Bourbaki

Proposition 13

Tag 0AUT, permanent. Cite it as [Bourbaki, Tag 0AUT].

Lie Groups and Lie Algebras, chapter VI, § 1. Root systems

Assume that R is irreducible, non-reduced and of rank ≥2\geq 2.

(i) The set R0R_0 of indivisible roots is a root system in V; this system is irreducible and reduced; and W(R0)=W(R)W(R_0) = W(R).

(ii) Let A be the set of roots α\alpha for which (α∣α)(\alpha|\alpha) takes the smallest value λ\lambda. Then any two non-proportional elements of A are orthogonal.

(iii) Let B be the set of β∈R\beta \in R such that (β∣β)=2λ(\beta|\beta) = 2\lambda. Then B≠∅B \neq \varnothing, R0=A∪BR_0 = A \cup B, R=A∪B∪2AR = A \cup B \cup 2A.

If α∈R−R0\alpha \in R - R_0, then 12α∈R\frac{1}{2} \alpha \in R, but 12(12α)∉R\frac{1}{2} (\frac{1}{2} \alpha) \notin R (Prop. 8), so 12α∈R0\frac{1}{2} \alpha \in R_0. This proves that R0R_0 satisfies (RS1_1). It is clear that, for all α∈R\alpha \in R, sα,α−(R0)=R0s_{\alpha, \alpha^-}(R_0) = R_0, so R0R_0 satisfies (RSII_{II}) and (RSIII_{III}). Since α∈R−R0\alpha \in R - R_0 implies that 12α∈R0\frac{1}{2} \alpha \in R_0 and since sα=sα/2s_\alpha = s_{\alpha/2}, we have W(R)=W(R0)W(R) = W(R_0). Thus R0R_0 is irreducible (Cor. of Prop. 5), and it is evidently reduced.

Since RR is not reduced, there exists α∈R0\alpha \in R_0 such that 2α∈R2\alpha \in R. Since R0R_0 is irreducible and dim⁡V≥2\dim V \geq 2, α\alpha cannot be proportional or orthogonal to every root. Let β∈R0\beta \in R_0 be such that n(β,α)≠0n(\beta, \alpha) \neq 0 and β\beta is not proportional to α\alpha. Changing β\beta to −β-\beta if necessary, we can assume that n(β,α)>0n(\beta, \alpha) > 0. Now 12n(β,α)=n(β,2α)∈Z\frac{1}{2} n(\beta, \alpha) = n(\beta, 2\alpha) \in \mathbf{Z}, so n(β,α)∈2Zn(\beta, \alpha) \in 2\mathbf{Z}. From the list in no. 3, n(β,α)=2n(\beta, \alpha) = 2, (β∣β)=2(α∣α)(\beta|\beta) = 2(\alpha|\alpha). Since R0R_0 is reduced, Prop. 12 shows that, for all γ∈R0\gamma \in R_0, either (γ∣γ)=(α∣α)(\gamma|\gamma) = (\alpha|\alpha) or (γ∣γ)=2(α∣α)(\gamma|\gamma) = 2(\alpha|\alpha). Also, the above shows that, for all γ∈R−R0\gamma \in R - R_0, the vector 12γ\frac{1}{2} \gamma is an element of R0R_0 such that (12γ∣12γ)=(α∣α)(\frac{1}{2} \gamma|\frac{1}{2} \gamma) = (\alpha|\alpha). Thus, λ=(α∣α)\lambda = (\alpha|\alpha), B≠∅B \neq \varnothing, R0=A∪BR_0 = A \cup B, and R⊆A∪B∪2AR \subseteq A \cup B \cup 2A; on the other hand, if γ∈A\gamma \in A, there exists g∈W(R)g \in W(R) such that γ=g(α)\gamma = g(\alpha) (Prop. 11), so 2γ=g(2α)∈R2\gamma = g(2\alpha) \in R; thus 2A⊆R2A \subseteq R and R=A∪B∪2AR = A \cup B \cup 2A. Finally, let γ,γ′\gamma, \gamma' be two non-proportional elements of AA. Then

n(2γ,γ′)=2n(γ,γ′)=4n(γ,2γ′)∈4Z,and∣n(γ,γ′)∣≤1 n(2\gamma, \gamma') = 2n(\gamma, \gamma') = 4n(\gamma, 2\gamma') \in 4\mathbf{Z}, \quad \text{and} \quad |n(\gamma, \gamma')| \leq 1

since γ\gamma and γ′\gamma' have the same length, so n(γ,γ′)=0n(\gamma, \gamma') = 0 and (γ∣γ′)=0(\gamma|\gamma') = 0.

Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.

Cited by

Cites

Transcribed from the printed volume. Not translated.