0% found this document useful (0 votes)
50 views3 pages

MATH 223A NOTES 2011 Lie Algebras 17

(1) A Lie algebra is semisimple if it has no nonzero solvable ideals. The Killing form of a Lie algebra is defined using the trace of adjoint maps. (2) A Lie algebra is semisimple if and only if its Killing form is nondegenerate. This is equivalent to having no nonzero abelian ideals and being a direct sum of simple ideals. (3) Any finite dimensional semisimple Lie algebra can be uniquely expressed as a product of simple ideals.

Uploaded by

eddyedwards511
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
50 views3 pages

MATH 223A NOTES 2011 Lie Algebras 17

(1) A Lie algebra is semisimple if it has no nonzero solvable ideals. The Killing form of a Lie algebra is defined using the trace of adjoint maps. (2) A Lie algebra is semisimple if and only if its Killing form is nondegenerate. This is equivalent to having no nonzero abelian ideals and being a direct sum of simple ideals. (3) Any finite dimensional semisimple Lie algebra can be uniquely expressed as a product of simple ideals.

Uploaded by

eddyedwards511
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

MATH 223A NOTES 2011 LIE ALGEBRAS 17

5. Semisimple Lie algebras and the Killing form


This section follows Procesi’s book on Lie Groups. We will define semisimple Lie
algebras and the Killing form and prove the following.
Theorem 5.0.9. The following are equivalent for L a finite dimensional Lie algebra over
any subfield F ⊆ C.
(1) L is semisimple.
(2) L has no nonzero abelian ideals.
(3) The Killing form of L is nondegenerate.
(4) L is a direct sum of simple ideals.
5.1. Definition. First we observe that the sum of two solvable ideals I, J in L is solvable.
This follows from the fact that I and (I + J)/I = J/(I ∩ J) are both solvable.
Definition 5.1.1. The solvable radical Rad L of L is defined to be the sum of all solvable
ideals. A Lie algebra is semisimple if its solvable radical is zero, i.e., if it has no nonzero
solvable ideal.
Proposition 5.1.2. L is semisimple iff L has no nonzero abelian ideals.
Proof. If L is semisimple then it has no abelian ideals. Conversely, if L is not semisimple,
then L has a solvable ideal J. Then DJ = [JJ] is also an ideal in L since
[x[JJ]] ⊆ [[xJ]J] + [J[xJ]] ⊆ [JJ]
k
We have D J = 0. So D k−1
J is a nonzero abelian ideal in L. �
5.2. Killing form. The Killing form κ : L × L → F is defined by
κ(x, y) = Tr(ad x ad y)
The Killing form is clearly symmetric: κ(x, y) = κ(y, x). The Killing form is also “asso-
ciative”:
κ([xy], z) = κ(x, [yz])
Proof. Since ad[xy] = [ad x, ad y], we have:
κ([xy], z) = Tr(ad [xy] ad z) = Tr([ad x, ad y]ad z) = Tr(ad x[ad y, ad z]) = κ(x, [yz])
Proposition 5.2.1. The Killing form is invariant under any automorphism ρ of L.
Proof. The equation ρ[xy] = [ρ(x)ρ(y)] for z = ρ(y) is ρ[x, ρ−1 (z)] = [ρ(x)z] which can be
rewritten as: ad ρ(x) = ρ ◦ ad x ◦ ρ−1 . So,
κ(ρ(x), ρ(y)) = Tr(ad ρ(x) ad ρ(y)) = Tr(ρ ◦ ad x ad y ◦ ρ−1 ) = Tr(ad x ad y) = κ(x, y). �
Lemma 5.2.2. The kernel (also called the radical) of the Killing form of L is an ideal.
The definition of the kernel of κ is:
S = {x ∈ L | κ(x, z) = 0 for all z ∈ L}
Proof. Suppose x ∈ S and y ∈ L. Then κ([xy], z) = κ(x, [yz]) = 0. �
Lemma 5.2.3. Every abelian ideal in L is contained in the kernel of its Killing form.
18 MATH 223A NOTES 2011 LIE ALGEBRAS

Proof. Suppose J ⊆ L is an abelian ideal, x ∈ J and y ∈ L. Then ad x ad y sends L into


