Lie Algebras and Lie Groups Homework VI

Pádraig Sheehy
Due: 01-11-2019

Problem 1
Let L be a Lie algebra. Show that
1. Z(L) ⊂ rad(L).
2. If I ⊂ L is a solvable ideal, then rad(L/I) ' rad(L)/I.
3. L/rad(L) is a semisimple Lie algebra.

Solution:
1. We have Z(L) is an abelian ideal by definition, and hence it is also solvable. radL ⊂ L is by definition
the maximal solvable ideal of L. So we must have Z(L) ⊂ radL
2. Let I be a solvable ideal. Consider π : L → L/I the natural projection. Recall that π is a Lie algebra
homomorphism. rad L is an ideal, by definition and so we have:
[rad(L), L] ⊂ rad(L)
=⇒ π([rad(L), L]) ⊂ π(rad(L))
=⇒ [π(rad(L)), π(L)] ⊂ π(rad(L))
=⇒ π(rad(L)) is an ideal in π(L)
=⇒ rad(L)/I is an ideal in L/I

However, rad(L) is solvable by definition, and thus since I is an ideal, rad(L)/I is solvable by Lemma 3.29.
Furthermore, rad(L/I) is the maximal solvable ideal in L/I and thus rad(L)/I ⊂ rad(L/I).
To see the opposite inclusion we note that:
L/I ⊂ L
=⇒ rad(L/I) ⊂ rad(L)
=⇒ π(rad(L/I)) ⊂ π(rad(L))
=⇒ rad(L/I)/I ⊂ rad(L)/I
=⇒ rad(L/I) ⊂ rad(L)/I

Thus, rad(L/I) ' rad(L)/I.

3. rad(L) ⊂ L is a solvable ideal, by definition. So, by part 2. rad(L/rad(L)) ' rad(L)/rad(L) = 0.
Thus the maximal solvable ideal of L/rad(L) is the zero ideal and thus L does not contain solvable ideals
except zero. Thus, by Lemma 3.37, L/rad(L) is semisimple.

Problem 2
Let L be a Lie algebra, W be an L-representation and U ⊂ W, V ⊂ W be subrepresentations. Show that there
is a canonical isomorphism of L-representations
V /(U ∩ V ) ' (U + V )/U

1
Solution:
Consider the map φ : V → U + V → (U + V )/U

φ is a homomorpism of L-representations.
Indeed for all x ∈ L, v ∈ V ,
φ(xv) = [xv]U
= x[v]U
= x(φ(v))
Consider ker φ ⊂ V .
x ∈ ker φ ⇐⇒ φ(x) = 0 ⇐⇒ x ∈ U
Thus ker φ = U ∩ V .

Consider Im φ ⊂ (U + V )/U
y ∈ (U + V )/U ⇐⇒ y ∈ U/U or y ∈ V /U ⇐⇒ y = 0 or φ−1 (y) ⊂ V But, 0 ∈ Im φ, as φ is a homomorphism
of representations.
Thus Im φ = (U + V )/U .

As L is a homomorphism of L-representations there exists an isomorphism φ̄ : V /ker φ → Im φ.

So we get:
V /(U ∩ V ) ' (U + V )/U

Problem 3
Let L be a Lie algebra over a field k and σ : L × L → k be an L-invariant bilinear form. Given an ideal I ⊂ L
show that
1. The subspace I ⊥ = {x ∈ L | σ(x, y) = 0 ∀ y ∈ I} is also an ideal in L.
2. dim I + dim I ⊥ ≥ dim L.
3. If σ is non-degenerate, then dim I + dim I ⊥ = dim L

Solution:
1. For all x ∈ L, y ∈ I, z ∈ I ⊥ ⊂ L, we have:
[x, y] ∈ I
=⇒ σ(z, [x, y]) = 0
=⇒ σ([z, x], y) = 0
=⇒ [z, x] ∈ I ⊥
So, the subspace I ⊥ is an ideal in L.
2. Consider the map φ : L → I ∗ : x 7→ σ(x, −).
We have kerφ = I ⊥ , by definition of I ⊥ . So, we have:
dimL = dim Kerφ + dim Imφ
= dimI ⊥ + dim Imφ
≤ dimI ⊥ + dimI ∗
= dimI ⊥ + dimI
Thus, dim I + dim I ⊥ ≥ dim L.
3. If σ is non-degenerate, then L⊥ = {x ∈ L | σ(x, L) = 0} = 0.
Considering ψ : L → L∗ : x 7→ σ(x, −), we have that ker ψ = L⊥ = 0, and so it is injective, and therefore
it’s bijective, since dimL = dimL∗
Therefore, since I ∗ ⊂ L∗ , we can consider the map φ from part 2 as a surjective map.
Thus Im φ = I ∗ and so dim Im φ = dim I ∗ we have from part 2 that dimI + dimI ⊥ = dimL.

