Bourbaki

Proposition 9

Tag 0A38, permanent. Cite it as [Bourbaki, Tag 0A38].

Lie Groups and Lie Algebras, chapter II, § 2. Free Lie algebras

*Let S1S_1 and S2S_2 be two disjoint sets and dd a mapping of S1×S2S_1 \times S_2 into L(S2)L(S_2). Let gg be the quotient Lie algebra of L(S1∪S2)L(S_1 \cup S_2) by the ideal generated by the elements [s1,s2]−d(s1,s2)[s_1, s_2] - d(s_1, s_2) with s1∈S1,s2∈S2s_1 \in S_1, s_2 \in S_2; let ψ\psi be the canonical mapping of L(S1∪S2)L(S_1 \cup S_2) onto gg.

(a) *For i=1,2i = 1, 2, the restriction ϕi\phi_i of ψ\psi to SiS_i can be extended to an isomorphism of L(Si)L(S_i) onto a subalgebra aia_i of gg.

(b) g=a1+a2,a1∩a2={0}g = a_1 + a_2, a_1 \cap a_2 = \{0\} and a2a_2 is an ideal of gg.

For i=1,2i = 1, 2, let ψi\psi_i denote the homomorphism of L(Si)L(S_i) into gg which extends ϕi\phi_i and aia_i its image. Clearly ϕi(Si)\phi_i(S_i) generates aia_i.

Let s1∈S1s_1 \in S_1; we write D=ad⁡ϕ1(s1)D = \operatorname{ad} \phi_1(s_1). The derivation DD of gg maps ϕ2(S2)\phi_2(S_2) into a2a_2 by the relation

[ϕ1(s1),ϕ2(s2)]=ψ2(d(s1,s2))for s2∈S2; [\phi_1(s_1), \phi_2(s_2)] = \psi_2(d(s_1, s_2)) \quad \text{for } s_2 \in S_2;

as the subalgebra a2a_2 of gg is generated by ϕ2(S2)\phi_2(S_2), therefore D(a2)⊂a2D(a_2) \subset a_2. The set of x∈gx \in g such that ad⁡x\operatorname{ad} x leaves a2a_2 invariant is a Lie subalgebra of gg which contains ϕ1(S1)\phi_1(S_1) by the above and hence also a1a_1. Hence

(23)(23)
[a1,a2]⊂a2.[a_1, a_2] \subset a_2.

Therefore a1+a2a_1 + a_2 is a Lie subalgebra of gg and, as it contains the generating set ϕ1(S1)∪ϕ2(S2)\phi_1(S_1) \cup \phi_2(S_2),

(24)(24)
a1+a2=g.a_1 + a_2 = g.

For all s1∈S1s_1 \in S_1 there exists a derivation Ds1D_{s_1} of L(S2)L(S_2) such that

Ds1(s2)=d(s1,s2) D_{s_1}(s_2) = d(s_1, s_2)

for all s2s_2 in S2S_2 (no. 8, Corollary to Proposition 8). The mapping s1↦Ds1s_1 \mapsto D_{s_1} can be extended to a homomorphism DD of L(S1)L(S_1) into the Lie algebra of derivations of L(S2)L(S_2). Let h\mathfrak{h} be the semi-direct product of L(S1)L(S_1) by L(S2)L(S_2) corresponding to DD (Chapter I, § 1, no. 8). As a module h\mathfrak{h} is equal to L(S1)×L(S2)L(S_1) \times L(S_2) and in particular

[(s1,0),(0,s2)]=(0,d(s1,s2)) [(s_1, 0), (0, s_2)] = (0, d(s_1, s_2))

for s1∈S1s_1 \in S_1 and s2∈S2s_2 \in S_2.

From (25) we deduce the existence of a homomorphism ff of gg into h\mathfrak{h} such that f(ϕ1(s1))=(s1,0)f(\phi_1(s_1)) = (s_1, 0) and f(ϕ2(s2))=(0,s2)f(\phi_2(s_2)) = (0, s_2) for s1∈S1s_1 \in S_1 and s2∈S2s_2 \in S_2. We deduce immediately the relation

f(ψ1(a1)+ψ2(a2))=(a1,a2) f(\psi_1(a_1) + \psi_2(a_2)) = (a_1, a_2)

for a1∈L(S1)a_1 \in L(S_1) and a2∈L(S2)a_2 \in L(S_2).

Relation (26) shows that ψ1\psi_1 and ψ2\psi_2 are injective and that a1∩a2={0}a_1 \cap a_2 = \{0\}. Formulae (23) and (24) then imply the proposition.

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.