) 6= 0, from
W (0, 0, X1 , 0)(X, 0, 0, 3) = (0, 0, 0, (X1 , X)),
115
PC .
nX
(Pi
Proof. Since [, P Q] = P Q + P Q (Proposition 4.3.2), a =
i
o
Qi ) | Pi , Qi PC is an ideal of e7 C . From the simplicity of e7 C (Theorem 4.4.1), we
have a = e7 C .
, i e7 C .
[, 1 ] + 2A B1 2A1 B
2
2
2
!
+ A1 1 + 1 A
A
2
3
3
=
,
2
2
t + B1 + t 1 + 1 B B2 7
3
3
2
(A, B1 ) (B, A1 )
116
1
(X W + Z Y )
2
(2Y
X)
4
(, P Q)7 = ,
B
(2X Z W Y )
7
1
((X, W ) + (Z, Y ) 3( + ))
8
= (, X W + Z Y )6 (A, 2X Z W Y ) + (2Y W Z X, B)
1
((X, W ) + (Z, Y ) 3( + ))
3
= (X, W ) + (Z, Y ) 2(A, X, Z) + (A, W ) + (A, Y ) + 2(Y, W, B)
1
1
(Z, B) (X, B) (X, W ) (Z, Y ) + ( + )
3
3
1
X X + 2B Y + A
Z
3
(
)
1
W
2A X t Y + Y + B ,
=
= {P, Q}.
(A, Y ) +
(B, X)
Hence
h
2
i
4 4
2
[0 , [0 , (, A, B, )] ] = 0 , 0, A, B, 0 = 0, A, B, 0 .
3
3
9 9
B7 (0 , 0 ) = tr((ad0 )2 ) =
117
4
27 2 = 24.
9
we have
X Y
X Y
0 0 (X, Y, , ) = 0 , , , =
, , , ,
3 3
9 9
tr(02 ) =
1
27 2 + 1 + 1 = 8.
9
Therefore k = 3.
Lemma 4.5.3. For P PC , P 6= 0, there exists Q PC such that P Q 6= 0.
Proof. Asumme that P Q = 0 for all Q PC . Then for any e7 C ,
0 = (, P Q)7 = (, Q P )7 = {Q, P } (Lemma 4.5.1.(2)). Since e7 C PC = PC
(Proposition 4.4.2.(2)), we have {PC , P } = 0, so that P = 0.
4.6. Roots of e7 C
Theorem 4.6.1 The rank of the Lie algebra e7 C is 7. The roots of e7 C relative to
some Cartan subalgebra h are given by
(k l ), (k + l ), 0 k < l 3,
1
k (2 3 ),
0 k 3,
2
1
1
(0 1 + 2 3 ) (3 1 ),
2
2
1
1
( 0 + 1 + 2 3 ) (3 1 ),
2
2
1
1
(0 + 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 3 ) (1 2 ),
2
2
1
1
(0 + 1 + 2 3 ) (1 2 ),
2
2
1
1
( 0 + 1 + 2 + 3 ) (1 2 ),
2
2
1
1
(0 1 + 2 + 3 ) (1 2 ),
2
2
2
j + , 0 j 3,
1 3 2
k
1 , 0 k 3,
2
3
1
1
2
(0 1 + 2 3 )
2 ,
2
3
21
1
2
( 0 + 1 + 2 3 )
2 ,
2
2
3
118
1
1
2
(0 + 1 + 2 + 3 )
2 ,
2
3
21
1
2
( 0 1 + 2 + 3 )
2 ,
2
2
3
1
2
1
3 ,
( 0 1 + 2 3 )
2
3
12
1
2
(0 + 1 + 2 3 )
3 ,
2
2
3
1
1
2
( 0 + 1 + 2 + 3 )
3 ,
2
2
3
1
1
2
(0 1 + 2 + 3 )
3
2
2
3
with 1 + 2 + 3 = 0.
Proof. We use the decomposition of of e7 C in Theorem 4.3.1
e7 C = e6 C JC JC C.
Let
3
3
X
n X
j Ej , 0, 0, e7 C
k Hk +
h=
k=0
j=1
o
k , C
j C, 1 + 2 + 3 = 0
j=1
Hence is a root of e7 C .
e 0, 0, ), (0, Ej , 0, 0)]
II [(h + H,
e + 2 Ej , 0, 0 = j + 2 (0, Ej , 0, 0),
= 0, h + H
3
3
e 0, 0, ), (0, 0, Ej , 0)] = 0, 0, (h + H)
e 2 Ej , 0
[((h + H,
3
2
2
e
= 0, 0, h H Ej , 0 = j (0, 0, Ej , 0).
3
3
2
C
Hence j + , 0 j 3, are roots of e7 .
3
e 0, 0, ), (0, F1 (a), 0, 0)]
III [(h + H,
a = ek ie4+k
2
e + F1 (a), 0, 0
= 0, h + H
3
2
1
= 0, F1 (h a) + (2 + 3 )F1 (a) + F1 (a), 0, 0
2
3
119
1
2
k 1 + (0, F1 (a), 0, 0).
2
3
2
1
Hence k 1 + , 0 k 3, are roots of e7 C . The remainders of roots can be
2
3
similarly found (h , h in Theorem 3.6.4 will be used).
=
7
1
4
4
2
=1
=1
=1
=0
=0
=0
0
1
1
1
1
0
0
0
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 + 1
0 + 2
0 + 3
1 + 2
1 + 3
2 + 3
1
0 + (2 3 ) = 1
2
1
1 + (2 3 ) = 0
2
1
2 + (2 3 ) = 0
2
1
3 + (2 3 ) = 0
2
120
=1
=1
=1
=0
=0
=0
2
1
1
1
1
0
2
2
1
2
1
1
1 1
1 1
1 1
1 1
1 1
1 1
0 1
1 1
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
0 (2 3 ) = 1 1 1 1
2
1
1 (2 3 ) = 0 1 1 1
2
1
2 (2 3 ) = 0 0 1 1
2
1
3 (2 3 ) = 0 0 0 1
2
1
1
(0 1 + 2 3 ) + (3 1 ) = 0
2
2
1
1
( 0 + 1 + 2 3 ) + (3 1 ) = 1
2
2
1
1
(0 + 1 + 2 + 3 ) + (3 1 ) = 0
2
2
1
1
( 0 1 + 2 + 3 ) + (3 1 ) = 1
2
2
1
1
( 0 + 1 2 + 3 ) + (3 1 ) = 1
2
2
1
1
(0 1 2 + 3 ) + (3 1 ) = 0
2
2
1
1
( 0 1 2 3 ) + (3 1 ) = 1
2
2
1
1
(0 + 1 2 3 ) + (3 1 ) = 0
2
2
1
1
( 0 1 + 2 3 ) (1 2 ) = 1
2
2
1
1
( 0 1 2 + 3 ) (1 2 ) = 1
2
2
1
1
( 0 + 1 + 2 + 3 ) + (1 2 ) = 0
2
2
1
1
( 0 + 1 2 3 ) (1 2 ) = 1
2
2
1
1
(0 + 1 2 + 3 ) (1 2 ) = 0
2
2
1
1
(0 + 1 + 2 3 ) (1 2 ) = 0
2
2
1
1
( 0 + 1 + 2 + 3 ) (1 2 ) = 1
2
2
1
1
(0 1 + 2 + 3 ) (1 2 ) = 0
2
2
2
1 = 1 2 3 4 2
3
2
2 + = 0 0 0 0 0
3
2
3 = 0 0 0 0 0
3
121
1 0
1 0
1 0
1 0
0 1
1 0
0 0
2 3
3 2
1 1
1 2
3 2
1 1
1 2
3 2
1 1
2 2
3 2
1 1
0 0
1 0
0 0
1 1
1 0
0 0
1 1
1 0
0 0
1 2
2 1
1 1
1 1
2 1
1 1
0 0
0 1
0 0
2 2
2 1
1 1
1 1
2 1
1 1
1 2
2 1
1 1
2 3
4 3
2 2
0 1
2 1
1 1
1 2
1 0
0 1
1
0 1 +
2
1
1 1 +
2
1
2 1 +
2
1
3 1 +
2
1
0 + 1
2
1
1 + 1
2
1
2 + 1
2
1
3 + 1
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
=1 1
1 1
1 1
=0 1
1 1
1 1
=0 0
1 1
1 1
=0 0
0 1
1 1
=1 1
1 1
1 0
=0 1
1 1
1 0
=0 0
1 1
1 0
=0 0
0 1
1 0
0 1
1 0
0 1
2 3
3 2
1 2
1 2
3 2
1 2
1 2
3 2
1 2
2 2
3 2
1 2
0 0
1 0
0 1
1 1
1 0
0 1
1 1
1 0
0 1
1 2
2 1
0 1
1 1
2 1
0 1
0 0
0 1
1 0
2 2
2 1
0 1
1
2
1
(0 1 + 2 3 ) + 2 = 0
2
2
3
1
2
1
( 0 + 1 + 2 3 ) + 2 = 1
2
2
3
1
2
1
(0 + 1 + 2 + 3 ) + 2 = 0
2
2
3
1
2
1
( 0 1 + 2 + 3 ) + 2 = 1
2
2
3
1
1
2
( 0 + 1 2 + 3 ) + 2 = 1
2
2
3
1
2
1
(0 1 2 + 3 ) + 2 = 0
2
2
3
1
2
1
( 0 1 2 3 ) + 2 = 1
2
2
3
1
1
2
(0 + 1 2 3 ) + 2 = 0
2
2
3
1
2
1
( 0 1 + 2 3 ) + 3 = 1
2
2
3
1
2
1
( 0 1 2 + 3 ) + 3 + = 1
2
2
3
1
2
1
( 0 + 1 + 2 + 3 ) 3 = 0
2
2
3
1
2
1
( 0 + 1 2 3 ) + 3 = 1
2
2
3
1
1
2
(0 + 1 2 + 3 ) + 3 = 0
2
2
3
1
2
1
(0 + 1 + 2 3 ) + 3 = 0
2
2
3
1
2
1
( 0 + 1 + 2 + 3 ) + 3 = 1
2
2
3
1
2
1
(0 1 + 2 + 3 ) + 3 = 0
2
2
3
122
1 2
1 0
2 2
1 0
3 4
3 1
1 2
1 0
1.
3
3
n X
X
=
j Ej , 0, 0, e7 C
k Hk +
j=1
k=0
o
k , R
j R, 1 + 2 + 3 = 0
3
3
X
X
j j + 4
B7 (h, h ) = 6 6
k k + 3
j=1
k=0
3
3
3
3
X
X
X
X
for h =
j Ej , 0, 0, , h =
k Hk +
j Ej , 0,
k Hk +
0,
j=1
k=0
j=1
k=0
X
X
X
3 X
+ 24
j Ej
k Hk +
j Ej ,
B6
k Hk +
2
j=1
j=1
3
B7 (h, h ) =
k=0
k=0
3
3
X
3 X
= 12 2
j j + 24 (Theorem 3.6.5)
k k +
2
j=1
k=0
3
3
X
X
j j + 4 .
= 6 6
k k + 3
k=0
j=1
1
(H0 H1 , 0, 0, 0),
36
1
(H1 H2 , 0, 0, 0),
36
1
(H2 H3 , 0, 0, 0),
36
1
((H0 H1 H2 + H3 ) + 2(E3 E1 ) , 0, 0, 0),
72
1
((H0 + H1 + H2 + H3 ) + 2(E1 E2 ) , 0, 0, 0),
72
1
3
(E1 + 2E2 E3 ) , 0, 0, ,
54
2
1
3
(E1 + E2 2E3 ) , 0, 0, .
54
2
1 1
1
2=
36 36
18
and the other inner products are similarly calculated. Consequently, the inner product
induced by the Killing form B7 between 1 , 2 , , 7 and are given by
(1 , 1 ) = B7 (H1 , H1 ) = 36
123
(i , i ) =
1
,
18
i = 1, 2, 3, 4, 5, 6, 7,
(1 , 2 ) = (2 , 3 ) = (3 , 4 ) = (4 , 5 ) = (4 , 7 ) = (5 , 6 ) =
(i , j ) = 0,
(, ) =
1
,
36
otherwise,
1
,
18
(, 6 ) =
1
,
36
(, i ) = 0, i = 1, 2, 3, 5, 7,
using them, we can draw the Dynkin diagram and the extended Dynkin diagram of
e7 C .
