Bourbaki

Lemma 5

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

Algebra, chapter X, § 3. Résolutions

Consider a commutative diagram

P′→uPp′↓↓p0→fM→M′′→0 \begin{array}{ccc} P' & \xrightarrow{u} & P \\ p' \downarrow & & \downarrow p \\ 0 & \xrightarrow{f} & M \to M'' \to 0 \end{array}

where (P,p)(P, p) (resp. (P′,p′)(P', p')) is a left resolution of MM (resp. M′M'), and where the bottom horizontal line is an exact sequence. There exists a homologism p′′:Con⁡(u)→M′′p'' : \operatorname{Con}(u) \to M''.

In fact, the exact sequence (X, p. 37, prop. 7)

0→P→πCon⁡(u)→δP′(−1)→0 0 \to P \xrightarrow{\pi} \operatorname{Con}(u) \xrightarrow{\delta} P'(-1) \to 0

gives an exact homology sequence

→Hn(P)→Hn(Con⁡(u))→Hn−1(P′)→⋯ \to H_n(P) \to H_n(\operatorname{Con}(u)) \to H_{n-1}(P') \to \cdots
⋯→H1(Con⁡(u))→H0(P′)→∂^H0(P)→H0(Con⁡(u))→0. \cdots \to H_1(\operatorname{Con}(u)) \to H_0(P') \xrightarrow{\hat{\partial}} H_0(P) \to H_0(\operatorname{Con}(u)) \to 0 .

By X, p. 38, lemma 3 a), we have ∂=−H0(u)\partial = - H_0(u). Since Hn(P)=0=Hn(P′)H_n(P) = 0 = H_n(P') for n>0n > 0 and H0(u):H0(P′)→H0(P)H_0(u) : H_0(P') \to H_0(P) is identified with f:M′→Mf : M' \to M, we conclude that Hn(Con⁡(u))=0H_n(\operatorname{Con}(u)) = 0 for n>0n > 0 and that H0(Con⁡(u))H_0(\operatorname{Con}(u)) is isomorphic to M′′M'', whence the lemma.

Let us now prove the theorem.

a) Let M be an A-module of type C\mathscr{C}. For every bounded left resolution (P,p)(P, p) of M of type C0\mathscr{C}_0, the element χC0(P)\chi_{\mathscr{C}_0}(P) of K(C0)K(\mathscr{C}_0) depends only on M. In fact, let (P1,p1)(P_1, p_1) and (P2,p2)(P_2, p_2) be two resolutions of this type. Consider the resolution

(P1×P2,p1×p2) (P_1 \times P_2, p_1 \times p_2)
of the A-module M×MM \times M and the homomorphism Δ:x↦(x,x)\Delta : x \mapsto (x, x) of M into M×MM \times M. By lemma 4, there exists a bounded left resolution (Q,q)(Q, q) of M of type C0\mathscr{C}_0 and a commutative diagram
Q→uP1×P2q↓↓p1×p2M→ΔM×M; \begin{array}{ccc} Q & \xrightarrow{u} & P_1 \times P_2 \\ q \downarrow & & \downarrow p_1 \times p_2 \\ M & \xrightarrow{\Delta} & M \times M ; \end{array}
we deduce from this a commutative diagram
Q→u∘priPiq↓↓piM→1MM,i=1,2. \begin{array}{ccc} Q & \xrightarrow{u \circ pr_i} & P_i \\ q \downarrow & & \downarrow p_i \\ M & \xrightarrow{1_M} & M , \quad i = 1, 2 . \end{array}
By lemma 5, Con (u∘pri)(u \circ pr_i) has zero homology, hence u∘priu \circ pr_i is a homologism and χC0(Q)=χC0(Pi)\chi_{\mathscr{C}_0}(Q) = \chi_{\mathscr{C}_0}(P_i) (X, p. 41, prop. 10); it follows that χC0(P1)=χC0(P2)\chi_{\mathscr{C}_0}(P_1) = \chi_{\mathscr{C}_0}(P_2) as announced.

b) For every A-module M of type C\mathscr{C}, let φ(M)∈K(C0)\varphi(M) \in K(\mathscr{C}_0) be the common value of the χC0(P)\chi_{\mathscr{C}_0}(P) for all bounded left resolutions P of M of type C0\mathscr{C}_0. Let us show that the function φ:C→K(C0)\varphi : \mathscr{C} \to K(\mathscr{C}_0) is additive. Let therefore

0→M′→fM→M′′→0 0 \to M' \xrightarrow{f} M \to M'' \to 0
be an exact sequence of A-modules of type C\mathscr{C}. By Lemma 4, there exists a commutative diagram
P′→uPp′↓↓p0→fM→gM′′→0 \begin{array}{ccc} P' & \xrightarrow{u} & P \\ p' \downarrow & & \downarrow p \\ 0 & \xrightarrow{f} & M \xrightarrow{g} M'' \to 0 \end{array}
where (P,p)(P, p) and (P′,p′)(P', p') are bounded left resolutions of type C0\mathscr{C}_0. Then we have
φ(M)=χC0(P),φ(M′)=χC0(P′) \varphi(M) = \chi_{\mathscr{C}_0}(P) , \qquad \varphi(M') = \chi_{\mathscr{C}_0}(P')
and by Lemma 5
φ(M′′)=χC0(Con (u))=χC0(P)−χC0(P′)=φ(M)−φ(M′); \varphi(M'') = \chi_{\mathscr{C}_0}(\mathrm{Con}\,(u)) = \chi_{\mathscr{C}_0}(P) - \chi_{\mathscr{C}_0}(P') = \varphi(M) - \varphi(M') ;
which was to be proved.

c) Let then β:K(C)→K(C0)\beta : K(\mathscr{C}) \to K(\mathscr{C}_0) be the homomorphism such that, with the preceding notations, we have β([M]C)=χC0(P)\beta([M]_{\mathscr{C}})=\chi_{\mathscr{C}_0}(P). Since pp is a homomorphism, we have χC(P)=[M]C\chi_{\mathscr{C}}(P)=[M]_{\mathscr{C}}, hence α∘β([M]C)=α(χC0(P))=χC(P)=[M]C\alpha\circ\beta([M]_{\mathscr{C}})=\alpha(\chi_{\mathscr{C}_0}(P))=\chi_{\mathscr{C}}(P)=[M]_{\mathscr{C}} and α∘β=1K(C)\alpha\circ\beta=1_{K(\mathscr{C})}. If MM is of type C0\mathscr{C}_0, then (M,1M)(M,1_M) is a resolution of MM, hence φ(M)=[M]C0\varphi(M)=[M]_{\mathscr{C}_0} and β∘α=1K(C0)\beta\circ\alpha=1_{K(\mathscr{C}_0)}, which completes the proof.

We shall apply this theorem to modules of "finite projective dimension" in § 8 (X, p. 137).

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.