Bourbaki

Exercise 15, § 13

Tag 0198, permanent. Cite it as [Bourbaki, Tag 0198].

Lie Groups and Lie Algebras, VIII, 268

Bourbaki marks this one as harder than the rest.

Let V be a vector space of finite dimension ≥2,g\geq 2,\mathfrak{g} a Lie subalgebra of End(V) and θ\theta an element of g\mathfrak{g}. We make the following assumptions:

(i) V is a semi-simple g\mathfrak{g}-module; (ii) θ\theta is of rank 1 (i.e. dim Im(θ\theta ) =1= 1);

(iii) the line Im(θ\theta ) generates the U(gU(\mathfrak{g})-module V.

a) Show that these assumptions are satisfied (for a suitable choice of θ\theta ) when g=sl\mathfrak{g}=\mathfrak{s}\mathfrak{l}(V), when g=gl\mathfrak{g}=\mathfrak{g}\mathfrak{l}(V), or when there exists a non-degenerate alternating bilinear form Ψ\Psi on V such that g=sp(Ψ)\mathfrak{g}=\mathfrak{s}\mathfrak{p}(\Psi ) or g=k.1⊕sp(Ψ)\mathfrak{g}=k.1\oplus \mathfrak{s}\mathfrak{p}(\Psi ). In each of these cases, θ\theta can be taken to be a nilpotent element; in the second case (and only in that case) θ\theta can be taken to be a semi-simple element.

b) We shall prove that the four cases above are the only ones possible. We are reduced immediately to the case in which kk is algebraically closed. Show that V is then a simple g\mathfrak{g}-module, and that g=c⊕s\mathfrak{g}=\mathfrak{c}\oplus \mathfrak{s}, where s\mathfrak{s} is semi-simple, and c=0\mathfrak{c}= 0 or c=k.1\mathfrak{c}=k.1; show that V is not isomorphic to a tensor product of g\mathfrak{g}-modules of dimension ≥2\geq 2; deduce that s\mathfrak{s} is simple.

c) Choose a Cartan subalgebra h\mathfrak{h} of s\mathfrak{s}, and a basis B of R(s,h)R(\mathfrak{s},\mathfrak{h}). Let λ\lambda be the highest weight (with respect to B) of the s\mathfrak{s}-module V, and let ee be a non-zero element of V of weight λ\lambda. The highest weight of the dual module V∗V^* is λ∗=−w0λ(\lambda^*=-w_0\lambda (§7, no. 5); let e∗e^* be a non-zero element of V∗V^* of weight λ∗\lambda^*. Identify V⊗V∗V\otimes V^* with End(V) as usual. Show that there exist x∈V,y∈V∗x\in V, y\in V^* such that x⊗y∈gx\otimes y\in \mathfrak{g} and ⟨x,e∗⟩≠0,⟨e,y⟩≠0\langle x, e^*\rangle \not= 0,\langle e, y\rangle \not= 0 (take the conjugate of θ\theta by ene^n, where nn is a suitable nilpotent element of g\mathfrak{g}). Use the fact that g\mathfrak{g} is an h\mathfrak{h}-submodule of V⊗V∗V\otimes V^* to conclude that g\mathfrak{g} contains e⊗e∗e\otimes e^*. Deduce that λ+λ∗=α~\lambda +\lambda^*= \widetilde{\alpha}, where α~\widetilde{\alpha} is the highest root of s\mathfrak{s}.

d) Show that s\mathfrak{s} is not of type Bl(l≥2),Dl(l≥4),E6,E7,E8,F4,G2B_l(l\geq 2), D_l(l\geq 4), E_6,E_7,E_8,F_4,G_2 (by Chap. VI, Tables,α~, \widetilde{\alpha} would be a fundamental weight, and hence could not be of the form λ+λ∗\lambda +\lambda^* above). Deduce that s\mathfrak{s} is either of type AlA_l or of type ClC_l, and that in the first case λ=ϖ1\lambda =\varpi_1 or ϖ1=ϖ1∗\varpi_1=\varpi_1^*, and in the second case α~=2ϖ1=ϖ1+ϖ1∗\widetilde{\alpha}= 2\varpi_1=\varpi_1+\varpi^*_1; since c=0\mathfrak{c}= 0 or k.1k.1, this indeed gives the four possibilities in a)21a)^{21}.

Solution

No solution has been written yet.

Cited by

Nothing.

Cites

Transcribed from the printed volume. Not translated.