According to Borel-Siebenthal theory, the Lie algebra e7 has four subalgebras as
maximal subalgebras with the maximal rank 7.
(1) The first one is a subalgebra of type T E6 which is obtained as the fixed
points of an involution of e7 .
(2) The second one is a subalgebra of type A1 D6 which is obtained as the fixed
points of an involution of e7 .
(3) The third one is a subalgebra of type A7 which is obtained as the fixed points
of an involution of e7 .
(4) The fourth one is a subalgebra of type A2 A5 which is obtained as the fixed
points of an automorphism w of order 3 of e7 .
These subalgebras will be realized as subgroups of the group E7 in Theorems
4.10.2, 4.11.15, 4.12.5 and 4.13.5, respectively.
4.7. Subgroups E6 and U (1) of E7
We shall study the following subgroup (E7 )(0,0,1,0) of E7 :
(E7 )(0,0,1,0) = { E7 | (0, 0, 1, 0) = (0, 0, 1, 0)}.
Lemma 4.7.1. If E7 satisfies (0, 0, 1, 0) = (0, 0, 1, 0), then also satisfies
(0, 0, 0, 1) = (0, 0, 0, 1), and conversely.
Proof. If E7 satisfies (0, 0, 1, 0) = (0, 0, 1, 0), then we have
(0, 0, 0, 1) = ( (0, 0, 1, 0)) = (0, 0, 1, 0) = (0, 0, 1, 0) = (0, 0, 0, 1).
The converse can be similarly proved.
Theorem 4.7.2.
(E7 )(0,0,1,0)
= E6 .
124
0
0
e=
0
0
0
0
0
0
1
0
0
0
(E7 )(0,0,1,0) E7 .
0
1
eP
eQ = (X, Y, , ) (Z, W, , )
= (using (X Y )1 = X Y, (1 ) = ( ) ( 1 ) etc.)
=
e(P Q)e
1
and he
P,
eQi = hP, Qi is evident. Hence
e E7 , moreover
e (E7 )(0,0,1,0) .
=
0
0
1
0
0
0
0
1
0
0
0
,
0
1
, 1 , , HomC (JC ).
Indeed, the fact that the left bottom parts are 0 follows from
1i
1i
1i
= hX,
= hX,
= 0,
hX,
1.i = hX,
1.i = hX,
1.i = 0.
hX,
Now, since
1
X + (X X)
X X
1
X
+
(X
X)
1
C
,
=
M
1 detX
1
2
det
X
and so
1
1
(X, X X) = (X, 1 (X X)) = det X,
3
3
C
Y i = hX,
Y i implies hX, Y i =
which implies that E6 . The equality hX,
hX, Y i and therefore E6 . Moreover from the relation
det (X) =
1 (X X) = X X = (X X),
we obtain 1 = . Indeed, putting X X instead of X, we have
(det X)1 X = (det X) X.
If det X 6= 0, then we have 1 X = X, Since 1 and are linear mappings (of
course are continuous mappings), we have 1 X = X even if det X = 0. Thus, the
proof of Theorem 4.7.2 is completed.
For C, 6= 0, we define a C-linear transformation 1 () : PC PC by
1 ()(X, Y, , ) = (1 X, Y, 3 , 3 ).
Then 1 () E7 C .
Theorem 4.7.3. The group E7 contains
U (1) = {1 () | C, ( ) = 1}
as a subargroup. This subgroup is isomorphic to the usual unitary group U (1) = {
C | ( ) = 1}.
Proof. It is easy to check that 1 () E7 .
4.8. Connectedness of E7
We denote by (E7 )0 the connected component of E7 containing the identity 1.
by
2 a sin |a| Ei
|a|
i (a) =
sin |a|
Ei
a
|a|
2a
sin |a|
Ei
|a|
sin |a|
Ei
|a|
0
0
a
sin |a|
Ei
|a|
cos |a|
0
sin |a|
Ei
|a|
cos |a|
sin |a|
if a = 0, then
means 1 , then i (a) (E7 )0 , where pi : JC JC is the
|a|
C-linear mapping defined by
126
1
pi x3
x2
x3
2
x1
1
x2
x1 = i3 x3
3
i2 x2
i3 x3
2
i1 x1
i2 x2
i1 x1 ,
3
where ij is the Kronecker delta symble. The mappings 1 (a1 ), 2 (a2 ), (a3 ), (ai C)
are commutative for each other.
Proof. For
0
2 aEi
i (a) = (0, aEi , aEi , 0) =
0
aEi
2aEi
0
aEi
0
0
aEi
0
0
aEi
0
e7
0
0
(Theorem 4.3.4), we have i (a) = exp i (a). Hence i (a) (E7 )0 . The relation
[i (ai ), j (aj )] = 0 shows that i (ai ) and j (aj ) are commutative.
Proposition 4.8.2. Any element P MC , P 6= 0 can be transformed to a
diagonal form by some element (E7 )0 :
P = (X, Y, , ),
1
1
(Y Y ) = Y Y
is also diagonal.
(2) Case P = (X, Y, 0, ), Y 6= 0. Choose
1 0
Y = 0 2
0 0
E6 so that
0
0 , i C
3
X
1 pi 2Ei
0
Ei
1 pi Ei 0 Y
2Ei
i (/2)P =
= , i 6= 0,
0
Ei
0
0
0
i
Ei
0
0
0
i 6= 0,
6= 0,
Theorem 4.8.3.
0
0
1 0 0
1 2 3
1
0
P =
3 1
0 , 0 2 0 , , 2 1 2 3 , > 0
0
0
1 2
0 0 3
1
1
(|2 3 |2 + |3 1 |2 + |1 2 |2 ) + (|1 |2 + |2 |2 + |3 |2 ) + 2 + 4 |1 2 3 |2 = 1,
2
that is,
1+
|1 |2
|2 |2
|3 |2
1
1
+
1
+
= 2.
2
2
2
Choose r1 , r2 , r3 R, 0 ri <
, such that
2
tan ri =
|i |
,
i = 1, 2, 3,
(i)
By letting
ai =
i
ri ,
|i |
i = 1, 2, 3
i =
ai
1 i |i |
ri
=
tan ri cos r1 cos r2 cos r3 .
|ai | |i |
|ai |
|a2 |
|a3 |
sin
|a
|
sin
|a
|
1
3
0
a1
cos |a2 |a3
0
|a
|
|a
|
1
3
sin
|a
|
sin
|a
|
1
2
0
0
a1
a2
cos |a3 |
|a1 |
|a2 |
sin |a1 |
0
0
a1 |a | cos |a2 | cos |a3 |
1
sin |a2 |
0
cos |a1 |a2
cos |a3 |
0
|a2 |
sin
|a
|
3
0
0
cos |a1 | cos |a2 |a3
|a3 |
for all E6 .
(0, 0, 0, 1) = (0, 0, 0, 1 ).
1
0
0 0
1
0 1 0 0
=
, = 1, or 2 .
0
0
1 0
0
0
0 1
(0, X, 0, 0) = ( X, 0, 0, 0) = (X, 0, 0, 0)
= (X, 0, 0, 0) = (0, X, 0, 0) = (0,
X, 0, 0),
E7 .
= { E7 | = } = (E7 ) .
2
2
Theorem
4.10.2 (E7 ) = (U (1) E6 )/Z 3 , Z 3 = {(1, 1), (, 1), ( , 1)},
1
3
= +
i C.
2
2
Proof. We define a mapping : U (1) E6 (E7 ) by
(, ) = 1 (),
1 1 0
1
0
1 () =
0
0
0
0
0
0
3
0
0
0
, =
0
0
3
0
0
0
0
0
0
1
0
0
0
.
0
1
, HomC (JC ),
0 M 0
0 0 N
a, b HomC (JC , C),
=
a 0 0 ,
M, N JC ,
0 b 0
, C.
The condition (0, 0, 1, 0), (0, 0, 0, 1) MC implies that
M = 0,
N = 0.
(i)
we have
1
1X X
C
M
=
1
2 detX
1
(detX)M
2
1
(X X) + N
,
a(X)
1
b(X X)
X +
1
1
(X X) + N
(X X) + N = a(X) X + 2 (detX)M ,
1
1
1
X + 2 (detX)M, (X X) + N = 3a(X) b(X X)
2(X X) N = a(X)X
(X X) (X X) = a(X)(det X)M
(ii)
(iii)
(iv)
(v)
C, ( ) = 1.
U (1), E6 .
132
This shows is onto. That Ker = {(1, 1), (, 1), ( 2 , 2 1)} = Z 3 is easily obtained. Thus we have the isomorphism (U (1) E6 )/Z 3
= (E7 ) .
Remark. (E7 ) is connected as a fixed points subgroup under the involution of
the simply connected Lie group E7 . Hence, to show that : U (1) E6 (E6 ) is
onto, it is sufficient to show that : u(1) e6 (e7 ) is onto, which is easily shown.
4.11. Involution and subgroup (SU (2) Spin(12))/Z 2 of E7
We define an involutive C-linear transformation of PC by
(X, Y, , ) = (X, Y, , ).
which is the extension of the C-linear transformation of JC . This is the same as
F4 regarding as F4 E6 E7 .
We shall now study the following subgroup (E7 ) of E7 :
(E7 ) = { E7 | = }.
To this end, we define two C-linear mappings , : PC PC by
= (2E1 E1 , 0, 0, 1),
= (0, E1 , E1 , 0).
X
1 X
Y Y
= 1 , 1 X = (E1 , X) 4E1 (E1 X),
X
2E1 Y + E1
Y 2E X + E1
= 1
.
(E1 , Y )
(E1 , X)
1 y3 y 2
1 x3 x2
(X, Y, , ) = x3 2 x1 , y 3 2 y1 , ,
x2 x1 3
y2 y 1 3
0
1
0
0
1 0
= 0
2 x1 , 0 2 y1 , , ,
0
x1 3
0 y 1 3
0
0
0
0
x1 , 1 , 1 .
(X, Y, , ) = 0 3 y1 , 0 3
2
0 x1
0 y1 2
=
=
Lemma 4.11.1. (1)
= ,
= ,
= .
133
(2) (e7 ), = { e7 | = , = }
(
(e6 ) , A (JC ) , (E1 , A) = 0, )
= (.A, A, ) e7
.
3
= (E1 , E1 )
2
(3) ((e7 ), )(0,E1 ,0,1) = { (e7 ), | ((0, E1 , 0, 1)) = 0}
e , E = 0,
6
1
= (, A, A, 0) e7
.
A JC , 2E1 A = A
Y JC .
X JC .
1
() x3
x2
x3
2
x1
x2
41
x1 = x3
3
3
x2
x3
22
2x1
x2
2x1
23
b
b( a) a( b)
2(b a)
,
=
,
a
b
2 (b a) a( b) b( a)
and
h
i
((), aE1 , aE1 , ), ((), bE1 , bE1 , )
( )
( )
A
aE1
A aE1
,
(e7 )
=
+
a1 (e7 ) ,
A
aE1
A + aE1
where =
1
1
+ (E1 , E1 ), a = (E1 , A), gives an isomorphism as Lie algebras.
3
2
135
= { E6 | = , E1 = E1 } E7
(Lemma 3.10.4). The group Spin(10) acts transitively on the 9 dimensional sphere
0
n 0 0
o
0
S9 =
x , 0, 0, 0 C, x C, xx + ( ) = 1 .
0 x
Lemma 4.11.6. For ((E7 ), )(0,E1 ,0,1) , we have
(0, E1 , 0, 1) = (0, E1 , 0, 1)
if and only if
In particular, we have
{ ((E7 ), )(0,E1 ,0,1) | (0, E1 , 0, 1) = (0, E1 , 0, 1)}
= Spin(10).
