Bourbaki

Corollary

Tag 063K, permanent. Cite it as [Bourbaki, Tag 063K].

Algebra, chapter III, § 7. Exterior algebras

Let M and N be two projective A-modules and f:M→Nf : M \to N an A-linear mapping. If f is injective, then Λ(f):Λ(M)→Λ(N)\Lambda(f) : \Lambda(M) \to \Lambda(N) is injective.

We prove this first under the assumption that M is free; let (eλ)λ∈L(e_\lambda)_{\lambda \in L} be a basis of M, so that (eJ)J∈B(L)(e_J)_{J \in \mathfrak{B}(L)} (formula (18)) is a basis of Λ(M)\Lambda(M). Suppose that the kernel of Λ(f)\Lambda(f) contains an element u=∑JαJeJ≠0u = \sum_J \alpha_J e_J \neq 0. Let K be an element minimal among the finite subsets J such that αJ≠0\alpha_J \neq 0 and let H be a finite subset of L disjoint from K and such that K∪HK \cup H contains all the J (finite in number) such that αJ≠0\alpha_J \neq 0; for all J≠KJ \neq K such that αJ≠0\alpha_J \neq 0, we have therefore by definition J∩H≠0J \cap H \neq 0 and consequently (no. 8, formula (20)).

u∧eH=+αKeK∪H u \wedge e_H = +\alpha_K e_{K \cup H}
belongs to the two-sided ideal of Λ(M)\Lambda(M), the kernel of Λ(f)\Lambda(f). We write eK∪H=eλ1∧eλ2∧⋯∧eλne_{K \cup H} = e_{\lambda_1} \wedge e_{\lambda_2} \wedge \cdots \wedge e_{\lambda_n}; then αKf(eλ1)∧f(eλ2)∧⋯∧f(eλn)=0\alpha_K f(e_{\lambda_1}) \wedge f(e_{\lambda_2}) \wedge \cdots \wedge f(e_{\lambda_n}) = 0; by virtue of Proposition 12, the elements f(eλi)f(e_{\lambda_i}) (1≤i≤n1 \leq i \leq n) of N are linearly dependent. But this contradicts the hypothesis that ff is injective (II, § 1, no. 11, Corollary 3 to Proposition 17).

We now consider the general case; let M′M' be an A-module such that M⊕M′=PM \oplus M' = P is free (II, § 2, no. 2, Proposition 4). Consider the linear mapping g:M⊕M′→N⊕M⊕M′g : M \oplus M' \to N \oplus M \oplus M' such that g(x,y)=(f(x),0,y)g(x, y) = (f(x), 0, y), so that there is a commutative diagram

M→fNj↓j′↓P→gN⊕P \begin{array}{ccc} M & \xrightarrow{f} & N \\ j \downarrow & & j' \downarrow \\ P & \xrightarrow{g} & N \oplus P \end{array}

where the vertical arrows are the canonical injections. Since gg is injective and PP is free, Λ(g)\Lambda(g) is an injective homomorphism as has been seen above. Now, Λ(j):Λ(M)→Λ(P)\Lambda(j): \Lambda(M) \to \Lambda(P) is an injective homomorphism since MM is a direct factor in PP (no. 2). The composite homomorphism

Λ(M)→Λ(j)Λ(P)→Λ(g)Λ(N⊕P) \Lambda(M) \xrightarrow{\Lambda(j)} \Lambda(P) \xrightarrow{\Lambda(g)} \Lambda(N \oplus P)

is therefore injective and, as it is also equal to the composite homomorphism

Λ(M)→Λ(f)Λ(N)→Λ(g′)Λ(N⊕P) \Lambda(M) \xrightarrow{\Lambda(f)} \Lambda(N) \xrightarrow{\Lambda(g')} \Lambda(N \oplus P)

we conclude that Λ(f)\Lambda(f) is injective.

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 edition, 1998, Springer. Not translated.