Bourbaki

§ 9. Koszul complexes

In this paragraph, all the rings considered are commutative.

1. The complexes K(u)\mathbf{K}(u), K.(u,C)\mathbf{K}.(u, C), K∗(u,C)\mathbf{K}^*(u, C)

Let AA be a ring, LL an AA-module, u:L→Au : L \to A a linear form, and Λ(L)\Lambda(L) the exterior algebra of the AA-module LL. For x∈Λ(L)x \in \Lambda(L), denote by du(x)d_u(x) the interior product x⋅ux \cdot u (III, p. 161, example). According to loc. cit., p. 162, formula (60), one has

(1)du(e1∧…∧en)=∑i=1n(−1)i+1u(ei)  e1∧…∧ei−1∧ei+1∧…∧en (1)\quad d_u(e_1 \wedge \ldots \wedge e_n) = \sum_{i=1}^n (-1)^{i+1} u(e_i) \; e_1 \wedge \ldots \wedge e_{i-1} \wedge e_{i+1} \wedge \ldots \wedge e_n

for e1,…,ene_1, \ldots, e_n in LL. According to III, p. 164 and 165, the mapping du:Λ(L)→Λ(L)d_u : \Lambda(L) \to \Lambda(L) is an antidifferentiation of degree (−1)(-1) and of square zero. It is the unique antidifferentiation of the AA-algebra Λ(L)\Lambda(L) which extends u:Λ1(L)→Λ0(L)u : \Lambda^1(L) \to \Lambda^0(L).

Definition 1

The complex (Λ(L),du)(\Lambda(L), d_u) is denoted by KA(u)\mathbf{K}^A(u) or K(u)\mathbf{K}(u).

One should note that Kn(u)=Λn(L)=K−n(u)\mathbf{K}_n(u) = \Lambda^n(L) = \mathbf{K}^{-n}(u). It is clear that K(u)\mathbf{K}(u) is zero on the right and that H0(K(u))=Coker (u)=A/qH_0(\mathbf{K}(u)) = \mathrm{Coker}\,(u) = A/q where qq is the ideal u(L)u(L) of AA.

For every complex of AA-modules CC, put

KA(u,C)=C⊗AKA(u),HA(u,C)=H(C⊗AKA(u))HrA(u,C)=Hr(C⊗AKA(u)), \begin{aligned} \mathbf{K}^A(u, C) &= C \otimes_A \mathbf{K}^A(u), \\ H^A(u, C) &= H(C \otimes_A \mathbf{K}^A(u)) \\ H_r^A(u, C) &= H_r(C \otimes_A \mathbf{K}^A(u)), \end{aligned}
KA∗(u,C)=HomgrA(KA(u),C),HA∗(u,C)=H(HomgrA(KA(u),C)),HAr(u,C)=Hr(HomgrA(KA(u),C)). \begin{aligned} \mathbf{K}^*_A(u, C) &= \mathrm{Homgr}_A(\mathbf{K}^A(u), C), \\ H^*_A(u, C) &= H(\mathrm{Homgr}_A(\mathbf{K}^A(u), C)), \\ H^r_A(u, C) &= H^r(\mathrm{Homgr}_A(\mathbf{K}^A(u), C)). \end{aligned}

There are therefore canonical homomorphisms of AA-modules (X, p. 62 and p. 82)

γ0:H0(C)⊗AA/q→H0A(u,C),λ0:HA0(u,C)→HomA(A/q,H0(C)). \begin{aligned} \gamma_0 : H_0(C) \otimes_A A/q &\to H^A_0(u, C), \\ \lambda^0 : H^0_A(u, C) &\to \mathrm{Hom}_A(A/q, H^0(C)). \end{aligned}

Lemma 1

If the complex CC is zero on the right (resp. on the left), then KA(u,C)\mathbf{K}^A(u, C) (resp. KA∗(u,c)\mathbf{K}^*_A(u, c)) is zero on the right (resp. on the left), and γ0\gamma_0 (resp. λ0\lambda^0) is bijective.

This follows from X, p. 62, prop. 1 and p. 82, prop. 1.

Proposition 1

Let x∈Lx \in L; denote by Rx:y↦x∧yR_x : y \mapsto x \wedge y the left multiplication by xx in the algebra Λ(L)\Lambda(L). Then du∘Rx+Rx∘du=u(x)⋅1Λ(L)=u(x)Λ(L)d_u \circ R_x + R_x \circ d_u = u(x) \cdot 1_{\Lambda(L)} = u(x)_{\Lambda(L)}.

Indeed (du∘Rx+Rx∘du)(y)=du(x∧y)+x∧du(y)(d_u \circ R_x + R_x \circ d_u)(y) = d_u(x \wedge y) + x \wedge d_u(y); since dud_u is an antidifferentiation, du(x∧y)+x∧du(y)=du(x)∧y=u(x)⋅yd_u(x \wedge y) + x \wedge d_u(y) = d_u(x) \wedge y = u(x) \cdot y.

Corollary 1

If uu is surjective, K(u)K(u) is homotopic to zero (X, p. 34) as well as KA(u,C)K^A(u, C) and KA∗(u,C)K_A^*(u, C) for every complex CC.

Indeed, there exists x∈Lx \in L such that u(x)=1u(x) = 1. Then K(u)K(u) is homotopic to zero by the proposition 1, and therefore also KA(u,C)K^A(u, C) (X, p. 64, prop. 3) and KA∗(u,C)K_A^*(u, C) (X, p. 83, prop. 3).

Corollary 2

Let CC be a complex, Ann (C) its annihilator. Then q+Ann(C)q + \mathrm{Ann}(C) annihilates HA(u,C)H^A(u, C) and HA∗(u,C)H_A^*(u, C).

For every λ∈q\lambda \in q, the homothety λK(u)\lambda_{K(u)} is homotopic to zero by the proposition, hence so also are 1C⊗λK(u)1_C \otimes \lambda_{K(u)} and Homgr (λK(u),1C\lambda_{K(u)}, 1_C) by X, p. 64, prop. 3 and X, p. 83, prop. 3; it follows that λ\lambda annihilates H(u,C)H(u, C) and H∗(u,C)H^*(u, C). If λ∈Ann(C)\lambda \in \mathrm{Ann}(C), then 1K(u)⊗λC1_{K(u)} \otimes \lambda_C and Homgr (1K(u),λC1_{K(u)}, \lambda_C) are zero.

Suppose LL projective (resp. K(u)K(u) acyclic in degrees > 0). Then the complex Λ(L)\Lambda(L) is projective by III, p. 87, cor. 2 (resp. is a resolution of A/qA/q); by X, p. 102 (resp. p. 100), one therefore has, for every AA-module MM, homomorphisms

(2) HrA(u,M)→TorrA(A/q,M)H_r^A(u, M) \to \mathrm{Tor}_r^A(A/q, M), ExtAr(A/q,M)→HrA(u,M)\mathrm{Ext}_A^r(A/q, M) \to H_r^A(u, M) resp. (3) TorrA(A/q,M)→HrA(u,M)\mathrm{Tor}_r^A(A/q, M) \to H_r^A(u, M), Hr(u,M)→ExtAr(A/q,M)H_r(u, M) \to \mathrm{Ext}_A^r(A/q, M).