Proof. If (E7 ), satisfies (0, E1 , 0, 1) = (0, E1 , 0, 1) and (0, E1 , 0, 1) =
(0, E1 , 0, 1), then we have (0, 0, 0, 1) = (0, 0, 0, 1) and (0, E1 , 0, 0) = (0, E1 , 0, 0),
which imply that (0, 0, 1, 0) = (0, E1 , 0, 0) = (0, E1 , 0, 0) = (0, E1 , 0, 0) =
(0, 0, 1, 0). The converse can be similarly proved. If E7 satisfies (0, 0, 1, 0) =
(0, 0, 1, 0), then E6 (Theorem 4.7.2), and from the condition E1 = E1 , we obtain
Spin(10), (Theorem 3.10.4). The converse also holds.
We define an 11 dimensional R-vector space V 11 by
V 11 = {P PC | P = P, P
n 0 0
0
=
x ,0
0 x
0
= P, P (0, E1 , 0, 1, 0) = 0}
0 0
o
0 0 , 0, x C, C, iR
0 0
136
1
{P, P } = xx + ( ) + ( ).
2
((E7 ), )(0,E1 ,0,1) /Spin(10) S 10 .
0
0 0
0 0
P = 0
x , 0 0 0 , 0, S 10 ,
0 x
0 0 0
choose a R, 0 a <
, such that
4
tan 2a =
2
.
If = 0, then we choose a =
. Applying 23 (a) of Lemma 4.11.5 on P , then
4
the -part of 23 (a)P becomes
2 sin2 a(E2 , E3 X) + sin a cos a (E3 , Y ) sin a cos a cos2 a
= sin2 a + ( ) sin a cos a cos2 a
1
= ( ) sin 2a cos 2a = 0.
2
Hence we have
23 (a)P S 9 .
Since the group Spin(10) acts transitively on S 9 (Proposition 3.10.3), there exists
Spin(10) = (E6 )E1 ((E7 ), )(0,E1 ,0,1) such that
23 (a)P = (i(E2 + E3 ), 0, 0, 0).
Again, applying 23 (/4) ((E7 ), )(0,E1 ,0,1) of Lemma 4.11.5 on the above, we
have
23 (/4)23 (a)P = (0, iE1 , 0, i).
This shows the transitivity of ((E7 ), )(0,E1 ,0,1) . The isotropy subgroup of
((E7 ), )(0,E1 ,0,1) at (0 iE1 , 0, i) is Spin(10) (Lemma 4.11.6). Thus we have the
homeomorphism ((E7 ), )(0,E1 ,0,1) /Spin(10) S 10 .
Theorem 4.11.8.
Spin(10)
p
SO(10)
S 10
=
S 10
we see that p : ((E7 ), )(0,E1 ,0,1) SO(11) is onto by the five lemma. Using the five
lemme again, we see that Ker p coincides with Ker p . Hence Ker p = {1, } (Theorem
3.10.4). Thus we have the isomorphism
((E7 ), )(0,E1 ,0,1) /{1, }
= SO(11).
Therefore the group ((E7 ), )(0,E1 ,0,1) is isomorphic to the group Spin(11) as the
universal covering group of SO(11).
Lemma 4.11.9. For t R, we define a mapping (t) : PC PC by
1 y3 y 2
1 x3 x2
(t) x3 2 x1 , y 3 2 y1 , ,
x2 x1 3
y2 y 1 3
2it
2it
it
it
e
1 eit y3 eit y 2
e 1 e x3 e x2
2
x1 , eit y 3
2
y1 , e2it , e2it ,
= eit x3
x1
3
y1
3
eit x2
eit y2
n 0 0
0
x ,0
=
0 x
0
= P}
0
0
0
0
o
0 , 0, x C, , C
0
1
{P, P } = xx + ( ) + ( ).
2
(E7 ), /Spin(11) S 11 .
138
0
0 0
0 0
x , 0 0 0 , 0, S 11 ,
P = 0
0 x
0 0 0
we choose t R such that e2it iR. Applying (t) of Lemma 4.11.9 on P , we get
(t)P S 10 .
Since the group Spin(11) acts transitively on S 10 (Proposition 4.11.7), there exists
Spin(11) = ((E7 ), )(0,E1 ,0,1) such that
(t)P = (0, iE1 , 0, i).
If we further apply (/4) (E7 ), of Lemma 4.11.9 on the above, then we have
(/4)(t)P = (0, E1 , 0, 1).
This shows the transitivity of (E7 ), . The isotropy subgroup of (E7 ), at (0, E1 , 0, 1)
is Spin(11) (Theorem 4.11.8). Thus we have the homeomorphism (E7 ), / Spin(11)
S 11 .
(E7 ),
= Spin(12).
Theorem 4.11.11.
Spin(11)
p
SO(11)
(E7 ),
p
SO(12)
S 11
=
S 11
we see that p : (E7 ), SO(12) is onto by the five lemma. Using the five lemme
again we see that Ker p coincides with Ker p . Hence Ker p = {1, } (Theorem 4.11.8).
Thus we have the isomorphism
(E7 ), /{1, }
= SO(12).
139
Therefore the group (E7 ), is isomorphic to the group Spin(12) as the universal
covering group of SO(12).
Theorem 4.11.12. The center z(Spin(12)) of Spin(12) is
z(Spin(12)) = {1, 1, }
= {1, 1} {1, }
= Z2 Z2.
And we have
Spin(12)/{1, }
= SO(12),
Spin(12)/{1, 1}
= Spin(12)/{1, }
= Ss(12).
Theorem 4.11.13. The group (E7 ) contains a subgroup
2 (SU (2)) = {2 (A) E7 | A SU (2)}
which is isomorphic to the special unitary group SU (2) = {A M (2, C) | t ( A)A =
E, det A = 1}. Here, for A SU (2), a mapping 2 (A) : PC PC is defined by
1 y3 y 2
1 x3 x2
2 (A) x3 2 x1 , y 3 2 y1 , ,
3
x2 x1 3
y2 y
1
1 y3 y 2
1 x3 x2
= x3 2 x1 , y3 2 y1 , , ,
x2 x1 3
y2 y 1 3
where
1
3
3
1
2
2
,
,
=A
=A
,
,
=A
=A
2
2
1
3
3
1
x3
x2
x1
x3
x2
x1
.
,
,
=
=
= A
y3
y3
y2
y2
y1
y1
a
1
1
=
,
=
,
1
a
1
a
3
3
a
2
2
,
=
,
=
2
a
2
3
a
3
x1
a
x1
x2
x3
0
=
,
=
=
.
y1
a
y2
y1
y3
0
a
Therefore, for A = exp
SU (2), we have
a
1
X = 0
0
0
0 ,
3
i C.
Proof. Recall (0, aEi , aEi , 0) (e7 ) , i = 1, 2, 3, then we can prove analogously as in Proposition 4.8.2.
Claim 3.
(E7 ) /(E6 ) (M1 ) = {P M1 | P = P }.
In particular, the group (E7 ) is connected.
Proof. Remark that i (a) of Lemma 4.8.1 belongs to ((E7 ) )0 , then this claim is
proved analogously as Theorem 4.8.3. The connectedness of (E7 ) follows from the
connectedness of (E6 )
= (U (1) Spin(10))/Z 4 (Theorem 3.10.7) and (M1 ) .
141
= { E7 | = } = (E7 ) .
(PC ) = {P PC | P = P }
= i(PC ) .
These spaces (PC ) , (PC ) are invariant under the action of (E7 ) and we have
the decomposition
PC = (PC ) (PC ) = (PC ) i(PC ) .
In paticular, PC is the complexification of (PC ) : PC = ((PC ) )C .
Analogious to Section 3.11, we can define the R-linear mapping k : M (4, H)
M (8, C),
a b
k a + be2 =
, a, b C.
b a
Lemma 4.12.1. Any element B su(8) is uniquely expressed by
B = k(D) + e1 k(T ),
Proof. For B su(8), let D1 =
have
B = D1 + e1 T 1 ,
B JBJ
B + JBJ
M (8, C), then we
, T1 =
2
2e1
D1 = D1 , JD1 = D1 J,
T1 = T1 , JT1 = T 1 J, tr(T1 ) = 0.
implies
C1 = T1 = 0.
tr(M )
ia
2
, the homomorphism : Sp(4) (E6 ) , (A)X =
1
ia
M tr(M )E
2
1
= [1 , 2 ] 2A1 A2 + 2A2 A1 ,
A = 1 A2 2 A1 .
Proposition 4.12.3.
(e7 )
= su(8).
P PC .
P = (X, Y, , )
k gX E J + e1 k g(Y ) E J
2
2
k(D)k gX E J + e1 k(D)k g(Y ) E J
2
2
t
t
+k gX E J k(D) + e1 k g(Y ) E J k(D)
2
2
= k D gX E J + e1 k D g(Y ) E J
2
2
+k gX E D J + e1 k g(Y ) E D J
2
2
= k(D(gX) + (gX)D )J + e1 k(D(g(Y ) + (g(Y ))D )J
= k(g( (D)X))J + e1 k(g( (D)(Y )))J
(recall (C)X = g 1 (C(gX) + (gX)C ))
(D)X
(D)X
0
0
(D)
(D)Y
=
=
0
0
0
0
0
0
= (( (D), 0, 0, 0)P ).
0
0
0
0
0
X
0 Y
0
P = (X, Y, , )
k gX E J + e1 k g(Y ) E J
2
2
e1 k(T )k gX E J k(T )k g(Y ) E J
2
2
t
t
+e1 k gX E J k(T ) k g(Y ) E J k(T )
2
2
= k(T g(Y ) g(Y )T + T )J + e1 k(T (gX) + (gX)T T )J
= k(2gA g(Y ) + gA)J + e1 k(2gA gX gA)J
1
= k 2g(A Y ) (A, Y )E + gA J
2
1
+e1 k 2g(A X) + (A, X)E gA J (Lemma
2
2A Y + A
0
2A
0
0
A
2A X A
2A
=
=
(A, Y )
0
A
0
(A, X)
A
0
0
= ((0, A, A, 0)P ).
3.12.1)
A
X
0 Y
0
(E7 )
= SU (8)/Z 2 , Z 2 = {E, E}.
P PC .
We first prove (A) (E7 ) . To prove this, for the differential mapping : su(8)
(e7 ) of ,
(D)P = 1 (D(P ) + (P )t D), P PC ,
it is sufficient to show that is well-defined, that is, (D) (e7 ) . However
this fact is already shown in Proposition 4.12.3. Evidently : SU (8) (E7 ) is a
homomorphism. Since : su(8) (e7 ) is onto and (E7 ) is connected (Lemma
4.12.4), : SU (8) (E7 ) is also onto. Ker = {E, E} = Z 2 is easily obtained.
Thus we have the isomorphism SU (8)/Z 2
= (E7 ) .
Remark. Without using Lemma 4.12.4, the fact that the mapping : SU (8)
(E7 ) is onto will be followed from two claims.
Claim 1. For a R, i (a) of Lemma 4.8.1 belongs to (SU (8)).
Proof. i (a) = exp((0, aEi , aEi , 0)) exp (su(8)) (Proposition 4.12.3) =
(exp(su(8))) (SU (8)).
X=
1
(Y Y ),
= ,
= .
Since Y (JC ) , we have Y (JC ) , so that g(Y ) J(4, H)0 . Hence, there
exists D Sp(4) such that
D(g(Y ))D
is real diagonal.
145
Then
(D)Y = g 1 (D(g(Y ))D ) is real diagonal,
1
1
(D)X = (D) Y Y = ((D)Y (D)Y ) is real diagonal.
In the case = 0, by the same proof of Proposition 4.8.2, we can choose (SU (8))
so that
P = (X, Y, , ), X, Y are real diagonal, 0 6= R.
If < 0, apply 1 () of Claim 1 on it, then becomes > 0.
Now, we will return to the proof of the surjection of : SU (8) (E7 ) using
Claims 1, 2. For a given (E7 ) , consider the element P = 1 (MC ) . We
first transform P to a diagonal form (Claim 2) by some (SU (8)), and we have,
in a similar way to Theorem 4.8.3,
E7
a, b R.
a C, v V.
a, b C.
