14.08.2013 Views

TH`ESE A NEW INVARIANT OF QUADRATIC LIE ALGEBRAS AND ...

TH`ESE A NEW INVARIANT OF QUADRATIC LIE ALGEBRAS AND ...

TH`ESE A NEW INVARIANT OF QUADRATIC LIE ALGEBRAS AND ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Consider<br />

Then dim(g) = 6<br />

∑<br />

i=3<br />

2.2. Singular quadratic Lie algebras<br />

g =<br />

⊥<br />

3≤i≤6<br />

ki times<br />

<br />

(<br />

⊥<br />

⊕ ... ⊥<br />

⊕ gi).<br />

gi<br />

iki and by Lemma 2.2.14, dup(g) = 0, so we obtain O(n) = /0 if n ≥ 6.<br />

Finally, let g be a non-Abelian quadratic Lie algebra of dimension 3 or 4 with associated<br />

3-form I. Then I is decomposable, so g is singular. Therefore O(3) = O(4) = /0.<br />

Remark 2.2.16. We shall prove in Appendix B and Appendix C by two different ways that<br />

O(5) = /0. So, generically a non-Abelian quadratic Lie algebra is ordinary if n ≥ 6.<br />

⊥<br />

Definition 2.2.17. A quadratic Lie algebra g is indecomposable if g = g1 ⊕ g2, with g1 and g2<br />

ideals of g, implies g1 or g2 = {0}.<br />

The proposition below gives another characterization of reduced singular quadratic Lie algebras.<br />

Proposition 2.2.18. Let g be a singular quadratic Lie algebra. Then g is reduced if and only if<br />

g is indecomposable.<br />

Proof. If g is indecomposable, by Proposition 2.1.5, g is reduced. If g is reduced and g =<br />

⊥<br />

g1 ⊕ g2, with g1 and g2 ideals of g, then Z(gi) ⊂ [gi,gi] for i = 1,2. So gi is reduced or gi =<br />

{0}. But if g1 and g2 are both reduced, by Lemma 2.2.14, one has dup(g) = 0. Hence g1 or<br />

g2 = {0}.<br />

2.2.3 Quadratic Lie algebras of type S1<br />

Let (g,B) be a quadratic vector space and I be a non-zero 3-form in A 3 (g). As in Subsection<br />

2.2.1, we define a Lie bracket on g by:<br />

[X,Y ] = φ −1 (ιX∧Y (I)), ∀ X,Y ∈ g.<br />

Then g becomes a quadratic Lie algebra with an invariant bilinear form B if and only if {I,I} = 0<br />

[PU07].<br />

In the sequel, we assume that dim(VI) = 1. Fix α ∈ VI and choose Ω ∈ A 2 (g) such that<br />

I = α ∧ Ω as follows: let {α,α1,...,αr} be a basis of WI. Then, I ∈ 3 (WI) by Proposition<br />

2.2.3. We set:<br />

X0 = φ −1 (α) and Xi = φ −1 (αi), 1 ≤ i ≤ r.<br />

So, we can choose Ω ∈ A 2 (V ) where V = span{X1,...,Xr}. Note that Ω is an indecomposable<br />

bilinear form, so dim(V ) > 3.<br />

We define C : g → g by<br />

B(C(X),Y ) = Ω(X,Y ).<br />

Therefore C is skew-symmetric with respect to B.<br />

28

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!