If LL is projective and K(u)K(u) acyclic in degrees > 0, the homomorphisms (2) and (3) above are bijective and reciprocal to one another (X, p. 102, prop. 1).

Proposition 2

Let (Li)i∈I(L_i)_{i \in I} be a family of AA-modules, where the set II is finite and totally ordered. Let uu be a linear form on ⨁i∈ILi\bigoplus_{i \in I} L_i, uiu_i its restriction to LiL_i.

The canonical isomorphism of AA-algebras (III, p. 84)

g:⨂i∈IΛ(Li)→Λ(⨁i∈ILi) g : \bigotimes_{i \in I} \Lambda(L_i) \to \Lambda(\bigoplus_{i \in I} L_i)

is an isomorphism of the complex ⨂i∈IK(ui)\bigotimes_{i \in I} K(u_i) (X, p. 63) onto the complex K(u)K(u).

In fact, by X, p. 64, remark 4, the differential DD of the complex ⨂i∈IK(ui)\bigotimes_{i \in I} K(u_i) is an antiderivation; the antiderivations dud_u and g∘D∘g−1g \circ D \circ g^{-1} of Λ(⨁Li)\Lambda(\bigoplus L_i) coincide on ⨁Li\bigoplus L_i with the mapping x↦u(x).1x \mapsto u(x).1 of ⨁Li\bigoplus L_i into Λ(⨁Li)\Lambda(\bigoplus L_i), hence are equal (III, p. 128, cor.).

Let CC and C′C' be two complexes of AA-modules. One has (X, p. 63 and p. 99) canonical isomorphisms of complexes

C⊗A(C′⊗AK(u))→(C⊗AC′)⊗AK(u) C \otimes_A (C' \otimes_A K(u)) \to (C \otimes_A C') \otimes_A K(u)
HomgrA(C′,HomgrA(K(u),C))→HomgrA(C′⊗AK(u),C), \mathrm{Homgr}_A(C', \mathrm{Homgr}_A(K(u), C)) \to \mathrm{Homgr}_A(C' \otimes_A K(u), C),

that is, isomorphisms

(4)

C⊗AKA(u,C′)→KA(u,C⊗AC′) C \otimes_A K^A(u, C') \to K^A(u, C \otimes_A C')

(5)

Homgr⁡A(C′,KAi(u,C))→Homgr⁡A(KA(u,C′),C) \operatorname{Homgr}_A(C', K^i_A(u, C)) \to \operatorname{Homgr}_A(K^A(u, C'), C)

Dans (4) and (5), take C′=K(u′)C' = K(u'), where u′:L′→Au' : L' \to A is a linear form on an AA-module L′L', and note that KA(u,K(u′))K^A(u, K(u')) which is equal by definition to K(u′)⊗AK(u)K(u') \otimes_A K(u) is identified, according to Prop. 2, with K(u′⊕u)K(u' \oplus u) where u′⊕u:L′⊕L→Au' \oplus u : L' \oplus L \to A is the linear form (x′,x)↦u′(x′)+u(x)(x', x) \mapsto u'(x') + u(x). We then obtain isomorphisms of complexes

(6)

KA(u′⊕u,C)→KA(u,KA(u′,C)) K^A(u' \oplus u, C) \to K^A(u, K^A(u', C))

(7)

KAi(u′,KAi(u,C))→KAi(u′⊕u,C) K^i_A(u', K^i_A(u, C)) \to K^i_A(u' \oplus u, C)

By passing to homology, we deduce isomorphisms of AA-modules

HrA(u′⊕u,C)→HrA(u,KA(u′,C)),r∈Z, H^A_r(u' \oplus u, C) \to H^A_r(u, K^A(u', C)) , \qquad r \in \mathbf{Z} ,
HAr(u′,KAi(u,C))→HAr(u′⊕u,C),r∈Z. H^r_A(u', K^i_A(u, C)) \to H^r_A(u' \oplus u, C) , \qquad r \in \mathbf{Z} .

Finally note that the homomorphism deduced from the product in the algebra Λ(L)\Lambda(L)

m:KA(u)⊗AKA(u)→KA(u) m : K^A(u) \otimes_A K^A(u) \to K^A(u)

is a morphism of complexes (since dud_u is an antiderivation). Supposing that LL is free of rank n and composing with the morphism of complexes KA(u)→ΛnL(−n)K^A(u) \to \Lambda^n L(-n) which is the identity in degree nn, we deduce a morphism of complexes

χ:KA(u)⊗AKA(u)→ΛnL(−n); \chi : K^A(u) \otimes_A K^A(u) \to \Lambda^n L(-n) ;

to this morphism corresponds canonically, according to X, p. 99, Prop. 12, a morphism of complexes

φ:KA(u)→Homgr⁡A(KA(u),ΛnL(−n)) \varphi : K^A(u) \to \operatorname{Homgr}_A(K^A(u), \Lambda^n L(-n))

which is bijective (III, p. 87, formula (20)). For every complex CC, we deduce a composed isomorphism

KA(u,C)=C⊗AKA(u)→1⊗φC⊗Homgr⁡A(KA(u),ΛnL(−n))→→Homgr⁡A(KA(u),C⊗AΛnL(−n))=KAi(u,C⊗AΛnL(−n)). K^A(u, C) = C \otimes_A K^A(u) \xrightarrow{1 \otimes \varphi} C \otimes \operatorname{Homgr}_A(K^A(u), \Lambda^n L(-n)) \to \\ \to \operatorname{Homgr}_A(K^A(u), C \otimes_A \Lambda^n L(-n)) = K^i_A(u, C \otimes_A \Lambda^n L(-n)) .

By passing to homology, we therefore have canonical isomorphisms

(8)

HrA(u,C)→HAn−r(u,C⊗AΛnL),r∈Z. H^A_r(u, C) \to H^{n-r}_A(u, C \otimes_A \Lambda^n L) , \qquad r \in \mathbf{Z} .

Remarks. — 1) \* The above remains valid when LL is projective of rank nn. \*

2) Since LL is free of rank nn, ΛnL\Lambda^n L is isomorphic to AA, and we have non-canonical isomorphisms HrA(u,C)→HAn−r(u,C)H^A_r(u,C)\to H^{n-r}_A(u,C).

2. Functoriality

Let f:C→C′f:C\to C' be a morphism of complexes. We denote

KA(u,f):KA(u,C)→KA(u,C′), K^A(u,f):K^A(u,C)\to K^A(u,C'),
KA(u,f):KA(u,C)→KA(u,C′), K_A(u,f):K_A(u,C)\to K_A(u,C'),

the morphisms of complexes f⊗1K(u)f\otimes 1_{K(u)} and Hom⁡A(1K(u),f)\operatorname{Hom}_{A}(1_{K(u)},f).

