Bourbaki marks this one as harder than the rest.
a) Let μ,ν∈P++ and let γ be a weight of E(μ). Let ρμ be the representation of g on E(μ). Let Xα∈gα -- {0},Yα∈g−α -- {0}. If α∈B, put να=ν(Hα), and
E+(μ,γ,ν)=E(μ)γ∩⋂α∈B Ker ρμ(Xα)να+1,
E−(μ,γ,ν)=E(μ)γ∩⋂α∈B Ker ρμ(Yα)να+1,
d+(μ,γ,ν)= dim E+(μ,γ,ν),d−(μ,γ,ν)= dim E−(μ,γ,ν).
For all λ∈h∗, put λ∗=−w0λ, where w0 is the element of W that takes B to −B. Show that
d+(μ,γ,ν)=d−(μ,−γ∗,ν∗)
b) Let λ1,λ2∈P++,V the g-module Homk(E(λ1∗),E(λ2)),U the set of φ∈V such that Yα.φ=0 for all α∈B, and w a primitive vector in E(λ1∗). Show that φ→φ(w) is an isomorphism from U to the set of v∈E(λ2) such that Yαλ1∗(Hα)+1.v=0 for all α∈B. (Use Exerc. 15 of §7 to prove surjectivity.) c) With the notations of Prop. 2, prove that, for λ1,λ2,λ∈P++,
m(λ1,λ2,λ)=d+(λ,λ2−λ1∗,λ1∗)=d−(λ,λ1−λ2∗,λ1)
=d+(λ1,λ−λ2,λ2)=d−(λ1,λ2∗−λ∗,λ2∗)
(Observe that m(λ1,λ2,λ) is the dimension of the space of g-invariant elements of
E(λ)∗⊗E(λ1)⊗E(λ2)
and hence is equal to m(λ1∗,λ,λ2).)
d) Let λ1,λ2∈P++, and let λ be the unique element of P++∩W.(λ1−λ2∗). We have
m(λ1,λ2,λ)=1
(Use c).) Deduce that E(λ1)⊗E(λ2) contains a unique submodule isomorphic to E(λ).
e) We retain the notations of d). Show the equivalence of the following conditions:
(i) λ=λ1+λ2
(ii) ∥λ∥=∥λ1+λ2∥
(iii) λ1 and λ2∗ are orthogonal
(iv) λ1 and λ2∗ are disjoint (Exerc. 13)
(v) λ1 and λ2 are disjoint.
Conclude that E(λ1)⊗E(λ2) is not a simple module unless λ1 and λ2 are disjoint (and hence another proof of Exerc. 18f) of §7).
Nothing.