Now, let 3 (C 6 ) be the third exterior product of C-vector space C 6 and we define
a C-C-linear isomorphism f : (PC )C 3 (C 6 ) by
1 y3 y 2
1 x3 x2
X
f x3 2 x1 , y 3 2 y1 , , =
xijk ei ej ek
x2 x1 3
y2 y 1 3
i<j<k
{e1 , e2 , , e6 } is the canonical basis of C 6 and xijk C are skew-symmetric tensor:
i j k
xi j k = sgn
xijk , where
i j k
147
x156 = h (1 ),
x164 = h (x3 ),
x145 = h (x2 ),
x256 = h (x3 ),
x264 = h (2 ),
x245 = h (x1 ),
x356 = h (x2 ),
x364 = h (x1 ),
x345 = h (3 ),
x423 = h (1 ),
x431 = h (y3 ),
x412 = h (y 2 ),
x523 = h (y 3 ),
x531 = h (2 ),
x512 = h (y1 ),
x623 = h (y2 ),
x631 = h (y 1 ),
x612 = h (3 ),
x123 = h (),
x456 = h ().
The inverse mapping f 1 : 3 (C 6 ) (PC )C of f
h(x156 )
h(x256 , x164 )
h(x356 , x145 )
X
h(x423 )
f 1
xijk ei ej ek =
h(x523 , x431 )
i<j<k
h(x623 , x412 )
is given by
h(x164 , x256 )
h(x264 )
h(x364 , x245 )
h(x431 , x523 )
h(x531 )
h(x631 , x512 )
h(x123 )
h(x456 )
h(x145 , x356 )
h(x245 , x364 )
h(x345 )
h(x412 , x623 )
,
h(x512 , x631 )
h(x612 )
a C, P (PC )C .
The group SU (6) acts naturally on 3 (C 6 ), that is, the action of A SU (6) on
a b c 3 (C 6 ) is defined by
A(a b c) = Aa Ab Ac.
Hence, the action of D su(6) on 3 (C 6 ) is given by
D(a b c) = Da b c + a Db c + a b Dc.
Lemma 4.13.2. (1) Any element D su(6) is uniquely expressed by
E
B
L
0
D=
+
, B, C su(3), L M (3, C), e1 R.
L C
3 0 E
148
(2) The Lie algebra e7,C is isomorphic to the Lie algebra su(6) as Lie algebras:
e7,C
= su(6).
This isomorphism is given by the mapping C : su(6) e7,C ,
E
B
L
0
+
C
= (C (B, C), h(L), h(L), ie1 )
L C
3 0 E
where C : su(3) su(3) e6,C is defined by C (B, C)X = h(B, C)X + Xh(B, C) ,
X (JC )C (Lemma 3.13.3).
B
L
Proof. (1) For D =
su(6), B , C u(3), L M (3, C), we let
L C
= tr(B ) = tr(C ),
B = B
E,
3
C = C +
E,
3
A11
A21
A12
A22
0
E
E
0
A11
A21
A12
A22
0
E
E
0
1
A22 A21
A12 A11
where E, Aij M (3, C), and let SU (6) Z 2 be the semi-direct product of the groups
SU (6) and Z 2 under this action.
Proposition
4.13.3. E7,C
= (SU (6)/Z 3 ) Z 2 , Z 3 = {E, 1 E, 1 2 E}, 1 =
1
3
+
e1 .
2
2
Proof. We define a mapping : SU (6) Z 2 E7,C by
(A, 1)P = f 1 (A(f P )),
P (PC )C .
We first have to show that (A, 1) E7,C . To prove this, it is sufficient to show that
the differential mapping : su(6) e7,C of :
(D)P = f 1 (D(f P )),
P (PC )C
b11
b12
b13
l11
l12
l13
b12
b22
b23
l21
l22
l23
b13
b23
b33
l31
l32
l33
l11
l21
l31
c11
c12
c13
l12
l22
l32
c12
c22
c23
l13
l23
l33
c13
c23
c33
E
+
3 0
0
E
su(6),
where bii = bii , cii = cii , b11 + b22 + b33 = c11 + c22 + c33 = 0, = .
(1) For P = (0, 0, 1, 0), (D)P is calculated as follows.
P = (0, 0, 1, 0)
f
e1 e2 e3
D
+ b33 +
e1 e2 e3 l31 e4 e1 e2 l32 e5 e1 e2 l33 e6 e1 e2
3
0 0 0
0 0 0
0
0
0
h(L)
h(l11 )
h(l21 , l12 ) h(l31 , l13 )
f 1
,
l
)
h(l
)
h(l
,
l
)
h(l
12
21
22
32
23
ie1
h(l33 )
h(l13 , l31 ) h(l23 , l32 )
h()
ie1
(B, C) +
C
3
2h(L)
=
0
h(L)
2 h(L)
C (B, C)
h(L)
0
ie1
3
h(L)
ie1
0
0
0
0
1
0
0
ie1
h(L)
e1 e5 e6
D
h(c12 , b12 ) h(c13 , b13 )
h b11 c11
3
h(b12 , c12 )
0
0
0
0
h(b13 , c13 )
f 1
0
0
0
0
h(l
)
h(l
33
23 , l32 )
0 h(l32 , l23 )
h(l22 )
h(l11 )
ie1
E1
E1
3
2h(L) E1
= (C (B, C), h(L), h(L), ie1 )
0
0
( h(L), E )
C (B, C)E1 +
e2 e4 e5 + e3 e6 e4
D
h(b23 + c23 )
h(c12 , b12 )
0
h(l11 )
h(l21 , l12 )
0
h(l23 + l32 )
151
ie1
F1 (1)
F1 (1)
3
2h(L) F1 (1)
= (C (B, C), h(L), h(L), ie1 )
0
0
(h(L), F (1))
(4) For the other generator of PC C , that is, P = (X, 0, 0, 0), (0, X, 0, 0) where
X = Ei , Fi (1), Fi (e1 ), i = 1, 2, 3 and P = (0, 0, 0, 1), we have also
f 1 (D(f P )) = (C (B, C), h(L), h(L), ie1 )P.
Thus we see that (A, 1) E7,C for A SU (6). Since (E, )P = P , we see
(E, ) = G2,C (= Aut(C)) F4,C E6,C E7,C . We shall show that
: SU (6) Z 2 E7,C is a homomorphism. For this purpose, we first show
f 1 (A(f P )) = f 1 (((AdJ3 )Af P )),
A SU (6), P (PC )C .
does not hold. Therefore E7,C has just two connected components (see Lemma 4.13.1).
Hence : SU (6) Z 2 E7,C is onto. Ker = {E, 1 E, 1 2 E} 1 = Z 3 1 easily
obtained. Thus we have the isomorphism (SU (6)/Z 3 ) Z 2
= E7,C .
We identify (PC )C (M (3, C)C M (3, C)C ) with PC (using the identification
JC C M (3, C)C with JC in Section 3.13) by
((X, Y, , ), (M, N )) = (X + M, Y + N, , ).
152
M21 M22
(M M )e
M21 + M12 (M22 + M11 )e1
M22 M11
21
12 1
,
=
+i
,
+i
2
2
2
2
where Mij M (3, C). The inverse mapping 1 : M (3, C)C M (3, C)C M (6, C)
of is given by
N2 N1 e1 M2 + M1 e1
1
, Mi , Ni M (3, C).
(M1 + iM2 , N1 + iN2 ) =
M 2 M 1 e1 N2 N1 e1
f M (6, C), we have
Lemma 4.13.4. For D su(6) and M
fD ) = (D)(M
f).
(M
Proof. Let
E
B
L
0
+
su(6),
L C
3 0 E
f = N2 N1 e1 M2 + M1 e1 , Mi , Ni M (3, C),
M
M 2 M 1 e1 N2 N1 e1
M = M1 + iM2 , N = N1 + iN2 .
D=
Then we have
f)
(D)(M
1
2 h(L)
C (B, C) + ie1
3
1
2h(L)
C (B, C) ie1
=
0
h(L)
h(L)
0
1
C (B, C)M + ie1 M 2 h(L) N
3
(h(L), N )
( h(L), M )
1
M h(B, C) + N h(L) + ie1 M
3
=
ie
N
M
h(L)
h(B,
C)
153
0
h(L)
ie1
0
M
N
0
0
0
0
ie1
h(L)
by simple calculations
N2 N1 e1 M2 + M1 e1
B
=
L
M 2 + M 1 e1 N2 + N1 e1
by
L
C
E
0
0
E
fD .
=M
f means
where QM
Q 0
0 Q
M11
M21
M12
M22
QM11
QM21
QM12
QM22
P PC .
We first have to prove that (Q, A) E7 . To prove this, since (Q, E) (E6 )w
(E7 )w , it suffices to show that (E, A) E7 . Moreover, it is sufficient to show that,
for the differential mapping : su(3) su(6) e7 of , (0, D) coincides with
(C (B, C), h(L), h(L), ie1 ) e7 .
However, this is already shown in Proposition 4.13.3 and Lemma 4.13.4. Since
(PC )w = {P PC | wP = P } = (PC )C , obviously we have w(Q, A) = (Q, A)w,
hence (Q, A) (E7 )w . Evidently is a homomorphism. We shall show that is
onto. Let (E7 )w . Since the restriction to (PC )w = (PC )C of belongs to
E7,C , there exists A SU (6) such that
P = f 1 (A(f P ))
or P = f 1 (A(f P )),
P (PC )C
(Proposition 4.13.3). In the former case, let = (E, A)1 , then |(PC )C = 1,
hence G2 . Furthermore, (G2 )w = SU (3) (Theorem 1.9.4), so there exists
Q SU (3) such that
(PC + (M + N )) = PC + Q(M + N ) = PC + (QM + QN )
= (Q, E)(PC + (M + N )),
154
PC + (M + N ) PC
Hence we have
= (E, A) = (E, A)(Q, E) = (Q, A).
In this case, this shows that is onto. In the latter case, consider the mapping
1 : PC PC , 1 (PC + (M + N )) = PC + (M + N ),
PC + (M + N ) PC .
d = 133.
Since E7 is simply connected (Theorem 4.9.2), E7 C is also simply connected. The Lie
algebra of the group E7 C is e7 C , so E7 C is a complex simple Lie group of type E7 .
4.15. Non-compact exceptional Lie groups E7(7) , E7(5) and E7(25) of
type E7
155
Let
P = J(3, C) J(3, C) R R,
P = J(3, C ) J(3, C ) R R.
For P, Q P or P , we define an R-linear mapping P Q : P P or P P as
similar to Section 4.1. And we define a Hermitian inner product hP, Qi in PC by
hP, Qi = hP, Qi.
Now, we define groups E7(7) , E7(5) and E7(25) by
E7(7) = { IsoR (P ) | (P Q)1 = P Q},
E7(5)
= (E7 C ) ,
E7(25)
= (E7 C ) .
Theorem 4.15.1. The polar decompositions of the groups E7(7) , E7(5) and
E7(25) are respectively given by
E7(7) SU (8)/Z 2 R70 ,
Proof. These are the facts corresponding to Theorems 4.12.5, 4.11.15 and 4.10.2.
Theorem 4.15.2. The centers of the groups E7(7) , E7(5) and E7(25) are the
group of order 2:
z(E7(7) ) = Z 2 ,
z(E7(5) ) = Z 2 ,
156
z(E7(25) ) = Z 2 .
then e8 C
= [1 , 2 ] + P1 Q2 P2 Q1
P = 1 P2 2 P1 + r1 P2 r2 P1 + s1 Q2 s2 Q1
Q = 1 Q2 2 Q1 r1 Q2 + r2 Q1 + t1 P2 t2 P1
1
1
r = {P1 , Q2 } + {P2 , Q1 } + s1 t2 s2 t1
8
8
P = (0, P, 0, 0, 0, 0),
r = (0, 0, 0, r, 0, 0),
t = (0, 0, 0, 0, 0, t).
157
a [P , [1 , [1 , R] ] ] = [P , [1 , (0, 0, 0, s, 0, 2r)] ]
= [P , (0, 0, 0, 2r, 2s, 0)] = (0, 2rP, 0, 0, 0, 0).
Hence this case is reduced to the case (ii) above.
(iv) Case R = (0, 0, 0, 0, s, t), s 6= 0. We have
a [1 , R] = (0, 0, 0, s, 0, 0).