We denote by HA(u,f):HA(u,C)→HA(u,C′)H^A(u,f):H^A(u,C)\to H^A(u,C'), HA(u,f):HA(u,C)→HA(u,C′)H_A(u,f):H_A(u,C)\to H_A(u,C'), HrA(u,f):HrA(u,C)→HrA(u,C′)H^A_r(u,f):H^A_r(u,C)\to H^A_r(u,C'), HrA(u,f):HrA(u,C)→HrA(u,C′)H^A_r(u,f):H^A_r(u,C)\to H^A_r(u,C') the morphisms induced in homology. The mapping f↦KA(u,f)f\mapsto K^A(u,f) is linear; if g:C′→C′′g:C'\to C'' is another morphism of complexes, we have KA(u,g∘f)=KA(u,g)∘KA(u,f)K^A(u,g\circ f)=K^A(u,g)\circ K^A(u,f); analogously for KAK_A, HAH^A, HAH_A, HrAH^A_r, HrAH^A_r.

Let 0→C′→fC→gC′′→00\to C'\xrightarrow{f}C\xrightarrow{g}C''\to0 be an exact sequence of complexes.

a) Suppose LL is flat; then Λ(L)\Lambda(L) is flat (X, p. 15, Cor.). The sequence

0→KA(u,C′)→KA(u,f)KA(u,C)→KA(u,g)KA(u,C′′)→0 0\to K^A(u,C')\xrightarrow{K^A(u,f)}K^A(u,C)\xrightarrow{K^A(u,g)}K^A(u,C'')\to0

is then exact, and gives rise (X, p. 30) to an exact sequence of homology

⋯→HnA(u,C′)→Hn(u,f)HnA(u,C)→Hn(u,g)HnA(u,C′′)→∂nHn−1A(u,C′)→⋯ . \cdots\to H^A_n(u,C')\xrightarrow{H_n(u,f)}H^A_n(u,C) \xrightarrow{H_n(u,g)}H^A_n(u,C'') \xrightarrow{\partial_n}H^A_{n-1}(u,C')\to\cdots .

b) Suppose LL is projective; then Λ(L)\Lambda(L) is projective. The sequence

0→KA(u,C′)→KA(u,f)KA(u,C)→KA(u,g)KA(u,C′′)→0 0\to K_A(u,C')\xrightarrow{K_A(u,f)}K_A(u,C) \xrightarrow{K_A(u,g)}K_A(u,C'')\to0

is then exact, and gives rise to an exact sequence of homology

⋯→HAn(u,C′)→Hn(u,f)HAn(u,C)→Hn(u,g)HAn(u,C′′)→δnHAn+1(u,C′)→⋯ . \cdots\to H^n_A(u,C')\xrightarrow{H^n(u,f)}H^n_A(u,C) \xrightarrow{H^n(u,g)}H^n_A(u,C'') \xrightarrow{\delta^n}H^{n+1}_A(u,C')\to\cdots .

Let ρ:A→A′\rho:A\to A' be a ring homomorphism, L′L' the A′A'-module A′⊗ALA'\otimes_A L, u′:L′→A′u':L'\to A' the linear form 1⊗u1\otimes u. The bijective canonical homomorphism (III, p. 83, Prop. 8)

ψ:ΛA′(A′⊗AL)→A′⊗AΛA(L) \psi:\Lambda_{A'}(A'\otimes_A L)\to A'\otimes_A\Lambda_A(L)

is an isomorphism of complexes of A′A'-modules. It follows that:

1) for every complex of A′A'-modules C′C', an isomorphism of complexes of AA-modules

KAA′(u′,C′)→KAA(u,C′), K^{A'}_{A}(u',C')\to K^A_A(u,C'),

composed of the diagram

C′⊗A′(ΛA(A′⊗AL))→1C⊗ψC′⊗A′A′⊗AΛA(L)→C′⊗AΛA(L) C' \otimes_{A'} (\Lambda_A(A' \otimes_A L)) \xrightarrow{1_C \otimes \psi} C' \otimes_{A'} A' \otimes_A \Lambda_A(L) \to C' \otimes_A \Lambda_A(L)

where φ\varphi is the canonical bijection (III, p. 85, Prop. 14);

2) for every complex of AA-modules CC, an isomorphism of complexes of A′A'-modules

KA(u,A′⊗AC)→A′⊗AKA(u,C), K^A(u, A' \otimes_A C) \to A' \otimes_A K^A(u, C),

whence homomorphisms of A′A'-modules

A′⊗AHnA(u,C)→HnA′(u′,A′⊗AC), A' \otimes_A H_n^A(u, C) \to H_n^{A'}(u', A' \otimes_A C),

which are bijective when A′A' is flat over AA (X, p. 66, Cor. 2).

Let L′L' be an AA-module, u′:L′→Au' : L' \to A a linear form, f:L→L′f : L \to L' an AA-homomorphism such that u′∘f=uu' \circ f = u. It follows from III, p. 161, formula (55), that the homomorphism Λ(f):Λ(L)→Λ(L′)\Lambda(f) : \Lambda(L) \to \Lambda(L') satisfies du∘Λ(f)=Λ(f)∘du′d_u \circ \Lambda(f) = \Lambda(f) \circ d_{u'}, hence defines a morphism of complexes Λ(u):KA(u)→KA(u′)\Lambda(u) : K^A(u) \to K^A(u'). If CC is an AA-complex, it follows that there are morphisms of complexes

1C⊗Λ(u):KA(u,C)→KA(u′,C)1_C \otimes \Lambda(u) : K^A(u, C) \to K^A(u', C) and Homgr(Λ(u),1C):KA∗(u′,C)→KA∗(u,C)\mathrm{Homgr}(\Lambda(u), 1_C) : K_A^*(u', C) \to K_A^*(u, C).

If ff is bijective, all these morphisms are isomorphisms.

3. Example 1: the complex S(L)⊗AΛ(L)S(L) \otimes_A \Lambda(L)

Let AA be a ring, LL an AA-module, S(L)S(L) its symmetric algebra, S(L)⊗ALS(L) \otimes_A L the S(L)S(L)-module deduced by extension of scalars, u:S(L)⊗AL→S(L)u : S(L) \otimes_A L \to S(L) the linear form such that u(s⊗x)=sxu(s \otimes x) = sx for s∈S(L),x∈Ls \in S(L), x \in L. By the canonical isomorphism of S(L)S(L)-modules (III, p. 83, Prop. 8)

ΛS(L)(S(L)⊗AL)→S(L)⊗AΛ(L), \Lambda_{S(L)}(S(L) \otimes_A L) \to S(L) \otimes_A \Lambda(L),

the differential of the complex KS(L)(u)K^{S(L)}(u) is carried over into the mapping

d:S(L)⊗AΛ(L)→S(L)⊗AΛ(L) d : S(L) \otimes_A \Lambda(L) \to S(L) \otimes_A \Lambda(L)

such that, for x1,...,xp,y1,...,yqx_1, ..., x_p, y_1, ..., y_q in LL, one has

(9)d((x1...xp)⊗(y1∧...∧yq))=∑i=1q(−1)i+1yix1...xp⊗(y1∧...∧yi−1∧yi+1∧...∧yq). \begin{align} (9) \quad d((x_1 ... x_p) \otimes (y_1 \wedge ... \wedge y_q)) \\ &= \sum_{i=1}^q (-1)^{i+1} y_i x_1 ... x_p \otimes (y_1 \wedge ... \wedge y_{i-1} \wedge y_{i+1} \wedge ... \wedge y_q). \end{align}