J and ad x ad y(J) ⊆ ad x(J) = 0. So, (ad x ad y)2 = 0. Since nilpotent endomorphisms
have zero trace, κ(x, y) = Tr(ad x ad y) = 0 showing that J ⊆ S. �
Theorem 5.2.4 (Cartan). Suppose that L is a finite dimensional Lie algebra over C.
Then L is semisimple iff its Killing form is nondegenerate (its kernel S = 0).
Proof. We will show that S �= 0 iff L is not semisimple. If L is not semisimple then it has
a nonzero abelian ideal. Any such ideal lies in the kernel S. So, S �= 0.
Conversely, suppose that S �= 0. Then Cartan’s criterion shows that the image adL S
of S under the adjoint representation adL : L → gl(L) is solvable since
κ(x, y) = Tr(ad x ad y) = 0
for all x, y ∈ S. Since adL S = S/Z(L), this implies that S is solvable. Therefore, L is not
semisimple. �
Corollary 5.2.5. Suppose that L is a finite dimensional Lie algebra over a subfield F of
C. Then L is semisimple iff its Killing form is nondegenerate.
Proof. Suppose that the Killing form of L is nondegenerate. Then L must be semisimple
since any abelian ideal is contained in the kernel of κ which is zero. Conversely, suppose
that the Killing form of L has a nonzero kernel S.
Let LC = L⊗F C be the complexification of L. Since L ⊆ LC , it is clear that L is abelian
iff LC is abelian. This implies the L is solvable iff LC is solvable since D(LC ) = (DL)C .
One can also see that the Killing form κC of LC is the complexification of the Killing form
κ of L. So, the kernel of κC is SC . We know that SC is solvable from the proof of Cartan’s
Theorem. Therefore S is solvable and L is not semisimple. �
Example 5.2.6. Let L = sl(2, R). This has basis
� � � � � �
0 1 0 0 1 0
x= , y= , h=
0 0 1 0 0 −1
with respect to this basis we have
     
0 0 −2 0 0 0 2 0 0
ad x = 0 0 0  , ad y =  0 0 2 , ad h = 0 −2 0
0 1 0 −1 0 0 0 0 0
Then
κ(x, x) = 0 κ(x, y) = 4 κ(x, h) = 0
κ(y, y) = 0 κ(y, h) = 0
κ(h, h) = 8
The matrix of the form κ is therefore
 
0 4 0
4 0 0
0 0 8
MATH 223A NOTES 2011 LIE ALGEBRAS 19

Since this matrix is invertible, κ is nondegenerate and L = sl(2, R) is semisimple. Since


this matrix has negative determinant and positive trace its signature (#+ eigenvalues −
#− eigenvalues) is 1.
Exercise 5.2.7. Compute the Killing form of the real cross product algebra. Conclude
that this algebra is semisimple but not isomorphic to sl(2, R).
5.3. Product of simple ideals.
Theorem 5.3.1. Suppose that L is a finite dimensional semisimple Lie algebra over any
subfield F ⊆ C. Then L can be expressed uniquely as a product of simple ideals.
Proof. First we should point out that if L is a direct sum of two ideals L = J1 ⊕ J2 then
L∼= J1 × J2 with the isomorphism given by the projection maps L → L/J2 , L → L/J1 .
If L is simple, the statement trivially holds. Otherwise, let J ⊆ L be a minimal ideal.
Define J ⊥ to be the set of all x ∈ L so that κ(x, J) = 0. Then J ⊥ is an ideal since
κ([xy], J) = κ(x, [yJ]) ⊆ κ(x, J) = 0 for all x ∈ J ⊥ , y ∈ L. Therefore, J ∩ J ⊥ is also an
ideal. By minimality of J we have either J ∩ J ⊥ = J or J ∩ J ⊥ = 0. The first case is
not possible since κ = 0 on J ∩ J ⊥ which, by Cartan’s criterion, would imply that JC is
solvable, so J would be solvable.
Since κ is nondegenerate, we have dim J + dim J ⊥ = dim L. Therefore, L =�J ⊕ J ⊥ .
By induction, J ⊥ is a product of simple ideals. So, L = J1 ⊕ J2 ⊕ · · · ⊕ Jn ∼ = Ji . To
prove uniqueness of this decomposition, suppose that I is another minimal ideal. Then
I = [IL] = [IJ1 ] ⊕ [IJ2 ] ⊕ · · · ⊕ [IJn ]
One of these summands must be nonzero. Say [IJi ] ⊆ I ∩ Ji �= 0. Then I = Ji . �
Corollary 5.3.2. Let L be a finite dimensional Lie algebra over F ⊆ C. Then L is
semisimple iff it is a product of simple ideals.
Exercise 5.3.3. (1) Show that L is nilpotent iff its Killing form is identically zero.
(2) Show that the Killing form of a nonabelian 2-dimensional Lie algebra is nontrivial.
(3) For F ⊆ C show that L is solvable iff [LL] is contained in the kernel of κ.
(4) Show that κ nondegenerate implies L semisimple over any field.
(5) For char F = 3 show that sl(3, F ) modulo its center is semisimple but its Killing
form is degenerate.

You might also like