Hence this case is reduced to the case (iii) above.
(v) Case R = (0, 0, 0, 0, 0, t), t 6= 0. The argument is similar to (iv).
159
R, Ri e8 C .
[, 1 ] + P Q1 P1 Q
2
P1 1 P + rP1 r1 P + sQ1 s1 Q
P
Q1 1 Q rQ1 + r1 Q + tP1 t1 P 2 !
Q2
1
1
{P,
Q
}
+
{P
,
Q}
+
st
s
t
=
1
1
1
1
8
8
r2 8
{P, P1 } + 2rs1 2r1 s
s2
4
1
t2
{Q, Q1 } 2rt1 + 2r1 t
4
= (using ([, 1 ], 2 )7 + (1 , [, 2 ])7 = 0 (Lemma 4.5.1.(1)),
(, P Q)7 = {P, Q} (Lemma 4.5.1.(2)) etc.)
= (R1 , [R, R2 ])8 .
Theorem 5.3.2. The Killing form B8 of the Lie algebra e8 C is given by
B8 (R1 , R2 )
= 15(R1 , R2 )8
= 15(1 , 2 )7 + 15{Q1, P2 } 15{P1 , Q2 } + 120r1 r2 + 60t1 s2 + 60s1 t2
5
= B7 (1 , 2 ) + 15{Q1, P2 } 15{P1 , Q2 } + 120r1 r2 + 60t1 s2 + 60s1 t2 ,
3
where Ri = (i , Pi , Qi , ri , si , ti ) e8 C and B7 is the Killing form of e7 C .
Proof. Since e8 C is simple (Theorem 5.2.1), there exist k C such that
B8 (R1 , R2 ) = k(R1 , R2 )8 ,
Ri e8 C .
1
B8 (R, R1 )R,
30
R1 e8 C ,
161
1
e 1 , R2 ).
B8 ( R
15
1
[Fe1 (ei ), Fe1 (ej )],
2
0 i 7,
0i<j7
162
e = R is equivalent to = , Q = P, r = r, t = s,
the condition R
hence we have the theorem.
Proposition 5.5.4. The complexification of the Lie algebra e8 is e8 C . Hence e8
is simple.
Proof. For R e8 C , the conjugate transposed mapping R of R with respect to
e
e e8 C , and for R e8 C , R belongs
the inner product hR1 , R2 i of e8 C is R = R
to e8 if and only if R = R. Now, any element R e8 C is represented by
R=
R R
R + R
+i
,
2
2i
R R R + R
,
e8 .
2
2i
d = 248.
(k l ), (k + l ), 0 k < l 3,
1
0 k 3,
k (2 3 ),
2
1
1
(0 1 + 2 3 ) (3 1 ),
2
2
1
1
(0 + 1 + 2 3 ) (3 1 ),
2
2
1
1
(0 + 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 + 3 ) (3 1 ),
2
2
1
1
( 0 1 + 2 3 ) (1 2 ),
2
2
1
1
(0 + 1 + 2 3 ) (1 2 ),
2
2
1
1
( 0 + 1 + 2 + 3 ) (1 2 ),
2
2
1
1
(0 1 + 2 + 3 ) (1 2 ),
2
2
2
0 k < l 3,
k + ,
3
1
2
k
1 , 0 k 3,
2
3
1
2
1
2 ,
(0 1 + 2 3 )
2
3
21
1
2
( 0 + 1 + 2 3 )
2 ,
2
2
3
1
1
2
(0 + 1 + 2 + 3 )
2 ,
2
3
21
1
2
( 0 1 + 2 + 3 )
2 ,
2
2
3
1
1
2
( 0 1 + 2 3 )
3 ,
2
3
12
1
2
(0 + 1 + 2 3 )
3 ,
2
2
3
2
1
1
3 ,
( 0 + 1 + 2 + 3 )
2
3
12
1
2
(0 1 + 2 + 3 )
3 ,
2
2
3
1
j + r ,
1 j 3,
1 3 1
k
1 + r, 0 k 3,
2
3
1
1
1
( 0 + 1 2 3 )
2 + r,
2
3
12
1
1
2 + r,
( 0 + 1 2 3 )
2
2
3
164
1
(
2
1
(
2
1
(
2
1
(
2
1
(
2
1
(
2
with 1 + 2 + 3 = 0.
1
2 + r,
3
12
1
0 + 1 2 + 3 )
2 + r,
2
3
1
1
0 + 1 + 2 + 3 )
3 + r
3
12
1
0 + 1 2 3 )
3 + r,
2
3
1
1
0 1 + 2 + 3 )
3 + r,
3
12
2
0 1 2 + 3 )
3 + r,
2
3
2r,
r
0 + 1 + 2 3 )
1
By putting Y = 0, = = 0, we obtain
1
the root k + r
3
1
1
the root k 1 + r
2
3
by letting X = Ek ,
by letting X = F1 (a), a = ek ie4+k .
8
2
=2
=2
=2
=0
=0
=0
3
3
3
0
0
0
4
4
4
0
0
0
3
3
3
0
0
0
2
2
2
0
0
0
1
1
2
0
1
1
0
1
1
1
1
0
2
2
2
0
0
0
1
0 + (2 3 ) = 2
2
1
1 + (2 3 ) = 0
2
0 + 1
0 + 2
0 + 3
1 + 2
1 + 3
2 + 3
=2
=2
=2
=0
=0
=0
4
4
4
1
1
1
6
6
6
2
2
2
6 5
3 2
1 3
2 2
1 1
1 1
166
5
5
5
2
2
2
4
4
4
2
2
2
3
3
2
2
1
1
2
1
1
1
1
0
3
3
3
1
1
1
1
2 + (2 3 ) = 0 1 2 2
2
1
3 + (2 3 ) = 0 1 2 2
2
1
0 (2 3 ) = 2 3 4 3
2
1
1 (2 3 ) = 0 0 0 0
2
1
2 (2 3 ) = 0 0 0 0
2
1
3 (2 3 ) = 0 0 0 0
2
1
1
(0 + 1 + 2 3 ) + (3 1 ) = 1
2
2
1
1
(0 + 1 2 + 3 ) + (3 1 ) = 1
2
2
1
1
(0 1 + 2 + 3 ) + (3 1 ) = 1
2
2
1
1
(0 1 2 3 ) + (3 1 ) = 1
2
2
1
1
(0 + 1 + 2 3 ) (3 1 ) = 1
2
2
1
1
(0 + 1 2 + 3 ) (3 1 ) = 1
2
2
1
1
(0 1 + 2 + 3 ) (3 1 ) = 1
2
2
1
1
(0 1 2 3 ) (3 1 ) = 1
2
2
1
1
(0 1 + 2 3 ) + (1 2 ) = 1
2
2
1
1
(0 1 2 + 3 ) + (1 2 ) = 1
2
2
1
1
(0 + 1 2 3 ) + (1 2 ) = 1
2
2
1
1
(0 + 1 + 2 + 3 ) + (1 2 ) = 1
2
2
1
1
(0 1 + 2 3 ) (1 2 ) = 1
2
2
1
1
(0 1 2 + 3 ) (1 2 ) = 1
2
2
1
1
(0 + 1 2 3 ) (1 2 ) = 1
2
2
1
1
(0 + 1 + 2 + 3 ) (1 2 ) = 1
2
2
2
1 + = 0 1 1 0 0
3
2
2 + = 0 0 1 1 0
3
2
3 = 0 1 1 1 0
3
167
1 1
0 1
1 0
0 1
3 2
1 2
1 1
1 0
1 1
0 0
1 0
0 0
1 2
2 2
2 1
1 2
2 2
1 1
1 2
2 2
1 0
0 0
0 0
0 0
3 4
3 2
2 1
3 4
3 2
1 1
3 4
3 2
1 0
2 2
1 0
0 0
2 2
1 1
1 0
2 2
1 1
0 0
2 2
1 1
1 1
3 4
3 3
2 1
1 2
2 1
1 0
1 2
2 1
0 0
1 2
2 1
1 1
2 4
4 3
2 1
0 1
0 1
0 0
1
0 + 1
2
1
0 1 +
2
1
1 + 1
2
1
1 1 +
2
3
2
3
2
3
2
1
2 + 1
2
1
2 1 +
2
1
3 + 1
2
1
3 1 +
2
3
2
3
2
3
2
1
(0 + 1 + 2 3 ) +
2
1
(0 + 1 2 + 3 ) +
2
1
(0 1 + 2 + 3 ) +
2
1
(0 1 2 3 ) +
2
1
(0 + 1 + 2 3 )
2
1
(0 + 1 2 + 3 )
2
1
(0 1 + 2 + 3 )
2
1
(0 1 2 3 )
2
1
(0 + 1 + 2 + 3 ) +
2
1
(0 + 1 2 3 ) +
2
1
(0 1 + 2 3 ) +
2
1
(0 1 2 + 3 ) +
2
1
(0 + 1 + 2 + 3 )
2
1
(0 + 1 2 3 )
2
1
(0 1 + 2 3 )
2
1
(0 1 2 + 3 )
2
=2
4 5
3 2
1 2
=2
3 5
3 2
1 3
=0
1 1
1 1
1 0
=0
0 1
1 1
1 1
=0
1 1
1 1
0 0
=0
0 1
1 1
0 1
=0
1 1
1 0
0 0
=0
0 1
1 0
0 1
1
2
2
1
2
2
1
2
2
1
2
2
1
2 +
2
1
2 +
2
1
2
2
1
2 +
2
1
3
2
1
3
2
1
3
2
1
3
2
1
3 +
2
1
3 +
2
1
3 +
2
1
3 +
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
3
2
168
=1
2 3
3 2
2 1
=1
2 3
3 2
1 1
=1
2 3
3 2
1 0
=1
1 1
1 0
0 0
=1
2 3
2 2
2 1
=1
2 3
2 2
1 1
=1
2 3
2 2
1 0
=1
1 1
0 0
0 0
=1
2 3
3 3
2 1
=1
1 1
1 1
1 1
=1
1 1
1 1
1 0
=1
1 1
1 1
0 0
=1
3 5
4 3
2 1
=1
2 3
2 1
1 1
=1
2 3
2 1
1 0
=1
2 3
2 1
0 0
1
1 + r = 0
3
1
2 + r = 0
3
1
3 + + r = 0
3
1
1 r = 0
3
1
2 + r = 0
3
1
3 r = 0
3
1 1
0 0
0 0
0 1
1 0
0 0
1 2
1 0
0 1
1 0
0 0
0 0
0 0
1 0
0 0
1 1
1 0
0 1
1
+r =2 4 6 4
3
1
r =2 4 5 4
3
1
+r =2 3 5 4
3
1
r =2 3 4 4
3
1
+r =0 1 2 1
3
1
r =0 1 1 1
3
1
+r =0 0 1 1
3
1
r =0 0 0 1
3
1
+r =0 1 2 1
3
1
r =0 1 1 1
3
1
+r =0 0 1 1
3
1
r =0 0 0 1
3
1
+r =0 1 2 1
3
1
r =0 1 1 1
3
1
+r =0 0 1 1
3
1
r =0 0 0 1
3
1
1
1
(0 + 1 + 2 3 ) + 2 + + r = 1
2
2
3
1
1
1
(0 + 1 + 2 3 ) + 2 + r = 1
2
2
3
1
0 + 1 +
2
1
0 + 1 +
2
1
0 1
2
1
0 1
2
1
1 + 1 +
2
1
1 + 1 +
2
1
1 1
2
1
1 1
2
1
2 + 1 +
2
1
2 + 1 +
2
1
2 1
2
1
2 1
2
1
3 + 1 +
2
1
3 + 1 +
2
1
3 1
2
1
3 1
2
169
2 1
2 1
2 1
2 1
1 1
1 1
1 1
1 1
1 0
1 0
1 0
1 0
0 0
0 0
0 0
0 0
4 3
2 2
1 2
3 3
2 2
1 2
1
(0 + 1 + 2 3 )
2
1
(0 + 1 + 2 3 )
2
1
(0 + 1 2 + 3 ) +
2
1
(0 + 1 2 + 3 ) +
2
1
(0 + 1 2 + 3 )
2
1
(0 + 1 2 + 3 )
2
1
(0 1 + 2 + 3 ) +
2
1
(0 1 + 2 + 3 ) +
2
1
(0 1 + 2 + 3 )
2
1
(0 1 + 2 + 3 )
2
1
(0 1 2 3 ) +
2
1
(0 1 2 3 ) +
2
1
(0 1 2 3 )
2
1
(0 1 2 3 )
2
1
(0 + 1 2 3 ) +
2
1
(0 + 1 2 3 ) +
2
1
(0 + 1 2 3 )
2
1
(0 + 1 2 3 )
2
1
(0 1 + 2 3 ) +
2
1
(0 1 + 2 3 ) +
2
1
(0 1 + 2 3 )
2
1
(0 1 + 2 3 )
2
1
(0 + 1 + 2 + 3 ) +
2
1
(0 + 1 + 2 + 3 ) +
2
1
2
2
1
2
2
1
2 +
2
1
2 +
2
1
2
2
1
2
2
1
2 +
2
1
2 +
2
1
2
2
1
2
2
1
2 +
2
1
2 +
2
1
2
2
1
2
2
1
3 +
2
1
3 +
2
1
3
2
1
3
2
1
3 +
2
1
3 +
2
1
3
2
1
3
2
1
3 +
2
1
3 +
2
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
170
=1 2
3 2
2 2
1 1
=1 2
2 2
2 2
1 1
=1 2
4 3
2 1
1 2
=1 2
3 3
2 1
1 2
=1 2
3 2
2 1
1 1
=1 2
2 2
2 1
1 1
=1 2
4 3
2 1
0 2
=1 2
3 3
2 1
0 2
=1 2
3 2
2 1
0 1
=1 2
2 2
2 1
0 1
=1 1
2 1
0 0
0 1
=1 1
1 1
0 0
0 1
=1 1
1 0
0 0
0 0
=1 1
0 0
0 0
0 0
=1 1
2 1
1 1
1 1
=1 1
1 1
1 1
1 1
=1 2
3 2
1 1
1 1
=1 2
2 2
1 1
1 1
=1 1
2 1
1 1
0 1
=1 1
1 1
1 1
0 1
=1 2
3 2
1 1
0 1
=1 2
2 2
1 1
0 1
=1 2
4 3
3 2
1 2
=1 2
3 3
3 2
1 2
1
1
(0 + 1 + 2 + 3 ) 3
2
2
1
1
(0 + 1 + 2 + 3 ) 3
2
2
1
1
(0 1 2 + 3 ) + 3 +
2
2
1
1
(0 1 2 + 3 ) + 3 +
2
2
1
1
(0 1 2 + 3 ) 3
2
2
1
1
(0 1 2 + 3 ) 3
2
2
r =0 0
2r
=0 0
+r =0 0
1
+r
3
1
r
3
1
+r
3
1
r
3
1
+r
3
1
r
3
0 0
1 0
1 0
=1 3
5 4
3 2
1 2
=1 3
4 4
3 2
1 2
=1 1
2 1
1 0
0 1
=1 1
1 1
1 0
0 1
=1 2
3 2
1 0
0 1
=1 2
2 2
1 0
0 1
0
0
0
0 0
0 0
0 0
1
0
1.