Note that dd maps Sp(L)⊗Λq(L)S^p(L) \otimes \Lambda^q(L) into Sp+1(L)⊗Λq−1(L)S^{p+1}(L) \otimes \Lambda^{q-1}(L), hence that the complex of AA-modules S(L)⊗Λ(L)S(L) \otimes \Lambda(L) decomposes into the direct sum of the complexes described by the following diagrams:

(En):0→S0L⊗AΛnL→S1L⊗AΛn−1L→...→SnL⊗AΛ0L→0,n∈N. (\mathscr{E}_n) : 0 \to S^0 L \otimes_A \Lambda^n L \to S^1 L \otimes_A \Lambda^{n-1} L \to ... \to S^n L \otimes_A \Lambda^0 L \to 0, \quad n \in \mathbf{N}.

If the A-module L is the direct sum of a finite family (Li)i∈I(L_i)_{i \in I} where I is totally ordered, the canonical bijection

⨂i∈I(S(Li)⊗AΛ(Li))→S(L)⊗AΛ(L) \bigotimes_{i \in I} (\mathbf{S}(L_i) \otimes_A \Lambda(L_i)) \to \mathbf{S}(L) \otimes_A \Lambda(L)

is an isomorphism of complexes of A-modules (this follows from prop. 2 of X, p. 148 or from formula (9) above).

Proposition 3

If the A-module L is flat, the sequences (En)(\mathscr{E}_n) above are exact for n>0n > 0.

a) Let us first note that, if pLp_L is the composite homomorphism

S(L)⊗Λ(L)→αS0(L)⊗Λ0(L)→βA, \mathbf{S}(L) \otimes \Lambda(L) \xrightarrow{\alpha} \mathbf{S}^0(L) \otimes \Lambda^0(L) \xrightarrow{\beta} A,

where α\alpha is the tensor product of the canonical projections and β\beta the canonical isomorphism, it remains to prove that H(pL)H(p_L) is bijective.

b) If L=0L = 0 or if L=AL = A, the proposition is obvious.

c) Suppose L is free of finite rank; write it as the direct sum L1⊕...⊕LnL_1 \oplus ... \oplus L_n of free A-modules of rank 1. According to the remark preceding the proposition, the complex S(L)⊗Λ(L)\mathbf{S}(L) \otimes \Lambda(L) is isomorphic to the tensor product of the nn free complexes S(Li)⊗Λ(Li)\mathbf{S}(L_i) \otimes \Lambda(L_i) whose homology is free according to b). According to X, p. 79, cor. 4, the canonical homomorphism

γ:⨁i=1nH(S(Li)⊗Λ(Li))→H(S(L)⊗Λ(L)) \gamma : \bigoplus_{i=1}^n H(\mathbf{S}(L_i) \otimes \Lambda(L_i)) \to H(\mathbf{S}(L) \otimes \Lambda(L))

is bijective. According to b) H(pLi)H(p_{L_i}) is bijective for every ii. Since ⨂i=1nH(pLi)=H(pL)∘γ\bigotimes_{i=1}^n H(p_{L_i}) = H(p_L) \circ \gamma,

H(pL) is bijective. H(p_L) \text{ is bijective}.

d) In the general case, L is an inductive limit of a filtered inductive system (Li)i∈I(L_i)_{i \in I} of free modules of finite rank (X, p. 14, th. 1). Since the canonical bijective homomorphism

lim←⁡S(Li)⊗Λ(Li)→S(L)⊗Λ(L) \varprojlim \mathbf{S}(L_i) \otimes \Lambda(L_i) \to \mathbf{S}(L) \otimes \Lambda(L)

is an isomorphism of complexes, the proposition follows from X, p. 28, prop. 1.

Remark 1

We shall see below (X, p. 158, example) another proof of part c) above.

Remark 2

If A is a Q\mathbf{Q}-algebra, the conclusion of prop. 3 remains true without assumption on L (cf. X, p. 206, exercise 1).

Remark 3

Let G be a group and ρ:G→GL(L)\rho : G \to \mathbf{GL}(L) a linear representation of G in a flat A-module L. Then the (En)(\mathscr{E}_n) are exact sequences of linear representations. Suppose L finitely generated projective, and denote by RA(G)R_A(G) the ring of representations of G in finitely generated projective A-modules. It follows from prop. 3 that one has in R_A(G) the relations

∑i=0n(−1)i[Si(L)][Λn−i(L)]=0,n>0. \sum_{i=0}^{n} (-1)^i [\mathbf{S}^i(L)] [\Lambda^{n-i}(L)] = 0 , \quad n > 0 .

If one considers the formal series

s(T)=∑i=0∞[Si(L)]Ti∈RA(G)[[T]], s(T) = \sum_{i=0}^{\infty} [\mathbf{S}^i(L)] T^i \in R_A(G)[[T]] ,
λ(T)=∑i=0∞[Λi(L)]Ti∈RA(G)[[T]], \lambda(T) = \sum_{i=0}^{\infty} [\Lambda^i(L)] T^i \in R_A(G)[[T]] ,

the relations (10) are written

s(T)λ(−T)=1∗. s(T) \lambda(-T) = 1_* .

4. Example 2 : the case of a free module

Let k be a ring, M a k-module, I a set and p an integer ⩾0\geqslant 0. A mapping m:Ip→Mm : I^p \to M is said to be alternating if it satisfies the following two conditions:

a) for every permutation σ∈Sp\sigma \in S_p and every sequence (α1,...,αp)∈Ip(\alpha_1, ..., \alpha_p) \in I^p, one has

m(ασ(1),...,ασ(p))=εσm(α1,...,αp), m(\alpha_{\sigma(1)}, ..., \alpha_{\sigma(p)}) = \varepsilon_{\sigma} m(\alpha_1, ..., \alpha_p) ,

b) for every sequence (α1,...,αp)∈Ip(\alpha_1, ..., \alpha_p) \in I^p such that two of the indices α1,...,αp\alpha_1, ..., \alpha_p are equal, one has m(α1,...,αp)=0m(\alpha_1, ..., \alpha_p) = 0. (In the case where I is a k-module and m is multilinear, one recovers the notion introduced in III, p. 80.)

Suppose I finite and denote by CIp(M)C_I^p(M) the k-module of alternating mappings of IpI^p into M.

