Bourbaki

Exercise 6, § 2

Commutative Algebra, I, 41

Let A be a ring and M a left A-module. An exact sequence

Ln→Ln−1→⋯→L1→L0→M→0 L_n \to L_{n-1} \to \cdots \to L_1 \to L_0 \to M \to 0
where LiL_i is a free left A-module (0≤i≤n)(0 \leq i \leq n), is called a presentation of M of length n or an n-presentation of M. The presentation is called finite if all the LiL_i are finitely generated free modules.

If M is a finitely generated left A-module, we denote by λ(M)\lambda(M) the least upper bound (finite or +∞+\infty) of the integers n≥0n \geq 0 such that M has a finite nn-presentation. If M is not finitely generated, we set λ(M)=−1\lambda(M) = -1.

(a) Let 0→P→N→M→00 \to P \to N \to M \to 0 be an exact sequence of left A-modules. Then λ(N)≥inf⁡(λ(P),λ(M))\lambda(N) \geq \inf(\lambda(P), \lambda(M)). (Starting with two n-presentations of P and M respectively, derive one for N using Exercise 5 of § 1.)

(b) Let M,→unMn−1→⋯→M0→u0M→0M, \xrightarrow{u_n} M_{n-1} \to \cdots \to M_0 \xrightarrow{u_0} M \to 0 be a finite n-presentation of M; show that, if λ(M)>n\lambda(M) > n, Ker⁡(un)\operatorname{Ker}(u_n) is a finitely generated A-module. (Let

Ln+1→vn+1Ln⟶⋯⟶L2→v2L1→v1L0→v0M⟶0 L_{n+1} \xrightarrow{v_{n+1}} L_n \longrightarrow \cdots \longrightarrow L_2 \xrightarrow{v_2} L_1 \xrightarrow{v_1} L_0 \xrightarrow{v_0} M \longrightarrow 0
be a finite (n+1)(n+1)-presentation of M and let P=Ker⁡(v0)P = \operatorname{Ker}(v_0) be such that there is an n-presentation of P:
Ln+1→vn+1Ln⟶⋯⟶L2→v2L1→v1P⟶0. L_{n+1} \xrightarrow{v_{n+1}} L_n \longrightarrow \cdots \longrightarrow L_2 \xrightarrow{v_2} L_1 \xrightarrow{v_1} P \longrightarrow 0.
Applying the method of (a) to the exact sequence 0→P→L0→M→00 \to P \to L_0 \to M \to 0 we obtain an exact sequence
Mn⊕Ln+1→wnMn−1⊕Ln→wn−1⋯→w1M0⊕L1→w0L0⟶0 M_n \oplus L_{n+1} \xrightarrow{w_n} M_{n-1} \oplus L_n \xrightarrow{w_{n-1}} \cdots \xrightarrow{w_1} M_0 \oplus L_1 \xrightarrow{w_0} L_0 \longrightarrow 0
and exact sequences 0→Ker⁡(vi+1)→Ker⁡(wi)→Ker⁡(ui)→00 \to \operatorname{Ker}(v_{i+1}) \to \operatorname{Ker}(w_i) \to \operatorname{Ker}(u_i) \to 0 (§1§ 1, no. 4, Proposition 2). Observe finally that Ker⁡(wi)\operatorname{Ker}(w_i) is a direct factor of Mi⊕Li+1M_i \oplus L_{i+1}.

(c) Show that with the hypotheses of (a)

λ(M)≥inf⁡(λ(N),λ(P)+1). \lambda(M) \geq \inf(\lambda(N), \lambda(P) + 1).
(If n≤inf⁡(λ(N),λ(P)+1)n \leq \inf(\lambda(N), \lambda(P) + 1), show by induction on nn that λ(M)≥n\lambda(M) \geq n, arguing as in (a) and using (b).)

(d) Show that with the hypotheses of (a)

λ(P)≥inf⁡(λ(N),λ(M)−1). \lambda(P) \geq \inf(\lambda(N), \lambda(M) - 1).
(Same method as in (c).) Deduce that, if λ(N)=+∞\lambda(N) = +\infty, then λ(M)=λ(P)+1\lambda(M) = \lambda(P) + 1.

(e) Deduce from (a), (c) and (d) that, if N=M⊕PN = M \oplus P, then

λ(N)=inf⁡(λ(M),λ(P)). \lambda(N) = \inf(\lambda(M), \lambda(P)).
In particular, for N to admit a finite presentation it is necessary and sufficient that M and P do so too.

(f) Let N1,N2N_1, N_2 be two submodules of an A-module M. Suppose that N1N_1 and N2N_2 admit finite presentations. For N1+N2N_1 + N_2 to admit a finite presentation it is necessary and sufficient that N1∩N2N_1 \cap N_2 be finitely generated.

Solution

No solution has been written yet.

Transcribed from the printed volume. Not translated.