Bourbaki

Lemma 6

Tag 0113, permanent. Cite it as [Bourbaki, Tag 0113].

Lie Groups and Lie Algebras, chapter VIII, § 4. Split semi-simple Lie algebra defined by a reduced root system

Let α,β∈B\alpha , \beta \in B with α≠β\alpha \not=\beta. Then

[a+,yαβ]=0[a−,xαβ]=0 [\mathfrak{a}_+, y_{\alpha \beta}] = 0[\mathfrak{a}_-, x_{\alpha \beta}] = 0

The second formula follows from the first by using the automorphism θ\theta. To prove the first, it suffices to show that [xγ,yαβ]=0[x_{\gamma}, y_{\alpha \beta}] = 0 for all γ∈B\gamma \in B. We distinguish three cases.

Case 1:γ≠α:\gamma \not=\alpha and γ≠β\gamma \not=\beta. In this case, xγx_{\gamma} commutes with x−αx_{-\alpha} and x−βx_{-\beta}, and hence with yαβy_{\alpha \beta}.

Case 2:γ=β:\gamma =\beta. In this case, xγx_{\gamma} commutes with x−αx_{-\alpha}, so

[xγ,yαβ]=[x_{\gamma}, y_{\alpha \beta}] = (ad x−αx_{-\alpha})1−n(β,α)[xγ,x−β]^{1-n(\beta ,\alpha)}[x_{\gamma}, x_{-\beta}]

=−=-(ad x−αx_{-\alpha})1−n(β,α)hβ=−n(α,β^{1-n(\beta ,\alpha)}h_{\beta}=-n(\alpha , \beta)(ad x−αx_{-\alpha})−n(β,α)x−α^{-n(\beta ,\alpha)}x_{-\alpha}.

If n(β,α)<0n(\beta , \alpha )<0, this expression is zero since (ad x−αx_{-\alpha}).x−α=0.x_{-\alpha}= 0. If n(β,α)=0n(\beta , \alpha ) = 0, then n(α,β)=0n(\alpha , \beta ) = 0. In both cases, [xγ,yαβ]=0[x_{\gamma}, y_{\alpha \beta}] = 0.

Case 3:γ=α:\gamma =\alpha. In the algebra of endomorphisms of a\mathfrak{a}, [[-ad hαh_{\alpha}, ad x−α]=2x_{-\alpha}] = 2 ad x−αx_{-\alpha}

and [ad xαx_{\alpha}, ad x−α]=x_{-\alpha}] =-ad hαh_{\alpha}; thus, by §1, Lemma 1,

[ad xαx_{\alpha}, (ad x−αx_{-\alpha})1−n(β,α)]=(1−n(β,α^{1-n(\beta ,\alpha)}] = (1-n(\beta , \alpha))(ad x−αx_{-\alpha})−n(β,α)(^{-n(\beta ,\alpha)}(-ad hα−n(β,α))h_{\alpha}-n(\beta , \alpha )).

Consequently,

[xγ,yαβ]=[x_{\gamma}, y_{\alpha \beta}] = [ad xαx_{\alpha}, (ad x−αx_{-\alpha})1−n(β,α)]x−β+^{1-n(\beta ,\alpha)}]x_{-\beta}+ (ad x−αx_{-\alpha})1−n(β,α)^{1-n(\beta ,\alpha)}(ad xαx_{\alpha})x−βx_{-\beta}

=−(1−n(β,α=-(1-n(\beta , \alpha))(ad x−αx_{-\alpha})−n(β,α)^{-n(\beta ,\alpha)}(ad hα+n(β,α)h_{\alpha}+n(\beta , \alpha ))x−βx_{-\beta}

+ (ad x−αx_{-\alpha})1−n(β,α)^{1-n(\beta ,\alpha)}(ad xαx_{\alpha})x−βx_{-\beta}.

Now [hα,x−β]+n(β,α)x−β=0[h_{\alpha}, x_{-\beta}] +n(\beta , \alpha )x_{-\beta}= 0 and [xα,x−β]=0[x_{\alpha}, x_{-\beta}] = 0, so [xγ,yαβ]=0[x_{\gamma}, y_{\alpha \beta}] = 0.

Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.

Cited by

Cites

Transcribed from the printed volume. Not translated.