Let L0L_0 be a k-module, (ei)i∈I(e_i)_{i \in I} a family of elements of L0L_0; one defines two k-linear mappings

g:Homk(ΛpL0,M)→CIp(M) g : \mathrm{Hom}_k (\Lambda^p L_0, M) \to C_I^p(M)
h:CIp(M)→M⊗kΛpL0 h : C_I^p(M) \to M \otimes_k \Lambda^p L_0

as follows: if f∈Homk(ΛpL0,M)f \in \mathrm{Hom}_k (\Lambda^p L_0, M), one sets

g(f)(α1,...,αp)=f(eα1∧...∧eαp); g(f)(\alpha_1, ..., \alpha_p) = f(e_{\alpha_1} \wedge ... \wedge e_{\alpha_p}) ;

let m∈CIp(M)m \in C_I^p(M), define h(m)∈M⊗kΛpL0h(m) \in M \otimes_k \Lambda^p L_0. For every element (α1,...,αp)(\alpha_1, ..., \alpha_p) of IpI^p, the element m(α1,...,αp)⊗(eα1∧...∧eαp)m(\alpha_1, ..., \alpha_p) \otimes (e_{\alpha_1} \wedge ... \wedge e_{\alpha_p}) of ΛpL0⊗kM\Lambda^p L_0 \otimes_k M is zero if Card{α1,...,αp}<p\mathrm{Card}\{ \alpha_1, ..., \alpha_p \} < p and is independent of the order of the indices α1,...,αp\alpha_1, ..., \alpha_p if

Card{α1,...,αp}=p. \mathrm{Card}\{ \alpha_1, ..., \alpha_p \} = p .
It depends only on the subset J=(α1,...,αp)J = (\alpha_1, ..., \alpha_p) of II; denote it by hJ(m)h_J(m); we have hJ(m)=0h_J(m) = 0 if Card(J)<p\mathrm{Card}(J) < p; we then set:
h(m)=∑JhJ(m), h(m) = \sum_J h_J(m) ,
where JJ runs through the subsets of II having pp elements.

Lemma 2

If the k-module L0L_0 is free with basis (ei)i∈I(e_i)_{i \in I}, the k-linear mappings g and h are bijective.* This follows from III, p. 79, th. 1.

Let MM be a kk-module, and let x=(xi)i∈Ix = (x_i)_{i \in I} be a family of kk-endomorphisms of MM, pairwise permutable. Consider the polynomial ring A=k[(Xi)i∈I]A = k[(X_i)_{i \in I}] and endow MM with the structure of an AA-module such that P(Xi)m=P(xi)mP(X_i) m = P(x_i) m for P∈AP \in A and m∈Mm \in M. Let moreover LL be the free AA-module AIA^I, (ei)i∈I(e_i)_{i \in I} its canonical basis and u:L→Au : L \to A the linear form which maps eie_i onto XiX_i for all i∈Ii \in I. Consider the complexes of kk-modules KA∗(u,M)K_A^*(u, M) and KA∗(u,M)K_A^*(u, M); one has canonical isomorphisms

KAp(u,M)=HomA(ΛAp(AI),M)→Homk(Λkp(kI),M), K_A^p(u, M) = \mathrm{Hom}_A(\Lambda_A^p(A^I), M) \to \mathrm{Hom}_k(\Lambda_k^p(k^I), M) ,
M⊗kΛkp(kI)→M⊗AΛAp(AI)=KAp(u,M); M \otimes_k \Lambda_k^p(k^I) \to M \otimes_A \Lambda_A^p(A^I) = K_A^p(u, M) ;
whence by composition with the isomorphisms gg and hh isomorphisms of k-modules
θp:KAp(u,M)→CIp(M), \theta^p : K_A^p(u, M) \to C_I^p(M) ,
θp:CIp(M)→KAp(u,M). \theta_p : C_I^p(M) \to K_A^p(u, M) .
We denote by
∂p:CIp(M)→CIp+1(M), \partial^p : C_I^p(M) \to C_I^{p+1}(M) ,
∂p:CIp(M)→CIp−1(M), \partial_p : C_I^p(M) \to C_I^{p-1}(M) ,
the kk-homomorphisms obtained by transporting the differentials of KA∗(u,M)K_A^*(u, M) and KA∗(u,M)K_A^*(u, M) by the isomorphisms θ\theta. For example, one has: (12)
(∂pm)(α1,...,αp+1)=∑j=1p+1(−1)j+1xαjm(α1,...,αj−1,αj+1,...,αp+1). (\partial^p m)(\alpha_1, ..., \alpha_{p+1}) = \sum_{j=1}^{p+1} (-1)^{j+1} x_{\alpha_j} m(\alpha_1, ..., \alpha_{j-1}, \alpha_{j+1}, ..., \alpha_{p+1}) .
The complex formed by the CIp(M)C_I^p(M) and the ∂p\partial^p (resp. the ∂p\partial_p) is denoted by K∗(x,M)K^*(x, M) (resp. K.(x,M)K.(x, M)) and is called the ascending (resp. descending) Koszul complex associated with the module M and the sequence of endomorphisms (x1,...,xn)(x_1, ..., x_n). One therefore has isomorphisms of complexes of kk-modules

θ∗:KA∗(u,M)→K∗(x,M), \theta^*: K_A^*(u, M) \to K^*(x, M),
θ∗:K∗(x,M)→KA∗(u,M). \theta_*: K_*(x, M) \to K_A^*(u, M).

Remark

Conversely, let B be a kk-algebra, L a free B-module with basis (ei)i∈I(e_i)_{i \in I}, and M a B-module. The datum of a linear form u:L→Bu : L \to B is equivalent to that of a family x=(xi)i∈Ix = (x_i)_{i \in I} of elements of B, by the relation xi=u(ei)x_i = u(e_i). The underlying complex of kk-modules of KB∗(u,M)K_B^*(u, M) (resp. KB(u,M)K^B(u, M)) is then identified, by the isomorphism θ∗\theta^* (resp. θ∗\theta_*) with the Koszul complex K∗(x,M)K^*(x, M) (resp. K∗(x,M)K_*(x, M)). For example KB(u)K^B(u) is identified with K∗(x,B)K^*(x, B).

The notations H∗(x,M)H_*(x, M), H∗(x,M)H_*(x, M), etc., are introduced as in No. 1 (X, p. 147), and all the results of Nos. 1 and 2 apply mutatis mutandis, the module AIA^I being free. We have for example isomorphisms

H0(x,M)→M/(x)M H_0(x, M) \to M/(x) M
H0(x,M)→Hom⁡A(A/(x),M), H^0(x, M) \to \operatorname{Hom}_A(A/(x), M),

where (x)(x) denotes the ideal ∑Axi\sum A x_i of A. For example also, if K∗(x,A)K_*(x, A) is acyclic in degrees >0> 0, we have isomorphisms

Hr(x,M)→Tor⁡rA(k,M), H_r(x, M) \to \operatorname{Tor}_r^A(k, M),
Ext⁡Ar(k,M)→Hr(x,M). \operatorname{Ext}_A^r(k, M) \to H^r(x, M).

