Bourbaki

Corollary

Tag 05R4, permanent. Cite it as [Bourbaki, Tag 05R4].

Algebra, chapter II, § 6. Inverse and direct limits of modules

Let (Pα)(P_\alpha) be a direct system of subsets of EαE_\alpha and let P=lim⁡Pα→P = \lim \overrightarrow{P_\alpha}. If, for all α∈I\alpha \in I, PαP_\alpha is a free subset (resp. basis) of EαE_\alpha, then PP is a free subset (resp. basis) of EE.

The second assertion follows immediately from the first and Proposition 4. It is therefore sufficient to prove that if the PαP_\alpha are free every subset {y(i)}1≤i≤n\{ y^{(i)} \}_{1 \leq i \leq n} consisting of distinct elements of PP, is free. There exists an α∈I\alpha \in I and elements xα(i)∈Pαx_\alpha^{(i)} \in P_\alpha such that y(i)=fα(xα(i))y^{(i)} = f_\alpha(x_\alpha^{(i)}) for 1≤i≤n1 \leq i \leq n (Set Theory, III, § 7, no. 5, Lemma 1); if ∑iλ(i)y(i)=0\sum_i \lambda^{(i)} y^{(i)} = 0, it may be assumed that λ(i)=ϕα(λα(i))\lambda^{(i)} = \phi_\alpha(\lambda_\alpha^{(i)}) for 1≤i≤n1 \leq i \leq n and hence fα(∑iλβ(i)xβ(i))=0f_\alpha \left( \sum_i \lambda_\beta^{(i)} x_\beta^{(i)} \right) = 0; this implies ∑iλβ(i)xβ(i)=0\sum_i \lambda_\beta^{(i)} x_\beta^{(i)} = 0 for some β≥α\beta \geq \alpha, where λβ(i)=ϕβα(λα(i))\lambda_\beta^{(i)} = \phi_{\beta \alpha}(\lambda_\alpha^{(i)}), xβ(i)=fβα(xα(i))x_\beta^{(i)} = f_{\beta \alpha}(x_\alpha^{(i)}) and the xβ(i)x_\beta^{(i)} belong to PβP_\beta and are distinct since y(i)=fβ(xβ(i))y^{(i)} = f_\beta(x_\beta^{(i)}); then λβ(i)=0\lambda_\beta^{(i)} = 0 for 1≤i≤n1 \leq i \leq n, whence

λ(i)=ϕβ(λβ(i))=0 \lambda^{(i)} = \phi_\beta(\lambda_\beta^{(i)}) = 0
for 1≤i≤n1 \leq i \leq n.

Suppose now that all the rings AαA_\alpha are equal to the same ring AA and the ϕβα\phi_{\beta \alpha} to 1A1_A; then, for every direct system (Eα,fβα)(E_\alpha, f_{\beta \alpha}) of AA-modules, E=lim⁡Eα→E = \lim \overrightarrow{E_\alpha} is an AA-module. Let FF be an AA-module and for all α\alpha let uα:Eα→Fu_\alpha : E_\alpha \to F be an AA-linear mapping such that (uα)(u_\alpha) is a direct system of mappings; then u=lim⁡uα→u = \lim \overrightarrow{u_\alpha} is an AA-linear mapping of EE into FF. Conversely, for every AA-linear mapping v:lim⁡Eα→→Fv : \lim \overrightarrow{E_\alpha} \to F, the family of vα=v∘fαv_\alpha = v \circ f_\alpha is a direct system of AA-linear mappings such that v=lim⁡vα→v = \lim \overrightarrow{v_\alpha}. On the other hand we note that for α≤β\alpha \leq \beta the mapping

Hom⁡(fβα,1F)=fˉαβ:Hom⁡A(Eβ,F)→Hom⁡A(Eα,F) \operatorname{Hom}(f_{\beta \alpha}, 1_F) = \bar{f}_{\alpha \beta} : \operatorname{Hom}_A(E_\beta, F) \to \operatorname{Hom}_A(E_\alpha, F)
is a Z\mathbf{Z}-module homomorphism such that (Hom⁡A(Eα,F),fˉαβ)(\operatorname{Hom}_A(E_\alpha, F), \bar{f}_{\alpha \beta}) is an inverse system of Z\mathbf{Z}*-modules*; as fˉαβ(vβ)=vβ∘fβα\bar{f}_{\alpha \beta}(v_\beta) = v_\beta \circ f_{\beta \alpha}, the above remarks can be expressed as follows:

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.