Bourbaki

Theorem 1

Tag 0AI2, permanent. Cite it as [Bourbaki, Tag 0AI2].

Lie Groups and Lie Algebras, chapter III, § 8. Lie groups over R or Qp

Let G and H be two Lie group germs over R\mathbf{R} or Qp\mathbf{Q}_p. Let f be a continuous morphism of G into H. Then f is analytic.

We give L(G)L(G) and L(H)L(H) norms which define their topologies and such that ∥[x,y]∥≤∥x∥∥y∥\| [x, y] \| \leq \|x\| \|y\| for all x,yx, y. There exist an open ball V of centre 0 in L(G)L(G) and an exponential mapping ϕ\phi of G defined on V such that: (1) ϕ(V)\phi(V) is an open neighbourhood of ee in G; (2) ϕ\phi is an isomorphism of the analytic manifold V onto the analytic manifold ϕ(V)\phi(V); (3) ϕ(nx)=ϕ(x)n\phi(nx) = \phi(x)^n for all x∈Vx \in V and all n∈Zn \in \mathbf{Z} such that nx∈Vnx \in V. We define similarly W and ψ\psi for H. By shrinking V if necessary, it can be assumed that f(ϕ(V))⊂ψ(W)f(\phi(V)) \subset \psi(W). Then g=ψ−1∘f∘ϕg = \psi^{-1} \circ f \circ \phi is a continuous mapping of V into W.

We show that

(x∈V,λ∈Q and λx∈V)⇒g(λx)=λg(x). (x \in V, \lambda \in \mathbf{Q} \text{ and } \lambda x \in V) \Rightarrow g(\lambda x) = \lambda g(x).
It can be assumed that λ≠0\lambda \neq 0. Let λ=rq\lambda = \frac{r}{q} with r,qr, q in Z−{0}\mathbf{Z} - \{0\}. Let y=rqxy = \frac{r}{q} x.

If K=RK = \mathbf{R}, we write z=xq=yr∈Vz = \frac{x}{q} = \frac{y}{r} \in V. Then x=qz,y=rzx = qz, y = rz, whence

g(x)=ψ−1(f(ϕ(qz)))=ψ−1(f(ϕ(z)q))=ψ−1(f(ϕ(z))q). g(x) = \psi^{-1}(f(\phi(qz))) = \psi^{-1}(f(\phi(z)^q)) = \psi^{-1}(f(\phi(z))^q).
We show that ψ−1(f(ϕ(z))q)=qψ−1(f(ϕ(z)))=qg(z)\psi^{-1}(f(\phi(z))^q) = q \psi^{-1}(f(\phi(z))) = qg(z). It suffices to verify that, if u∈ψ(W)u \in \psi(W) is such that uq∈ψ(W)u^q \in \psi(W), then ψ−1(uq)=qψ−1(u)\psi^{-1}(u^q) = q \psi^{-1}(u); but if u=ψ(v)u = \psi(v), uq=ψ(vq)u^q = \psi(v^q), then v1/q∈Wv^1/q \in W and (ψ(v1/q))q=uq(\psi(v^1/q))^q = u^q, hence ψ(v1/q)=u=ψ(v)\psi(v^1/q) = u = \psi(v) and therefore v1=qvv^1 = qv.

Similarly, g(y)=rg(z)g(y) = rg(z), whence (1).

If K=QpK = \mathbf{Q}_p, we write z=rx=qy∈Vz = rx = qy \in V, whence g(z)=rg(x)=qg(y)g(z) = rg(x) = qg(y), whence again (1).

