Bourbaki

Theorem 2

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

Lie Groups and Lie Algebras, chapter IX, § 6. Integration on compact Lie groups

The canonical injection ι:Ω(X)G→Ω(X)\iota :\Omega (X)^G\rightarrow \Omega (X) is a homotopism of complexes (Algèbre, Chap. X, p. 33, déf. 5); the map ω→ω♯=∫Gτg∗ωdg\omega \rightarrow \omega^{\sharp}=\int_G\tau_g^*\omega dg is a homotopism, inverse to it up to homotopy. In particular, the map H(ι):H(Ω(X)G)→H(Ω(X))H(\iota ) : H(\Omega (X)^G)\rightarrow H(\Omega (X)) is bijective.

By Cor. 3 of no. 4, the map ω→ω♯\omega \rightarrow \omega^{\sharp} is a morphism of complexes from Ω(X)\Omega (X) to Ω(X)G\Omega (X)^G that induces the identity on the subcomplex Ω(X)G\Omega (X)^G; thus, to prove the theorem it suffices to construct a homomorphism s:Ω(X)→Ωs:\Omega (X)\rightarrow \Omega(X), graded of degree −1-1, such that

ω♯−ω=(d∘s+s∘d)(ω)\omega^{\sharp}-\omega = (d\circ s+s\circ d)(\omega ) for all ω∈Ω(X)\omega \in \Omega (X). (17)

By Lemma 1 of Integration, Chapter IX, §2, no. 4 and Remark 1 of §2, no. 2, there exists a positive measure dξd\xi on g\mathfrak{g} of compact support whose image under the exponential map is equal to dgdg. For ω∈Ω\omega \in \Omega(X), put

s(ω)=∫01{∫gτexptξ∗(i(ξ)ω).dξ}dt s(\omega ) =\int_0^1\{\int_{\mathfrak{g}}\tau_{expt\xi}^*(i(\xi )\omega ).d\xi \}dt

we have to show that formula (17) is satisfied. As in the proof of Cor. 1 (no. 4), we verify the formula

ds(ω)=∫01{∫gτexptξ∗d(i(ξ)ω).dξ}dt ds(\omega ) =\int_0^1\{\int_{\mathfrak{g}}\tau_{expt\xi}^*d(i(\xi )\omega ).d\xi \}dt

We now deduce from formulas (15) and (16) the equalities

ds(ω)+s(dω)=∫01{∫gτexptξ∗(d(i(ξ)ω)+i(ξ)dω).dξ}dt ds(\omega ) +s(d\omega ) =\int_0^1\{\int_{\mathfrak{g}}\tau_{expt\xi}^*(d(i(\xi )\omega ) +i(\xi )d\omega ).d\xi \}dt
=∫01{∫gτexptξ∗(θ(ξ)ω).dξ}dt =\int_0^1\{\int_{\mathfrak{g}}\tau_{expt\xi}^*(\theta (\xi )\omega ).d\xi \}dt
=∫g{∫01ddt(τexptξ∗ω)dt}dξ =\int_{\mathfrak{g}}\{\int_0^1\frac{d}{dt}(\tau_{expt\xi}^*\omega )dt\}d\xi
=∫g(τexpξ∗ω−ω)dξ =\int_{\mathfrak{g}}(\tau_{exp\xi}^*\omega -\omega )d\xi
=ω♯−ω =\omega^{\sharp}-\omega

hence Th. 2.

We apply the theorem in the case X = G, for the action of G by left translations. Recall (Chap. III, §3, no. 13, Prop. 50) that associating to a differential form on G its value at the identity element gives an isomorphism from Ω(G)G\Omega (G)^G to the graded algebra Alt(g\mathfrak{g}) of alternating multilinear forms on g\mathfrak{g}. Identify Ω(G)G\Omega (G)^G with Alt(g\mathfrak{g}) by means of this isomorphism. The operator dd is then given by the formula (Chap. III, §3, no. 14, Prop. 51)

dω(a1,...,ap+1) d\omega (a_1, . . . , a_{p+1})
=∑i<j(−1)i+jω([ai,aj],a1,...,ai−1,ai+1,...,aj−1,aj+1,...,ap+1) =\sum_{i<j}(-1)^{i+j}\omega ([a_i, a_j], a_1, . . . , a_{i-1}, a_{i+1}, . . . , a_{j-1}, a_{j+1}, . . . , a_{p+1})

for ω\omega in Altp(g)^p(\mathfrak{g}) and a1,...,ap+1a_1, . . . , a_{p+1} in g\mathfrak{g}.

For ξ∈g\xi \in \mathfrak{g}, let LξL_{\xi} be the corresponding left-invariant vector field (defined by means of the action of G on itself by right translations, cf. Chap. III, §3, no. 6). The operators θ(Lξ),i(Lξ)\theta (L_{\xi}), i(L_{\xi}) commute with the action of G on Ω(G)\Omega (G) defined by left translation, and hence induce operators θ(ξ),i(ξ)\theta (\xi ), i(\xi ) on Ω(G)G\Omega (G)^G; with the preceding identifications, these are expressed by the formulas (Differentiable and Analytic Manifolds, Results, 8.3.2 and 8.4.2)

(θ(ξ)ω)(a1,...,ap)=−∑iω(a1,...,ai−1,[ξ,ai],ai+1,...,ap) (\theta (\xi )\omega )(a_1, . . . , a_p) =-\sum_i\omega (a_1, . . . , a_{i-1},[\xi , a_i], a_{i+1}, . . . , a_p)
(i(ξ)ω)(a1,...,ap−1)=ω(ξ,a1,...,ap−1) (i(\xi )\omega )(a_1, . . . , a_{p-1}) =\omega (\xi , a_1, . . . , a_{p-1})

for ω\omega in Altp(g)^p(\mathfrak{g}) and a1,...,apa_1, . . . , a_p in g\mathfrak{g}.

The subcomplex GΩ(G)G^G\Omega (G)^G of biinvariant forms (Chap. III, §3, no. 13) can be identified with the subcomplex Alt(g\mathfrak{g})G^G of alternating multilinear forms on g\mathfrak{g} invariant under the adjoint representation (that is, such that θ(ξ)ω=0\theta (\xi )\omega = 0 for all ξ∈g\xi \in \mathfrak{g}). Thus, we have a commutative diagram of complexes

GΩ(G)G⟶Ω(G)G⟶Ω(G) ^G\Omega (G)^G\longrightarrow \Omega (G)^G\longrightarrow \Omega (G)

(18)

Alt(g\mathfrak{g})G⟶^G\longrightarrow Alt(g\mathfrak{g})

where the horizontal arrows are the canonical injections, and the vertical arrows are the isomorphisms induced by the map ω→ω(e)\omega \rightarrow \omega (e).

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

  • Algèbre, Chap. X, p. 33 not in this corpus: A X
  • Lemma 1 of Integration, Chapter IX, §2, no. 4 not in this corpus: INT IX
  • Remark 1 of §2, no. 2 not in this corpus: INT IX
  • Chap. III, §3, no. 13
  • Chap. III, §3, no. 6
  • Chap. III, §3, no. 13, Prop. 50 not in this corpus
  • Chap. III, §3, no. 14, Prop. 51 not in this corpus
  • Cor. 1
  • Cor. 3
Transcribed from the printed volume. Not translated.