3
3
X
X
j Ej , 0, 0, (h7 )R (Theorem 4.6.2) . The
k Hk +
where (h) = {
j=1
k=0
3
X
k k + 30
3
X
j j + 40 + 120rr ,
j=1
k=0
3
X
e
e
for h = ((h), 0, 0, r, 0), h = ((h ), 0, 0, r , 0) h, where (h) =
k Hk +
3
X
j=1
j Ej
k=0
3
3
X
X
, 0, 0, , (h ) = (
j Ej , 0, 0, (h7 )R , Indeed,
k Hk +
j=1
k=0
5
B8 (e
h, e
h ) = B7 ((h), (h )) + 120rr (Theorem 5.3.2)
3
3
3
X
5 X
j j + 4 + 120rr (Theorem 4.6.2)
k k + 3
= 6 6
3
j=1
k=0
= 60
3
X
k=0
k k + 30
3
X
j j + 40 + 120rr .
j=1
1
H3 = 0, 0, 0, , 0, 0 ,
60
1
1
1
, 0, 0,
(E1 + 2E2 E3 ) , 0, 0,
, 0, 0 ,
H4 =
90
120
120
1
(H3 (E2 E3 ) ), 0, 0, 0 , 0, 0, 0, 0, 0 ,
H5 =
60
1
H6 =
(H2 H3 ), 0, 0, 0 , 0, 0, 0, 0, 0 ,
60
1
(H1 H2 ), 0, 0, 0 , 0, 0, 0, 0, 0 ,
H7 =
60
1
1
, 0, 0,
, 0, 0 .
H8 = 0, 0, 0,
40
120
Thus we have
(1 , 1 ) = B7 (H1 , H1 ) = 60
1 1
1
1 1
4 + 30
8=
,
120 120
120 120
30
and the other inner products are similarly calculated. Consequently, the inner product
induced by the Killing form B8 between 1 , 2 , , 8 and are given by
(i , i ) =
1
,
30
i = 1, 2, 3, 4, 5, 6, 7, 8,
1
1
, i = 1, 2, , 6, (3 , 8 ) = ,
60
60
(i , j ) = 0, otherwise,
1
1
(, ) =
, (, 7 ) = , (, i ) = 0, i = 1, 2, 3, 4, 5, 6, 8,
30
60
(i , i+1 ) =
using them, we can draw the Dynkin diagram and the extended Dynkin diagram of
e8 C .
According to Borel-Siebenthal theory, the Lie algebra e8 has five subalgebras as
maximal subalgebras with the maximal rank 8.
(1) The first one is a subalgebra of type A1 E7 which is obtained as the fixed
points of an involution of e8 .
(2) The second one is a subalgebra of type D8 which is obtained as the fixed points
e of e8 .
of an involution
(3) The third one is a subalgebra of type A2 E6 which is obtained as the fixed
points of an automorphism w of order 3 of e8 .
(4) The fourth one is a subalgebra of type A8 which is obtained as the fixed points
of an automorphism w3 of order 3 of e8 .
(5) The fifth one is a subalgebra of type A4 A4 which is obtained as the fixed
points of an automorphism z5 of order 5 of e8 .
Proposition 5.7.1.
Ad
0
0
e =
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
,
0
0
1
where (Ad) = 1 , e7 C . It is easy to check that e (E8 C )1,1 ,1 . Conversely, if E8 C satisfies 1 = 1, 1 = 1 and 1 = 1 , then is of the
form
1 HomC (e7 C ),
1 12 13 0 0 0
21 2 23 0 0 0
2 , 3 , 23 , 32 HomC (PC ),
21 , 31 HomC (e7 C , PC ),
31 32 3 0 0 0
=
,
a1
b1
c1 1 0 0
12 , 13 HomC (PC , e7 C ),
a2
b2
c2 0 1 0
ai HomC (e7 C , C),
a3
b3
c3 0 0 1
bi , ci HomC (PC , C).
From the relation [, 1] = [, 1] = [, 1] = 0, that is,
0 = [(1 , 21 , 31 , a1 , a2 , a3 ), (0, 0, 0, 1, 0, 0)]
= (0, 21 , 31 , 0, 2a2 , 2a3 ),
we obtain 21 = 31 = 0 and a2 = a3 = 0. Furthermore, from [, 1 ] = [, 1 ] =
[, 1 ] = 0, that is,
0 = [(1 , 0, 0, a1 , 0, 0), (0, 0, 0, 0, 1, 0)] = (0, 0, 0, 0, 2a1, 0),
we obtain a1 = 0. Using the fact that [P , 1] = [P , 1] = [P , 1] = P ,
that is,
(12 P, 2 P, 32 P, b1 P, b2 P, b3 P )
= [(12 P, 2 P, 32 P, b1 P, b2 P, b3 P ), (0, 0, 0, 1, 0, 0)]
= (0, 2 P, 32 P, 0, 2b2P, 2b3 P ),
we obtain 12 = 32 = 0 and b1 = b2 = b3 = 0. Similarly, from [Q , 1] =
[Q , 1] = [Q , 1] = Q , we obtain 13 = 23 = 0 and c1 = c2 = c3 = 0.
173
1
0
=
0
0
0
2
0
0
0
0
3
0
0
0
.
0
E
1
By applying on [(0, P, 0, 0, 0, 0), (0, 0, Q, 0, 0, 0)] = P Q, 0, 0, {P, Q}, 0, 0 , we
8
obtain
1 (P Q) = 2 P 3 Q,
{2 P, 3 Q} = {P, Q},
(i)
1
since [(0, 2 P, 0, 0, 0, 0), (0, 0, 3Q, 0, 0, 0)] = 1 (P Q), 0, 0, {P, Q}, 0, 0 . Again,
8
1
by applying on [(0, P, 0, 0, 0, 0), (0, Q, 0, 0, 0, 0)] = 0, 0, 0, {P, Q}, 0, 0 , we obtain
4
{2 P, 2 Q} = {P, Q}.
(ii)
(iii)
P, Q PC ,
e = }
e (Proposition 5.7.1)
= { E7 C |
1 0
0
0
0
0
0 a1 b1
0
0
0
0 b1 a1
0
0
0
.
3 (A) =
0
( a)a ( b)b ( a)b a( b)
0 0
0 0
0
2a( b)
a2
( b)2
0 0
0
2( a)b
b2
( a)2
175
a b
i
= exp
b a
exp((0, 0, 0, i, , )) (E8 ) .
Proof. For A =
(E8 )
= (SU (2) E7 )/Z 2 , Z 2 = {(E, 1), (E, 1)}.
Ker = {(E, 1), (E, 1)} = Z 2 is easily obtained. Thus we have the isomorphism
(SU (2) E7 )/Z 2
= (E8 ) .
Remark. We can prove directly that is onto without using the connectedness
of (E8 ) (Lemma 5.7.5), (see Imai and Yokota [13]).
e and subgroup Ss(16) of E8
5.8. Involution
e : e8 C e8 C by
We define a C-linear mapping
e
(,
P, Q, , s, t) = (1 , Q, P, r, t, s).
e E8 and ()
e 2 = 1.
Then
e
e
= { E8 |
= }
(E8 )e
= { E8 | = } = (E8 ) .
J = diag(J, , J), J =
0 1
1 0
B + JBJ
B JBJ
+
and use (1) above.
2
2
The following Lemmas 5.8.2 and 5.8.3 are properties of the mappings : (PC )
S(8, C) and : su(8) (e6 ) of Section 4.12, and will be used in the proof of
Theorem 5.8.4.
Lemma 5.8.2 The Lie isomorphism : sp(4) (e6 ) defined by ( D)X =
g 1 (D(gX) + (gX)D ), X JC of Theorem 3.1.2 satisfies
([gX1 , gX2 ]) = 2(X1 X2 X2 X1 ),
177
X1 , X2 JC .
X JC
1
= g((X2 , X)X1 + (X1 , X2 )X 4X2 (X1 X)) (Lemma 3.4.1)
3
1
= (X2 , X)gX1 + (X1 , X2 )gX 4gX2 (gX1 gX) + (X1 , X)gX2
3
+(X2 , X1 , X)E (Lemma 3.12.1),
we have
g(2(X1 X2 X2 X1 )X)
(1)
(2)
(3)
1 (S) = 1 (iS).
tr(S1 S2 S2 S1 ) = 4i{1S1 , 1 S2 }.
1
(S1 S2 S2 S1 ) tr(S1 S2 S2 S1 )E
8
= 4(1 S1 1 S2 1 S2 1 S1 ).
1
1
2
= k gX1 E + ik g(Y1 ) E
k gX2 E
2
2
2
2 2
ik g(Y2 ) E J
2
2
1
2
1
= k gX1 E gX2 E + g(Y1 ) E g(Y2 ) E
2
2
2
2
1
2
1
2
ik g(Y1 ) E gX2 E gX1 E g(Y2 ) E .
