Bourbaki

Proposition 42

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

Lie Groups and Lie Algebras, chapter III, § 3. Passage from a Lie group to its Lie algebra

Let GG be a Lie group, XX an analytic manifold, (g,x)↦gx(g, x) \mapsto gx (resp. xgxg) a law of analytic left (resp. right) operation of GG on XX and x0x_0 a point of XX which is invariant under GG. For all g∈Gg \in G, let τ(g)\tau(g) be the automorphism x↦gxx \mapsto gx (resp. xgxg) of XX and let π(g)\pi(g) be the automorphism of Tx0(X)T_{x_0}(X) tangent at x0x_0 to τ(g)\tau(g). (i) π\pi is an analytic representation of GG (resp. G∨G^\vee) on Tx0(X)T_{x_0}(X). (ii) For all a∈L(G)a \in L(G) and all ξ0∈Tx0(X)\xi_0 \in T_{x_0}(X), L(π)a.ξ0L(\pi)a.\xi_0 can be calculated as follows: let DaD_a be the vector field defined by aa on XX and ξ\xi a vector field of class C1C^1 in an open neighbourhood of x0x_0 such that ξ(x0)=ξ0\xi(x_0) = \xi_0; then

L(π)a.ξ0=−[Da,ξ](x0). L(\pi)a.\xi_0 = -[D_a, \xi](x_0).

τ(gg′)=τ(g)τ(g′)\tau(gg') = \tau(g)\tau(g') (resp. τ(g′)τ(g)\tau(g')\tau(g)) and hence π(gg′)=π(g)π(g′)\pi(gg') = \pi(g)\pi(g') (resp. π(g′)π(g)\pi(g')\pi(g)). On the other hand, since TXTX is a vector G-bundle of class CωC^\omega (§ 1, no. 8, Proposition 16), π\pi is analytic, whence (i).

To prove (ii), suppose that G operates on the left. There exists an open neighbourhood I of 0 in K and an analytic mapping γ\gamma of I into G such that γ(0)=e,T0(γ)1=a\gamma(0) = e, T_0(\gamma)1 = a. Then DaD_a is the vector field on X defined by the mapping ϕ:(λ,x)↦γ(λ)x\phi : (\lambda, x) \mapsto \gamma(\lambda)x of I×XI \times X into X (§ 2, no. 2). If ϕλ\phi_\lambda denotes the bijection x↦γ(λ)xx \mapsto \gamma(\lambda)x of X into X, then

[Da,ξ](x0)=(ddλ(Tϕλ(x0)(ϕλ−1)ξ(ϕλ(x0))))λ=0(Diff. & Anal. Man., R, [D_a, \xi](x_0) = \left( \frac{d}{d\lambda} (T_{\phi_\lambda(x_0)}(\phi_\lambda^{-1})\xi(\phi_\lambda(x_0))) \right)_{\lambda=0} \quad (\text{Diff. \& Anal. Man., R,}
=(ddλ(Tx0(ϕλ−1)ξ0))λ=0 = \left( \frac{d}{d\lambda} (T_{x_0}(\phi_\lambda^{-1})\xi_0) \right)_{\lambda=0}
=(ddλ(π(γ(λ))−1ξ0))λ=0. = \left( \frac{d}{d\lambda} (\pi(\gamma(\lambda))^{-1}\xi_0) \right)_{\lambda=0}.

As the mappings λ↦γ(λ)−1\lambda \mapsto \gamma(\lambda)^{-1} and λ↦γ(−λ)\lambda \mapsto \gamma(-\lambda) are tangent at 0, this is also equal to

−(ddλ(π(γ(λ))ξ0))λ=0 - \left( \frac{d}{d\lambda} (\pi(\gamma(\lambda))\xi_0) \right)_{\lambda=0}
=−(ddλ(π∘γ)(λ))λ=0ξ0 = - \left( \frac{d}{d\lambda} (\pi \circ \gamma)(\lambda) \right)_{\lambda=0} \xi_0
=−L(π)a.ξ0. = -L(\pi)a.\xi_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.