Bourbaki

Proposition 14

Tag 0DT1, permanent. Cite it as [Bourbaki, Tag 0DT1].

Algebra, chapter X, § 2. Complexes de A-modules

(i) There exists a unique mapping kk-linéaire ∇\nabla from the right ΩA/k\Omega_{A/k}-module M⊗AΩA/kM \otimes_A \Omega_{A/k} into itself, graded of degree 1, which extends ∇0\nabla^0 in degree 0 and satisfies the identity :

(14)(14)
∇(xω)=(∇x)ω+(−1)px(dω)forx∈M⊗AΩA/kp, ω∈ΩA/k. \nabla(x \omega) = (\nabla x) \omega + (-1)^p x(d\omega) \quad \text{for} \quad x \in M \otimes_A \Omega^p_{A/k},\ \omega \in \Omega_{A/k}.

(ii) The composite mapping ∇∘∇\nabla \circ \nabla is ΩA/k\Omega_{A/k}-linéaire ; in particular the mapping R=∇1∘∇0R = \nabla^1 \circ \nabla^0 from MM into M⊗AΩA/k2M \otimes_A \Omega^2_{A/k} is AA-linéaire, and we have

∇∘∇(m⊗ω)=R(m).ωform∈M, ω∈ΩA/k. \nabla \circ \nabla(m \otimes \omega) = R(m).\omega \quad \text{for} \quad m \in M,\ \omega \in \Omega_{A/k}.

The homomorphism RR is sometimes called curvature homomorphism of the connection ∇0\nabla^0; if it is zero, the pair (M⊗AΩA/k,∇)(M \otimes_A \Omega_{A/k}, \nabla) is a complex, also called de Rham complex of (M,∇0)(M, \nabla^0) over kk.

Let us prove (i). The uniqueness of ∇\nabla is obvious. Let us define a kk-homomorphisme ∇‾\overline{\nabla} from M⊗kΩA/kM \otimes_k \Omega_{A/k} into M⊗AΩA/kM \otimes_A \Omega_{A/k} by

∇‾(m⊗kω)=(∇0m)ω+m⊗dωform∈M, ω∈ΩA/k. \overline{\nabla}(m \otimes_k \omega) = (\nabla^0 m) \omega + m \otimes d\omega \quad \text{for} \quad m \in M,\ \omega \in \Omega_{A/k}.

It follows from (13) that ∇‾(am⊗ω)=∇‾(m⊗aω)\overline{\nabla}(am \otimes \omega) = \overline{\nabla}(m \otimes a \omega), so that one obtains by passing to the quotient a kk-homomorphisme ∇\nabla from M⊗AΩA/kM \otimes_A \Omega_{A/k} into itself, graded of degree 1, extending ∇0\nabla^0 in degree 0. Let us verify (14): we have for m∈Mm \in M, α∈ΩA/kp\alpha \in \Omega^p_{A/k}, ω∈ΩA/k\omega \in \Omega_{A/k} :

∇((m⊗α).ω)=∇(m⊗(α∧ω))=∇0(m).(α∧ω)+m⊗d(α∧ω)=∇0(m)α.ω+(m⊗dα)ω+(−1)p(m⊗α)dω=(∇(m⊗α))ω+(−1)p(m⊗α)dω \begin{align*} \nabla((m \otimes \alpha).\omega) &= \nabla(m \otimes (\alpha \wedge \omega)) = \nabla^0(m).(\alpha \wedge \omega) + m \otimes d(\alpha \wedge \omega) \\ &= \nabla^0(m) \alpha.\omega + (m \otimes d\alpha) \omega + (-1)^p (m \otimes \alpha) d\omega \\ &= (\nabla(m \otimes \alpha)) \omega + (-1)^p (m \otimes \alpha) d\omega \end{align*}
which proves (14) for x=m⊗αx = m \otimes \alpha; the general case follows by linearity.

Let us prove (ii). Let x∈M⊗AΩA/kp,ω∈ΩA/kx \in M \otimes_A \Omega_{A/k}^p, \omega \in \Omega_{A/k}; by repeated application of (14), we obtain :

∇∘∇(xω)=∇(∇(x)ω)+(−1)p∇(xdω)=(∇∘∇(x))ω+(−1)p+1∇(x)(dω)+(−1)p∇(x)(dω)=(∇∘∇(x))ω, \nabla \circ \nabla(x \omega) = \nabla(\nabla(x) \omega) + (-1)^p \nabla(x d \omega) = (\nabla \circ \nabla(x)) \omega + (-1)^{p+1} \nabla(x)(d \omega) + (-1)^p \nabla(x)(d \omega) = (\nabla \circ \nabla(x)) \omega,
which proves the first assertion of (ii); the others follow immediately.

Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.

Cited by

Nothing.

Cites

Nothing.

Translated with machine help, not yet reviewed by a person.