Finally, suppose that I is (finite and) totally ordered, for example I={1,...,n}I = \{ 1, ..., n \}; identify Λn(AI)\Lambda^n(A^I) with A thanks to the basis element e1∧...∧ene_1 \wedge ... \wedge e_n and translate the isomorphism KpA(u,M)→KAn−p(u,M)K_p^A(u, M) \to K_A^{n-p}(u, M) of X, p. 149. It becomes, by transport by (θp)(\theta_p) and (θn−p)(\theta^{n-p}), the isomorphism

C1p(M)→C1n−p(M) C_1^p(M) \to C_1^{n-p}(M)

which associates with m∈C1p(M)m \in C_1^p(M) the element m~\tilde{m} of C1n−p(M)C_1^{n-p}(M) such that

m(α1,...,αp)=m~(β1,...,βn−p) m(\alpha_1, ..., \alpha_p) = \tilde{m}(\beta_1, ..., \beta_{n-p})

if (α1,...,αp,β1,...,βn−p)(\alpha_1, ..., \alpha_p, \beta_1, ..., \beta_{n-p}) is an even permutation of {1,...,n}\{ 1, ..., n \}. Let us also remark that when I={1,...,n}I = \{ 1, ..., n \}, we can identify C1p(M)C_1^p(M) with the set of families m(α1,...,αp)m(\alpha_1, ..., \alpha_p) of elements of M where α1<α2<...<αp\alpha_1 < \alpha_2 < ... < \alpha_p; formula (12) remains valid, as does relation (13).

Example

Take M=k[T1,...,Tn]M = k[T_1, ..., T_n]; the Koszul complex K(∂/∂T,M)K (\partial/\partial T, M) associated with the sequence of endomorphisms (∂/∂T1,...,∂/∂Tn)(\partial/\partial T_1, ..., \partial/\partial T_n) is identified with the de Rham complex of k[x1,...,xn]k[x_1, ..., x_n] over kk (X, p. 44): with m∈C{1,...,n}p(M)m \in C_{\{1,...,n\}}^p(M), we associate the differential form

ω(m)=∑α1<...<αpm(α1,...,αp) dxα1∧...∧dxαp, \omega(m) = \sum_{\alpha_1 < ... < \alpha_p} m(\alpha_1, ..., \alpha_p)\ dx_{\alpha_1} \wedge ... \wedge dx_{\alpha_p},

cf. formula (12) and Example 1, p. 44.

5. Example 3 : the case L=AL = A

If L=AL = A, put u(1)=x∈Au(1) = x \in A. The complex K(u)K(u) is then of length 1, we have K0(u)=K1(u)=AK_0(u) = K_1(u) = A and d1(a)=xad_1(a) = xa, hence for every AA-module MM, K0(u,M)K_0(u, M), K1(u,M)K_1(u, M), K0(u,M)K^0(u, M) and K1(u,M)K^1(u, M) are identified with MM, the differentials

d1:K1(u,M)→K0(u,M)andd0:K0(u,M)→K1(u,M) d_1 : K_1(u, M) \to K_0(u, M) \quad \text{and} \quad d^0 : K^0(u, M) \to K^1(u, M)

being m↦xmm \mapsto xm. We therefore have isomorphisms

H0(x,M)→A/xA⊗AM←H1(x,M), H_0(x, M) \to A/xA \otimes_A M \leftarrow H^1(x, M),
H1(x,M)→HomA(A/xA,M)←H0(x,M). H_1(x, M) \to \mathrm{Hom}_A(A/xA, M) \leftarrow H^0(x, M).

Lemma 3

Let KK be a complex such that Ki=0K_i = 0 for i≠0,1i \neq 0, 1, and let CC be a complex and pp an integer.

a) If KK is flat, there is for every p∈Np \in \mathbf{N} an exact sequence

0→H0(K⊗AHp(C))→Hp(K⊗AC)→H1(K⊗AHp−1(C))→0. 0 \to H_0(K \otimes_A H_p(C)) \to H_p(K \otimes_A C) \to H_1(K \otimes_A H_{p-1}(C)) \to 0.

b) If KK is projective, we have for every p∈Np \in \mathbf{N} an exact sequence

0→H1(HomgrA(K,Hp−1(C)))→Hp(HomgrA(K,C)) 0 \to H^1(\mathrm{Homgr}_A(K, H^{p-1}(C))) \to H^p(\mathrm{Homgr}_A(K, C))
→H0(HomgrA(K,Hp(C)))→0. \to H^0(\mathrm{Homgr}_A(K, H^p(C))) \to 0.

Let us prove a), the proof of b) being analogous. For every ii, denote by K(i)K_{(i)} the complex Ki(−i)K_i(-i). We have an exact sequence of complexes, split as a sequence of AA-modules

0→K(0)→αK→βK(1)→0; 0 \to K_{(0)} \xrightarrow{\alpha} K \xrightarrow{\beta} K_{(1)} \to 0;

the sequence

0⟶K(0)⊗AC→α⊗1K⊗AC→β⊗1K(1)⊗AC⟶0 \begin{array}{ccccccccc} 0 & \longrightarrow & K_{(0)} \otimes_A C & \xrightarrow{\alpha \otimes 1} & K \otimes_A C & \xrightarrow{\beta \otimes 1} & K_{(1)} \otimes_A C & \longrightarrow & 0 \end{array}

is exact, and KK being flat, the homomorphisms

γ0,p(K(0),C):K(0)⊗AHp(C)→Hp(K(0)⊗AC) \gamma_{0,p}(K_{(0)}, C) : K_{(0)} \otimes_A H_p(C) \to H_p(K_{(0)} \otimes_A C)
γ1,p−1(K(1),C):K(1)⊗AHp−1(C)→Hp(K(1)⊗AC) \gamma_{1,p-1}(K_{(1)}, C) : K_{(1)} \otimes_A H_{p-1}(C) \to H_p(K_{(1)} \otimes_A C)

are bijective (X, p. 66, cor. 2). Let us calculate the connecting homomorphism ∂(α⊗1,β⊗1)\partial(\alpha \otimes 1,\beta \otimes 1); by definition, it maps the class of the cycle ∑ai⊗bi\sum a_i \otimes b_i onto that of ∑dai⊗bi\sum d a_i \otimes b_i, which means that

∂(α⊗1,β⊗1)∘γ(K(1),C)=γ(K(0),C)∘(dK⊗1). \partial(\alpha \otimes 1,\beta \otimes 1)\circ\gamma(K_{(1)},\mathbf C)=\gamma(K_{(0)},\mathbf C)\circ(d_K\otimes 1).

The exact homology sequence associated with (14) therefore has the form

K1⊗Hp(C)→dk⊗1K0⊗Hp(C)⟶Hp(K⊗C) K_1\otimes H_p(\mathbf C)\xrightarrow{d_k\otimes 1}K_0\otimes H_p(\mathbf C)\longrightarrow H_p(K\otimes\mathbf C)
⟶K1⊗Hp−1(C)→dk⊗1K0⊗Hp−1(C), \hspace{7cm}\longrightarrow K_1\otimes H_{p-1}(\mathbf C)\xrightarrow{d_k\otimes 1}K_0\otimes H_{p-1}(\mathbf C),

whence a).

Applying Lemma 3, a) to the complex K(u)K(u), and using the commutation isomorphisms, we obtain:

Proposition 4

For every complex CC, there are exact sequences .

0⟶A/xA⊗AHp(C)⟶Hp(x,C)⟶Hom⁡A(A/xA,Hp−1(C))⟶0. 0\longrightarrow A/xA\otimes_A H_p(C)\longrightarrow H_p(x,C)\longrightarrow \operatorname{Hom}_A(A/xA,H_{p-1}(C))\longrightarrow 0.