2
Problem 4
Given a representation (V, ρ) of a Lie algebra L, define a bilinear form

σV : L × L → k : σV (x, y) = tr(ρ(x)ρ(y)), ∀x, y ∈ L.

Show that σV is L-invariant. Determine the matrix of σV for the algebra L = sl2 (k) (in the standard basis
(x, y, h)) and the following representations:
1. V = k2 , the standard representation of sl2 (k).
2. V = sl2 (k), the adjoint representation of sl2 (k).

Solution:
We have:

σV ([x, y], z) = tr(ρ([x, y])ρ(z))

= tr([ρ(x), ρ(y)]ρ(z))
= tr((ρ(x)ρ(y) − ρ(y)ρ(x))ρ(z))
= tr(ρ(x)ρ(y)ρ(z) − ρ(y)ρ(x)ρ(z))
= tr(ρ(x)ρ(y)ρ(z)) − tr(ρ(y)ρ(x)ρ(z))
= tr(ρ(x)ρ(y)ρ(z)) − tr(ρ(x)ρ(z)ρ(y))
= tr(ρ(x)(ρ(y)ρ(z) − ρ(y)ρ(x)))
= tr(ρ(x)[ρ(y), ρ(z)])
= tr(ρ(x)ρ([y, z]))
= σV (x, [y, z])

Thus σV is L-invariant.
To compute the matrix A of σV w.r.t. a representation, we evaluate it on the basis elements.
1. ρ(x) = I2 x
We get:

2 0 0
σV (x, x) = tr(x ) = tr( )=0
0 0
 
1 0
σV (x, y) = tr(xy) = tr( )=1
0 0
 
0 −1
σV (x, h) = tr(xh) = tr( )=0
0 0
σV (y, x) = tr(yx) = tr(xy) = 1
 
2 0 0
σV (y, y) = tr(y ) = tr( )=0
0 0
 
0 0
σV (y, h) = tr(yh) = tr( )=0
1 0
σV (h, x) = tr(hx) = tr(xh) = 0
σV (h, y) = tr(hy) = tr(yh) = 0
 
1 0
σV (h, h) = tr(h2 ) = tr( )=2
0 1

Thus,  
0 1 0
A = 1 0 0
0 0 2

3
 2. ρ(x) = adx
From Assignment 1, we have:
     
0 0 −2 0 0 0 2 0 0
adx = 0 0 0  , ady =  0 0 2 , adh = 0 −2 0
0 1 0 −1 0 0 0 0 0

Let X = adx , Y = ady , H = adh . Then:

 
0 −2 0
σV (X, X) = tr(X 2 ) = tr(0 0 0) = 0
0 0 0
 
2 0 0
σV (X, Y ) = tr(XY ) = tr(0 0 0) = 4
0 0 2
 
0 0 0
σV (X, H) = tr(XH) = tr(0 0 0) = 0
0 −2 0
σV (Y, X) = tr(Y X) = tr(XY ) = 4
 
0 0 0
σV (Y, Y ) = tr(Y 2 ) = tr(−2 0 0) = 0
0 0 0
 
0 0 0
σV (Y, H) = tr(Y H) = tr( 0 0 0) = 0
−2 0 0
σV (H, X) = tr(HX) = tr(XH) = 0
σV (H, Y ) = tr(HY ) = tr(Y H) = 0
 
4 0 0
σV (H, H) = tr(H 2 ) = tr(0 4 0) = 8
0 0 0

Thus,  
0 4 0
A = 4 0 0
0 0 8

We note that matrix of σV in the adjoint representation is precisely 4 times the matrix of σV in the standard
representation.