2
2
2
2
Therefore
S1 S2 S2 S1
(denote
+ik(g(2X1 Y2 2X2 Y1 + 1 X2 2 X1 1 Y2 + 2 Y1 ))
i
+ ((X1 , Y2 ) (X2 , Y1 ) + 1 2 2 1 ))E (Lemma 3.12.1)
2
D = [gX1 , gX2 ] [g(Y1 ), g(Y2 )] sp(4),
A = 2X1 Y2 2X2 Y1 + 1 X2 2 X1 1 X2 + 2 Y1 JC )
i
= kD + ik(gA) + {P1 , P2 }E.
2
Taking the trace of both sides, we obtain
tr(S1 S2 S2 S1 ) = 4i{P1 , P2 } = 4i{1 S1 , 1 S2 }.
and the expression above equal to
1
S1 S2 S2 S1 tr(S1 S2 S2 S1 ) = kD + ik(gA),
8
On the other hand, we have
Therefore,
X2
Y1
X1 Y2
P1 P2 =
2
1
2
1
1
(Y1 Y2 X2 X1 )
(2X1 Y2 + 1 X2 + 2 Y1 )
4
=
.
1
(2Y1 X2 + 1 Y2 + 2 X1 )
4
1
179
P1 P2 P2 P1
1
(X1 X2 + X2 X1 Y1 Y2 + Y2 Y1 )
(2X1 Y2 2X2 Y1 + 1 X2 2 X1 + 2 Y1 1 Y2 )
= 4
(2X2 Y1 2X1 Y2 + 2 X1 1 X2 + 1 Y2 2 Y1 )
4
0
1
1
1
= D, A, A, 0 (Lemma 5.8.2)
4
4
4
1
= (kD + ik(gA))
4
1
1
= S1 S2 S2 S1 tr(S1 S2 S2 S1 )E .
4
8
is given by
of the Lie group (E )e
Theorem 5.8.4. The Lie algebra (e8 )e
8
e
e
(e8 )e
= { (e8 C ) |
= }
1
({21 S1 , 21 S2 } + {21 S2 , 21 S1 }),
8
1
1
{21 S1 , 21 S2 }, {21 S1 , 21 S2 } ,
4
4
which equals to the above.
e
of the simply connected Lie group E8 . Therefore, by Theorem 5.8.4, (E8 )e
is
isomorphic to one of the following groups
Spin(16),
SO(16),
Ss(16),
SO(16)/Z 2 .
181
e
e
e
(e8 C )e
= { Der(e8 C ) |
= }
= R} = (e8 C )e
,
= {R e8 C | R
e
e
e
= R},
(e8 C )e
= { Der(e8 C ) |
= }
= {R e8 C | R
Since we have
,
((e8 C )e
) = ((e8 C ) )e
= (e8(8) )e
= (e8(8) )e
e
((e8 C )e
) = ((e8 C ) )e
= (e8(8) )e
,
e
e8(8) = (e8(8) )e
(e8(8) )e
.
Since (e8(8) )e
= Ss(16) (Theorem 5.8.4), this is the Cartan decompposition of e8(8) .
e
to
] (e8(8) )e
, we obtain a representation of (e8(8) )e
Since [(e8(8) ) , (e8(8) )e
(e8(8) )e
:
which is irreducible. (See, for example, (8.5.1) of Goto and Grosshans [11]). Furthermore, the complex representation C of to (((e8(8) )e
)C = (e8 C )e
is also
2:
) = {1, e
}.
z((E8 )e
e
Proof. Evidently, {1, }
z((E8 )e
). Conversely, let z((E8 )e
). Since
e
C
the representation of (E8 ) to (e8 )e
irreducible
(Lemma
5.8.5),
we
see,
by using
is constant.
Schurs lemma in the theory of groups, that the action of on (e8 C )e
R (e8 C )e
.
182
which implies that k 2 = 1. By Theorem 5.8.4, we have (e8 C )e
= so(16, C) which is
C e
C e
simple, and hence we see that (e8 ) is generated by (e8 ) :
nX
o
[Rk , Rl ] | Rk , Rl (e8 C )e
=
(e8 C )e
.
k,l
SO(16),
Ss(16).
21
135
2 22 3
4
5
6
7 8
120 5304 560 1820 4368 8008 128 128
has
16 + and 16 are not representation of SO(16). Now, by Lemma 5.8.5, (E8 )e
C e
(E8 )e
= Ss(16).
group (E8 )e
, the group (E8 ) have to the semi-spinor group Ss(16). However the
e
proof of (E8 )
= (E8 ) is not concretely. M. Gomyo [10] find the group
= Ss(16)
183
Ss(16) explicitly in the group E8 (although the definition of the group E8 is different
from E8 in Section 5.5).
5.9. Center z(E8 ) of E8
Theorem 5.9.1. The center z(E8 ) of the group E8 is trivial :
z(E8 ) = {1}.
X1
o
n
(JC )3 = X = X2 Xi JC .
X3
X2 Y3 Y2 X3
X Y = X3 Y1 Y3 X1 (JC )3 ,
X1 Y2 Y1 X2
X Y = X1 Y1 + X2 Y2 + X3 Y3 e6 C ,
X1
Y1
where X = X2 , Y = Y2 (JC )3 . Further, for HomC (JC ), D = dij
X3
Y3
X1
M (3, C) and X = X2 (JC )3 , elements X, DX (JC )3 are naturally defined
X3
184
by
X1
X = X2 ,
X3
1
1
D = [D1 , D2 ] + X 1 Y 2 X 2 Y 1
4
4
1
1
= [1 , 2 ] + X 1 Y 2 X 2 Y 1
2
2
X = 1 X 2 2 X 1 + D1 X 2 D2 X 1 Y 1 Y 2
Y = t 1 Y 2 + t 2 Y 1 t D1 Y 2 + t D2 Y 1 + X 1 X 2 ,
C
Proof. Let ee8 = e7 C PC PC C C C be the C-Lie algebra constructed
C
in Theorem 5.1.1. We define a mapping f : ee8 e8 C by
1
1
2
2
2
2A
2B
1
1
= r
, , Z , Y ,
t
2
3
1
X
W
1
s
+r
2
3
then we can prove that f is an isomorphism as Lie algebras by straightforward calC
culations. Thus we have the isomorphism ee8
= e8 C .
C
Using the Killing form of ee8 which is obtained in Theorem 5.3.2, we see that the
Killing form B8 of e8 C is given by
5
B8 (R1 , R2 ) = 60tr(D1 D2 ) + B6 (1 , 2 ) + 15(X 1 , Y 2 ) + 15(X 2 , Y 1 )
2
Then we have
5
hR1 , R2 i = 60tr(D1 ( t D2 )) + B6 (1 , t2 ) + 15hX 1 , X 2 i + 15hY 1 , Y 2 i.
2
Now, as in Theorem 5.5.3, we see that
E8 = { Aut(e8 C ) | hR1 , R2 i = hR1 , R2 i}
is a simply connected compact Lie group of type E8 .
We define a C-linear transformation w of e8 C by
w(D, , X, Y ) = (D, , X, 2 Y ),
3
1
i C. Then w E8 and w3 = 1.
where = +
2
2
Now, we shall study the following subgroup (E8 )w of E8 :
(E8 )w = { E8 | w = w}.
Theorem 5.10.2. (E8 )w
= (SU (3) E6 )/Z 3 , Z 3 = {(E, 1), (E, 2 1), ( 2 E,
1)}.
Proof. We first define a mapping 1 : SU (3) (E8 )w by
1 (A)(D, , X , Y ) = (ADA1 , , AX, tA1 Y ).
We have to prove that 1 (A) (E8 )w . Indeed, since the action of D1 = (D1 , 0, 0, 0)
su(3) sl(3, C) e8 C is given by
(adD1 )(D, , X, Y ) = ((adD1 )D, 0, D1 X, tD1 Y ),
for A = exp D1 , we have 1 (A) = exp(ad(D1 )) Aut(e8 C ). And from
tr(AD1 tA( t (AD2 tA))) = tr(AD1 ( tD2 )A1 ) = tr(D1 ( t D2 )),
hAX, AY i = hX, Y i,
we see that 1 (A) E8 . Evidently, w1 (A) = 1 (A)w, hence, 1 (A) (E8 )w . Next,
we define a mapping 2 : E6 (E8 )w by
2 ()(D, , X , Y ) = (D, 1 , X, t 1 Y ).
We have to prove that 2 () (E8 )w . Indeed, since the action of an element =
(0, , 0, 0) e6 e6 C e8 C is given by
(ad )(D, , X, Y ) = (0, (ad ), X, t Y ),
for = exp , we have 2 () = exp(ad( )) Aut(e8 C ). And from
186
B6 (1 1 , t (2 1 ) ) = B6 (1 1 , t 2 1 ) = B6 (1 , t 2 ),
hX, t 1 Y i = hX, Y i,
we see that 2 () E8 . Evidently, w2 () = 2 ()w, hence, 2 () (E8 )w . Now,
we define a mapping : SU (3) E6 (E8 )w by
(A, ) = 1 (A)2 ().
Since 1 (A) and 2 () commute, is a homomorphism. It is not difficult to show
that Ker = {(E, 1), (E, 2 1), ( 2 E, 1)} = Z 3 . Finally, since (E8 )w is connected
as the fixed points subgroup by automorphims w of the simply connected group E8
and
e = R}
(e8 )w = {R e8 C | wR = R, R
= {(D, , 0, 0) e8 C | D su(3), e6 }
= su(3) e6 ,
a, b 0 (C n ) = C.
v nk (C n ).
u k (C n ).
A1 = 1.
k
X
j=1
x1 Dxj xk ,
D1 = 0.
nk
(u, v)x,
n
x Cn.
(2) t (u v) = v u,
(u v) = (u) (v).
D = [D1 , D2 ] + u1 v 2 u2 v 1
u = D1 u2 D2 u1 + (v1 v 2 )
v = t D1 v 2 + t D2 v 1 (u1 u2 ),
Proof. In order to prove the Jacobi identity, we need the following Lemma.
Lemma 5.11.4. For u, v, w 3 (C 9 ), we have
(1) u (v w) + v (w u) + w (u v) = 0,
188
(i)
Denote x =
P9
(u1 u9 , e1 e9 )x = (ii).
i=1
xi ei , uj =
P9
k=1
ujk ek and U = ujk M (9, C). Hence we have
9
X
k=1
u
ejk xk ,
where u
ejk is the cofactor of ujk of the matrix U . Therefore
X
X
(ii) =
xk u
ejk uj (detU )x =
xk u
ejk uji ei (detU )x
j,k
X
j,k
i,j,k
189
+(v, w)(u, a)
3 X
3
X
i=1 j=1
From Lemmas 5.11.1, 5.11.2 and 5.11.4, we can prove that e8 C is a C-Lie algebra.
Therefore Theorem 5.11.3 is proved.
We will show that the Lie algebra e8 C given in Theorem 5.11.3 is also the Lie
algebra of type E8 . Since we can not give explicit isomorphism between this e8 C and
e8 C of Theorem 5.1.1, we shall show that this e8 C is simple.
Theorem 5.11.5. e8 C = sl(9, C) 3 (C 9 ) 3 (C 9 ) is a simple C-Lie algebra of
type E8 .
Proof. We use the decomposition
e8 C = sl(9, C) q,
q = 3 (C 9 ) 3 (C 9 ).
0 = Dv = 3
I69
vJ eJ + 6
J69
I9
vJ eJ ,
J9
1
(Eii + Ejj + Ekk ) Ell ,
3
l 6= i, j, k.
Then we can reduce this case to the case (2). In case v 6= 0, we ca similarly reduce
to the case (2).
Thus the simplicity of g has been proved. Since the dimension of g is 248, we see that
g is a Lie algebra of type E8 .
Proposition 5.11.6. The Killing form B8 of the Lie algebra e8 C = sl(9, C)
3 (C 9 ) 3 (C 9 ) is given by
B8 ((D1 , u1 , v 1 ), (D2 , u2 , v 2 )) = 60(tr(D1 D2 ) + (u1 , v 2 ) + (u2 , v 1 )).
Proof. We consider a symmetric bilinear form B of e8 C :
B((D1 , u1 , v 1 ), (D2 , u2 , v 2 )) = tr(D1 D2 ) + (u1 , v2 ) + (u2 , v 1 ).