Corollary 1

For Hp(x,C)=0H_p(x,C)=0 to hold, it is necessary and sufficient that the homothety with ratio xx in Hp(C)H_p(C) be surjective and that the homothety with ratio xx in Hp−1(C)H_{p-1}(C) be injective.

Corollary 2

Let x=(x1,…,xn)x=(x_1,\ldots,x_n) be a sequence of elements of AA, MM an AA-module, x′x' the sequence (x1,…,xn−1)(x_1,\ldots,x_{n-1}). We have exact sequences

0⟶A/xnA⊗AHp(x′,M)⟶Hp(x,M)⟶Hom⁡A(A/xnA,Hp−1(x′,M))⟶0. 0\longrightarrow A/x_nA\otimes_A H_p(x',M)\longrightarrow H_p(x,M)\longrightarrow \operatorname{Hom}_A(A/x_nA,H_{p-1}(x',M))\longrightarrow 0.

Corollary 3

For Hi(x,M)H_i(x,M) to be zero for all i>0i>0, it is necessary and sufficient that the homothety with ratio xnx_n in Hi(x′,M)H_i(x',M) be bijective for i>0i>0, and that the homothety with ratio xnx_n in M/(x′)MM/(x')M be injective.

6. Completely transverse families

Let AA be a ring, MM an AA-module, x=(xi)i∈Ix=(x_i)_{i\in I} a family of elements of AA.

Definition 2

The family xx is said to be completely transverse for MM if one has Hi(x,M)=0H_i(x,M)=0 for i>0i>0.

If II is finite, it amounts to the same thing (X, p. 150) to say that one has Hi(x,M)=0H^i(x,M)=0 for i<Card⁡(I)i<\operatorname{Card}(I).

The following proposition makes it possible to give examples of completely transverse families.

Proposition 5

Let x=(x1,…,xn)x=(x_1,\ldots,x_n) be a sequence of elements of AA. If for i=1,…,ni=1,\ldots,n, the homothety of ratio xix_i in the AA-module M/(x1M+⋯+xi−1M)M/(x_1M+\cdots+x_{i-1}M) is injective, the sequence xx is completely secant for MM.

A sequence satisfying the conditions of the proposition is said to be regular for M, or M-regular. We shall note that this property depends in general on the order of the xix_i; for example the sequence (1,0)(1, 0) is always M-regular, whereas the sequence (0,1)(0, 1) is so only if M is zero. On the other hand, the fact that a sequence is completely secant does not depend on the order of the terms.

Let us prove the proposition by induction on nn, the case n=0n = 0 being immediate. Put x′=(x1,...,xn−1)x' = (x_1, ..., x_{n-1}); if the sequence xx is M-regular, the sequence x′x' is M-regular and multiplication by xnx_n in M/(x′)MM/(x')M is injective; by the induction hypothesis, we have Hi(x′,M)=0H_i(x', M) = 0 for i>0i > 0; it then follows from cor. 3 of X, p. 157, that Hi(x,M)=0H_i(x, M) = 0 for i>0i > 0.

Example

Let kk be a ring; take A=k[X1,...,Xn]A = k[X_1, ..., X_n] and x=(X1,...,Xn)x = (X_1, ..., X_n). The sequence xx is A-regular and the proposition gives again the acyclicity in degrees >0> 0 of the complex Sk(kn)⊗kΛk(kn)S_k(k^n) \otimes_k \Lambda_k(k^n) (cf. X, p. 152, prop. 3). Likewise in the ring of formal series A^=k[[X1,...,Xn]]\hat{A} = k[[X_1, ..., X_n]], the sequence (X1,...,Xn)(X_1, ..., X_n) is A^\hat{A}-regular, hence completely secant for A^\hat{A}.

Proposition 6

a) If ∑i∈IxiA=A\sum_{i \in I} x_i A = A, the family (xi)i∈I(x_i)_{i \in I} is completely secant for M. b) Let x=(x1,...,xn)x = (x_1, ..., x_n) be a sequence of elements of A. Let (aij)∈GLn(A)(a_{ij}) \in \mathrm{GL}_n(A); put

yi=∑jaijxj. y_i = \sum_j a_{ij} x_j.
If the sequence xx is completely secant for M, the same is true of the sequence (y1,...,yn)(y_1, ..., y_n). c) Let k1,...,knk_1, ..., k_n be integers ⩾1\geqslant 1; in order that the sequence (x1k1,...,xnkn)(x_1^{k_1}, ..., x_n^{k_n}) be completely secant for M, it is necessary and sufficient that the sequence xx be completely secant for M. Assertion a) follows from cor. 1, p. 148. Let us prove b). Let f:An→Anf : A^n \to A^n be the isomorphism defined by the matrix t(aij)^t(a_{ij}); it follows from X, p. 151, that 1M⊗Λ(f)1_M \otimes \Lambda(f) is an isomorphism of the complex K.(y,M)K.(y, M) onto the complex K.(x,M)K.(x, M), whence b). To prove c), it is obviously enough to prove that if kk is an integer ⩾1\geqslant 1, the sequence (x1,...,xn−1,xnk)(x_1, ..., x_{n-1}, x_n^k) is completely secant for M if and only if the same is true of the sequence xx. Let x′=(x1,...,xn−1)x' = (x_1, ..., x_{n-1}); by cor. 3, p. 157, the first condition (resp. the second) means that the homothety of ratio xnkx_n^k (resp. xnx_n) is bijective in Hi(x′,M)H_i(x', M) for i⩾1i \geqslant 1 and injective in M/(x′)MM/(x')M. These two conditions are clearly equivalent, whence c).

Remark 1

The assertion analogous to c) for regular sequences is true (X, p. 207, exercise 5).

Proposition 7

a) Let N be a flat A-module. If the family x is completely secant for M, it is so for M⊗ANM \otimes_A N.

b) Let 0→M′→M→M′′→00 \to M' \to M \to M'' \to 0 be an exact sequence of AA-modules. If the family xx is completely transverse for M′M' and for M′′M'', it is so for MM.

The complex K.(x,M⊗AN)K.(x, M \otimes_A N) is isomorphic by definition to K.(x,M)⊗ANK.(x, M) \otimes_A N; since NN is flat, one deduces from this an isomorphism H.(x,M)⊗AN→H.(x,M⊗AN)H.(x, M) \otimes_A N \to H.(x, M \otimes_A N) (X, p. 66, cor. 2), whence a).

The complex K.(x)K.(x) being flat, one has an exact sequence of complexes

0→K.(x,M′)→K.(x,M)→K.(x,M′′)→0; 0 \to K.(x, M') \to K.(x, M) \to K.(x, M'') \to 0 ;

assertion b) follows from the associated exact homology sequence.

Remark 2

The analogous assertions to a) and b) for regular sequences are immediate.