As Q\mathbf{Q} is dense in KK, (1) implies that

(x∈V,λ∈K and λx∈V)⇒g(λx)=λg(x). (x \in V, \lambda \in K \text{ and } \lambda x \in V) \Rightarrow g(\lambda x) = \lambda g(x).
Let x∈L(G)x \in L(G) and λ,λ′\lambda, \lambda' be elements of K∗K^* such that λx∈V,λ′x∈V\lambda x \in V, \lambda' x \in V. Then
g(λ′x)=g(λ′λλx)=λ′λg(λx) g(\lambda' x) = g\left( \frac{\lambda'}{\lambda} \lambda x \right) = \frac{\lambda'}{\lambda} g(\lambda x)
by (2) and hence 1λg(λx)=1λ′g(λ′x)\frac{1}{\lambda} g(\lambda x) = \frac{1}{\lambda'} g(\lambda' x). Thus an extension hh of gg to L(G)L(G) is defined by writing h(x)=1λg(λx)h(x) = \frac{1}{\lambda} g(\lambda x) for all λ\lambda such that λx∈V\lambda x \in V. Clearly hh is continuous. We show that
(x∈L(G) and λ∈K)⇒h(λx)=λh(x). (x \in L(G) \text{ and } \lambda \in K) \Rightarrow h(\lambda x) = \lambda h(x).
Let λ′∈K∗\lambda' \in K^* be such that λ′x∈V\lambda' x \in V and λ′λx∈V\lambda' \lambda x \in V. Then
h(λx)=1λ′g(λ′λx)=1λ′λg(λ′x)=λ1λ′g(λ′x)=λh(x). h(\lambda x) = \frac{1}{\lambda'} g(\lambda' \lambda x) = \frac{1}{\lambda'} \lambda g(\lambda' x) = \lambda \frac{1}{\lambda'} g(\lambda' x) = \lambda h(x).
Let x,yx, y be in L(G)L(G). Then, by Proposition 4 of § 4, no. 3,
h(x)+h(y)=lim⁡λ∈K∗,λ→0λ−1ψ−1(ψ(λh(x))ψ(λh(y)))=lim⁡λ∈K∗,λ→0λ−1ψ−1(ψ(h(λx))ψ(h(λy))). \begin{align*} h(x) + h(y) &= \lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} \psi^{-1}(\psi(\lambda h(x)) \psi(\lambda h(y))) \\ &= \lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} \psi^{-1}(\psi(h(\lambda x)) \psi(h(\lambda y))). \end{align*}
For ∣λ∣|\lambda| sufficiently small, λx∈V\lambda x \in V and λy∈V\lambda y \in V and hence the above expression is equal to
lim⁡λ∈K∗,λ→0λ−1ψ−1(f(ϕ(λx))f(ϕ(λy)))=lim⁡λ∈K∗,λ→0λ−1(ψ−1∘f)(ϕ(λx)ϕ(λy))=lim⁡λ∈K∗,λ→0λ−1g(ϕ−1(ϕ(λx)ϕ(λy)))=lim⁡λ∈K∗,λ→0h(λ−1ϕ−1(ϕ(λx)ϕ(λy)))=h(lim⁡λ∈K∗,λ→0λ−1ϕ−1(ϕ(λx)ϕ(λy)))=h(x+y). \begin{align*} &\lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} \psi^{-1}(f(\phi(\lambda x)) f(\phi(\lambda y))) \\ &= \lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} (\psi^{-1} \circ f)(\phi(\lambda x) \phi(\lambda y)) \\ &= \lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} g(\phi^{-1}(\phi(\lambda x) \phi(\lambda y))) \\ &= \lim_{\lambda \in K^*, \lambda \to 0} h(\lambda^{-1} \phi^{-1}(\phi(\lambda x) \phi(\lambda y))) \\ &= h(\lim_{\lambda \in K^*, \lambda \to 0} \lambda^{-1} \phi^{-1}(\phi(\lambda x) \phi(\lambda y))) \\ &= h(x + y). \end{align*}
Thus hh is continuous and linear, hence g=h∣Vg = h|V is analytic, hence ff is analytic on ϕ(V)\phi(V) and hence ff is analytic (§ 1, no. 10).

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.