Bourbaki

Proposition 3

Tag 01IU, permanent. Cite it as [Bourbaki, Tag 01IU].

Lie Groups and Lie Algebras, chapter IX, Appendix 2. Representations of real, complex or quaternionic type

Assume that ρ\rho is irreducible.

a) The representation ρ\rho is of real type if and only if there exists a non-zero symmetric bilinear form B on W, invariant under G. In this case the form B is separating; the set V of w∈Ww\in W such that H(w,x)=B(w,x)H(w, x) = B(w, x) for all x∈Wx\in W is an R\mathbf{R}-structure on W invariant under G.

b) The representation ρ\rho is of complex type if and only if there exists no non-zero bilinear form on W invariant under G.

c) The representation ρ\rho is of quaternionic type if and only if there exists a non-zero alternating bilinear form on W, invariant under G; such a form is necessarily separating.

For θ∈\theta \in HomC(G)(W,W)_{\mathbf{C}^{(G)}}(W,W) and x,y∈Wx, y\in W, put Bθ(x,y)=H(θx,y)B_{\theta}(x, y) = H(\theta x, y). Then BθB_{\theta} is a bilinear form on W, invariant under G, and separating if θ\theta is non-zero. Denote by B(W)G\mathscr{B}(W)^G the space of bilinear forms on W invariant under G; the map θ→Bθ\theta \rightarrow B_{\theta} from HomC(G)(W,W)_{\mathbf{C}^{(G)}}(W,W) to B(W)G\mathscr{B}(W)^G is an isomorphism of C\mathbf{C}-vector spaces. This implies, in particular, assertion b).

Let θ\theta be a C(G)\mathbf{C}^{(G)}-isomorphism from W to W‾\overline{W} such that θ∘θ=αW\theta \circ \theta =\alpha_W, with α∈{−1,+1}\alpha \in \{-1,+1\} (Prop. 2); since B(W)G\mathscr{B}(W)^G is of dimension 1, there exists ε∈C\varepsilon \in \mathbf{C} such that

Bθ(y,x)=εBθ(x,y)B_{\theta}(y, x) =\varepsilon B_{\theta}(x, y) for all x,yx, y in WW.

Iterating, we obtain Bθ(y,x)=εBθ(x,y)=ε2Bθ(y,x)B_{\theta}(y, x) =\varepsilon B_{\theta}(x, y) =\varepsilon^2B_{\theta}(y, x), so ε2\varepsilon^2 = 1 and ε∈{−1,+1}\varepsilon \in \{-1,+1\}. Moreover, for xx in W,

H(θx,θx)=Bθ(x,θx)=εBθ(θx,x)=εH(θ∘θ(x),x)=εαH(x,x) H(\theta x, \theta x) = B_{\theta}(x, \theta x) =\varepsilon B_{\theta}(\theta x, x) =\varepsilon H(\theta \circ \theta (x), x) =\varepsilon \alpha H(x, x)

so εα>0\varepsilon \alpha >0 since H is positive, that is, ε=α\varepsilon =\alpha. Assertions a) and c) now follow from Prop. 2.

Denote by dgdg the Haar measure of total mass 1 on G.

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.