Using Lemmas 5.11.2, 5.11.4, we see that B is e8 C -adjoint invariant. Since e8 C is
simple, there exists k C such that B8 (R1 , R2 ) = kB(R1 , R2 ) for all Ri e8 C . To
determined k, let R = R1 = R2 = (E11 E22 , 0, 0) e8 C . Then we have
B8 (R, R) = 120,
B(R, R) = 2.
Therefore k = 60.
e of e8 C by
We define a complex-conjugate linear transformation
e
(D,
u, v) = ( t D, v, u).
Then we have
1
3
i C. Then, w3 E8 and w3 3 = 1.
where = +
2
2
Now, we study the following subgroup (E8 )w3 of E8 :
(E8 )w3 = { E8 | w3 = w3 }.
(E8 )w3
= SU (9)/Z 3 , Z 3 = {E, E, 2 E}.
Theorem 5.11.7.
the differential of is onto. Since (E8 )w3 is connected, is also onto. It is not
difficult to see that ker = {E, E, 2 E} = Z 3 . Thus we have the isomorphism
SU (9)/Z 3
= (E8 )w3 .
E8
[g0 , g0 ] g0
[g0 , g1 ] g1
[g0 , g2 ] g2
[g0 , g2 ] g2
[g0 , g1 ] g1
[g1 , g1 ] g0
[g2 , g2 ] g0
[g1 , g1 ] g2
[g1 , g1 ] g2
[g2 , g2 ] g1
[g2 , g2 ] g1
[g1 , g2 ] g2
[g1 , g2 ] g2
[g2 , g1 ] g1
[g2 , g1 ] g1
(1) a (b c) + b (c a) + c (a b) = 0,
(7) (a x) y (a y) x + a (x y) = 0,
1
(8) (a b)c = ((a c) b) (a, b)c (b, c)a,
5
3
(9) (x y)a = (y (x a)) + (x, y)a.
5
Proof. (1) Let a = a1 a2 , b = a3 a4 , c = a5 a6 and ai =
((a (b c)), x) = (a, (b c) x)
P5
j=1
aij ej . Since
5 X
6
X
j=1 i=1
5
X
j=1
a1j
a2j
det
a6j
a11
a21
a61
a15
a25
ej = 0.
a65
3
= (v a, w (b x)) ((a b), x)(v, w)
5
194
(i)
3
= (x a, w (b v)) ((a b), x)(v, w)
5
= (v, w)((a b), v) + (a1 , w)(b x w a2 , e1 e5 )
3
(a2 , w)(b x w a1 , e1 e5 ) ((a b), x)(v, w).
5
Using (i), we have
1
(a (b x))v + (b (a x))v = (x, w) (a b) ((a b), x)v
5
= (x (a b))v.
(7) We have
3
((a (x y))v, w) = (x y w, v a) (a, x y)(v, w)
5
2
= (x, v)(y w, a) (y, v)(x w, a) + (a, x y)(v, w)
5
2
= (x, v)((a y), w) (y, v)((a x), w) + (a, x y)(v, w).
5
On the other hand, using (i), we have
1
((a x) y)v = (y, v) (a x) ((a x), v)y,
5
1
= (y, v) (a x) (a, x v)y,
5
1
((a x) y)v = (x, v) (a y) (a, x v)y.
5
Hence (7) is proved.
(8) Let a = a1 a2 and c = c1 c2 . Since
3
((a b)v, w) = (a v, b w) (a, b)(v, w)
5
2
= (a1 , w)(x a2 , b) + (a2 , w)(x a1 , b) + (a, b)(v, w),
5
we have
(a b)c = (a1 c1 , b)a1 c2 + (a1 c2 , b)a2 c1
4
+(a2 c1 , b)a1 c2 (a2 c2 , b)a1 c1 + (a, b)c.
5
195
q = g1 g2 g2 g1 .
B(R, R) = 2.
Therefore k = 60.
e of e8 C by
We define a complex-conjugate linear transformation
e
(C,
D, x a, b y, c z, w d)
= ( t C, t D, w d, c z, b y, x a).
197
Then, we have
e8(24) = e7(25) P P R R R,
E8(24)
= (E8 C ) .
Theorem 5.13.1. The polar decompositions of the Lie groups E8(8) and E8(24)
are respectively given by
z(E8(24) ) = {1}.
199
References
[1] K. Abe and I. Yokota, Realization of spaces E6 /(U (1)Spin(10)), E7 /(U (1)E6 ),
E8 /(U (1)E7 ) and their volumes, Tokyo J. Math. 20(1997), 73-86.
[2] K. Abe and I. Yokota, Volumes of compact symmetric spaces, Tokyo J. Math.
20(1997), 87-105.
[3] M.F.Atiyah, R.Bott and A.Shapiro,
1(1964).
[4] A. Borel and J. de Siebenthal, Les sous-groupes fermes de rang maximum des
groupes de Lie clos, Comment. Math. 23(1949-50), 200-221.
[5] C. Chevalley, Theory of Lie groups I, Princeton Univ. Press, 1946.
[6] H. Freudenthal, Oktaven, Ausnahmegruppen und Oktavengeometry, Math. Inst.
Rijkuniv. te Utrecht, 1951.
[7] H. Freudenthal, Beziehung der E7 und E8 zur Oktavenevene I, II, X, Nedel,
Akad. Weten. Proc.Ser. A.57(1954), 218-230, 363-365, 66(1963), 457-471.
[8] F. Gantmacher. On the classifications of real simple Lie groups, Rec. Math. J.
5(1939).
[9] S. Gomyo, Realization of maximal subgroups of rank 8 of the simply connected
compact simple Lie group of type E8 , Tsukuba J. Math. 21(1997), 595-616.
[10] S. Gomyo, Realizations of subgroups of type D8 of connected exceptional simple
Lie groups of type E8 , Tsukuba J. Math. 23(1999), 585-615.
[11] M. Goto and F. D. Grosshans, Semi-simpleLie algebras, Dekker, 1978.
[12] T. Imai, On bounded symmetric domains of exceptional type, J. Fac. Sci. Shinshu
Univ. 16(1981), 65-96.
[13] T. Imai and I. Yokota, Non-compacct simple Lie group E8(24) of type E8 , J.
Fac. Sci. Shinshu Univ. 15(1980), 53-76.
[14] T. Imai and I. Yokota, Another definitions of exceptional simple Lie groups of
type E7(25) and E7(133) , J. Fac. Sci. Shinshu Univ. 15(1980), 47-52.
[15] T. Imai and I. Yokota, Simply connected compact simple Lie group E7(133)
of type E7 , J. Math. Kyoto Univ. 21(1981), 383-395.
[16] T. Imai and I. Yokota, Simply connected compact simple Lie group E8(248) of
type E8 , J. Math. Kyoto Univ. 21(1981), 741-762.
200
[17] S. Kaneyuki, On the subalgebra g0 and gev of semisimple Lie algebra, J. Math.
Soc.Japan, 45(1993), 1-19.
[18] Y. Matsushima, Some remarks on exceptional Lie group F4 , Nagoya Math. J.
4(1954), 83-88.
[19] Y. Matsushima, Theory of Lie algebras (in Japaneae), Kyoritsu-shuppan, Tokyo,
1957.
[20] K. Mitsuishi and I. Yokota, Volume formula of compact simple Lie groups and
values of , J. Fac. Sci. Shinshu Uni. 29(1994), 69-70.
[21] T. Miyasaka, Adjoint orbit types of compact exceptional Lie group G2 in their
Lie algebra, Math. J. Okayama Univ. 43(2001), 17-180.
[22] T. Miyasaka, O. Schukuzawa and I. Yokota, Spinor-generators of compact exceptional Lie groups F4 , E6 and E7 , Tsukuba J. Math. 22(1998), 705-721.
[23] T. Miyasaka, O. Yasukura and I. Yokota, Diagonalization of an element P of
PC by the compact Lie group E7 , Tsukuba J. Math. 22(1998), 687-703.
[24] T. Miyasaka and I. Yokota, Orbit types of the compact Lie group E7 in the
Freudenthal vector space PC , Tsukuba J. Math. 23(1999), 229-234.
[25] T. Miyashita, Decomposition of spinor groups by the involution in exceptional
Lie groups, Math. J. Okayama Univ, 44(2001), 151-158.
[26] T. Miyashita, Fixed points subgroups by two involutive automorphisms ,
of compact exceptional Lie groups F4 , E6 and E7 , Tsukuba J. Math. 27(2003),
199-215.
[27] T. Miyashita and I. Yokota, 2-graded decompositions of exceptional Lie algebra g
and group realizations of gev , g0 , Part III, G = E8 , Japanese J. Math. 26-1(2000),
31-50.
[29] T. Miyashita and I. Yokota, An explicit isomorphism between (e8 ), and so(8)
so(8) as Lie algebras, Math. J. Toyama Univ. 24(2001), 151-158.
[30] T. Miyashita and I. Yokota, 3-graded decompositions of exceptional Lie algebra
g and group realizations of gev , g0 and ged , Part II, G = E7 , Case 1, J. Kyoto
Univ.46-2(2006), 383-413, Case 2, 3, 46-4(2006), 805-832, Case 5, 47-1(2007),
121-128.
201
202
[46] I. Yokota, Simply connected compact simple Lie group E6(78) of type E6 and
its involutive automorphisms, J. Math. Kyoto Univ. 20(1980), 447-473.
[47] I. Yokota, Subgroup (SU (8))/Z 2 of compact simple Lie group E7 and noncompact simple Lie group E7(7) of type E7 , Math. J. Okayama Univ. 24(1982),
53-71.
[48] I. Yokota, Realization of automorphisms of order 3 and G of compact exceptional Lie groups G, I, G = G2 , F4 , E6 , J. Fac. Sci. Shinshu Univ. 20(1985),
131-144.
[49] I. Yokota, Realization of involutive automorphisms of exceptional Lie groups
G, Part I, G = G2 , F4 and E6 , Part II, G = E7 , Part III, G = E8 , Tsukuba J.
Math. 14(1990), 185-223, 14(1990), 379-404, 15(1991), 301-314.
[50] I. Yokota, Orbit types of the compact Lie group E6 in the complex exceptional
Jordan algebra JC , Proc. Symp. on non-associative algebras, Japan, 1991, 353359.
[51] I. Yokota, 2-graded decompositions of exceptional Lie algebra g and group realizations of gev , g0 , Part I, G = G2 , F4 , E6 , Part II, G = E7 , Japan. J. Math.
24(1998), 258-296, 25(1999), 155-179.
[52] I. Yokota, 3-graded decompositions of exceptional Lie algebra g and group realizations of gev , g0 and ged , Part I, G = G2 , F4 , E6 , J. Math. Kyoto Univ.
41-3(2001), 449-494.
[53] I. Yokota, T, Imai and O. Yasukura, On the homogeneous space E8 /E7 , J. Math.,
Kyoto Univ. 23(1983), 467-473.
[54] I. Yokota, T. Ishihara and O. Yasukura, Subgroup ((SU (3) SU (6))/Z2 ).Z2
of the simply connected compact simple Lie group E7 , J. Math. Kyoto Univ.
23(1983), 715-737.
[55] I. Yokota and O. Shukuzawa, Automorphisms of order 4 of the simply connected
compact Lie group E6 , Tsukuba J. Math. 15(1991), 451-463.
[56] I. Yokota and O. Yasukura, Subgroup (SU (2) Spin(12))/Z 2 of compact simple
Lie group E7 and non-compact simple group E7, of type E7(5) , Hiroshima Univ.
J. 12(1982), 59-76.
[57] I. Yokota and O. Yasukura, Non-compact simple Lie group E8(8) , Tsukuba J.
Math. 10(1986), 331-349.
[58] I. Yokota, Groups and Topology (in Japanese), Shokabo, Yokyo, 1971.
203
[59] I. Yokota, Classical linear simple Lie groups (in Japaneae), Gendai-sugakushya,
Kyoto, 1990.
[60] I. Yokota, Exceptional linear simple Lie groups (in Japaneae), Gendai-sugakusya,
Kyoto, 1992.
has some errors
204