Remark 3

If the family xx is completely transverse for AA, the complex K.(x,A)K.(x, A) defines a free resolution of the AA-module A/xA/x, with x=∑i∈IxiAx = \sum_{i \in I} x_i A; therefore for every integer i⩾0i \geqslant 0 and for every AA-module MM there are isomorphisms

ExtAn−i(A/x,M)→Hn−i(x,M)→Hi(x,M)→Tor⁡iA(A/x,M). \text{Ext}_A^{n-i}(A/x, M) \to H^{n-i}(x, M) \to H_i(x, M) \to \operatorname{Tor}_i^A(A/x, M) .

Remark 4

One says that the sequence x=(x1,...,xn)x = (x_1, ..., x_n) is MM-coregular if (denoting by (xi)M(x_i)_M the homothety of ratio xix_i in MM) the homothety of ratio xix_i in the module

Ker⁡(x1)M∩...∩Ker⁡(xi−1)M \operatorname{Ker}(x_1)_M \cap ... \cap \operatorname{Ker}(x_{i-1})_M

is surjective for i=1,...,ni = 1, ..., n. One then has Hi(x,M)=0H_i(x, M) = 0 for i<ni < n: this is proved in the same way as prop. 5.

Take for example A=k[D1,...,Dn]A = k[D_1, ..., D_n], where kk is a Q\mathbf{Q}-algebra, and M=k[T1,...,Tn]M = k[T_1, ..., T_n], endowed with the AA-module structure such that DiP=∂P/∂TiD_i P = \partial P / \partial T_i for every P∈MP \in M (1⩽i⩽n1 \leqslant i \leqslant n). One verifies immediately that the sequence (D1,...,Dn)(D_1, ..., D_n) is MM-coregular; taking into account the example p. 155, one deduces that the de Rham complex of k[T1,...,Tn]k[T_1, ..., T_n] over kk is acyclic in degrees >0> 0.

7. A criterion for completely transverse sequences

Let AA be a ring, MM an AA-module, xx an ideal of AA. The xx-adic topology on MM is defined to be the topology compatible with the group structure of MM for which the set of submodules xrMx^r M (r⩾0r \geqslant 0) is a fundamental system of neighbourhoods of zero (TG, III, p. 5, example). This topology is separated if and only if

⋂r⩾0xr=0. \bigcap_{r \geqslant 0} x^r = 0 .

Suppose now that the ideal xx is generated by a sequence x=(x1,...,xn)x = (x_1, ..., x_n) of elements of A. Consider the graded AA-module ⨁r≥0xrM\bigoplus_{r \geq 0} x^r M and the graded AA-homomorphism of degree 0

aMx:A[X1,...,Xn]⊗AM→⨁r≥0xrM a_M^x : A[X_1, ..., X_n] \otimes_A M \to \bigoplus_{r \geq 0} x^r M

such that aMx(P⊗m)=P(x1,...,xn)ma_M^x(P \otimes m) = P(x_1, ..., x_n) m if P is a homogeneous polynomial in X1,...,XnX_1, ..., X_n and m∈Mm \in M. Let d\mathfrak{d} be the ideal of A[X1,...,Xn]A[X_1, ..., X_n] generated by the elements (xiXj−xjXi)(x_i X_j - x_j X_i) for 1≤i<j≤n1 \leq i < j \leq n. We have P(x1,...,xn)=0P(x_1, ..., x_n) = 0 if P∈dP \in \mathfrak{d}, so that aMxa_M^x gives by passing to the quotient an A-homomorphism graded of degree 0

αMx:(A[X1,...,Xn]/d)⊗AM→⨁r≥0xrM. \alpha_M^x : (A[X_1, ..., X_n]/\mathfrak{d}) \otimes_A M \to \bigoplus_{r \geq 0} x^r M .

By tensor product with A/xA/x, we deduce from αMx\alpha_M^x an A-homomorphism graded of degree 0

βMx:(A/x)[X1,...,Xn]⊗AM→⨁r≥0(xrM/xr+1M). \beta_M^x : (A/x)[X_1, ..., X_n] \otimes_A M \to \bigoplus_{r \geq 0} (x^r M/x^{r+1} M) .

The homomorphisms aMx,αMxa_M^x, \alpha_M^x and βMx\beta_M^x are surjective.

Theorem 1

Consider the following conditions: (i) The sequence xx is M-regular (X, p. 158). (ii) The sequence xx is completely secant for M (X, p. 157, def. 2). (iii) One has H1(x,M)=0H_1(x, M) = 0. (iv) The homomorphism αMx:(A[X1,...,Xn]/d)⊗AM→⨁r≥0xrM\alpha_M^x : (A[X_1, ..., X_n]/\mathfrak{d}) \otimes_A M \to \bigoplus_{r \geq 0} x^r M is bijective. (v) The homomorphism βMx:(A/x)[X1,...,Xn]⊗AM→⨁r≥0(xrM/xr+1M)\beta_M^x : (A/x)[X_1, ..., X_n] \otimes_A M \to \bigoplus_{r \geq 0} (x^r M/x^{r+1} M) is bijective.

Then one has the implications (i) ⇒\Rightarrow (ii) ⇒\Rightarrow (iii) ⇔\Leftrightarrow (iv) ⇒\Rightarrow (v). If for 1≤i≤n1 \leq i \leq n the A-module M/(x1M+⋯+xi−1M)M/(x_1 M + \cdots + x_{i-1} M) is separated for the xx-adic topology, the conditions (i) to (v) are equivalent.

The theorem will be proved in Nos 8 and 9.

*COROLLARY 1. — If A is noetherian, if the A-module M is of finite type and if the xix_i belong to the radical of A, the conditions (i) to (v) of the theorem are equivalent. Indeed, on each of the modules M/(x1M+⋯+xi−1M)M/(x_1 M + \cdots + x_{i-1} M) the xx-adic topology is separated (AC III, § 3, No. 3, prop. 6).

Corollary 2

Suppose that A is a graded ring with positive degrees, M a bounded below graded A-module, and the xix_i homogeneous elements of degree > 0 of A. Then the conditions (i) to (v) of the theorem are equivalent. The xx-adic topology is indeed separated for every bounded below graded A-module N, since if Nn=0N_n = 0 for n<n0n < n_0 one has xaN⊂∑j≥n0+aNjx^a N \subset \sum_{j \geq n_0 + a} N_j for all a≥0a \geq 0.

Corollary 3

Suppose that the modules M/(x1M+⋯+xi−1M)M/(x_1M+\cdots+x_{i-1}M) are separated for the x\mathfrak{x}-adic topology (1≤i≤n)(1\leq i\leq n) ; let p be an integer between 11 and nn. In order that the sequence x\mathfrak{x} be completely secant for MM, it is necessary and sufficient that the sequence (x1,…,xp)(x_1,\ldots,x_p) be completely secant for MM and that the sequence (xp+1,…,xn)(x_{p+1},\ldots,x_n) be completely secant for M/(x1M+⋯+xpM)M/(x_1M+\cdots+x_pM).

Indeed the corollary is evident if in the statement one replaces “completely secant sequences” by “regular sequences”; but the two notions coincide here by the theorem.

Remark

Let x=(x1,…,xn)\mathfrak{x}=(x_1,\ldots,x_n) be a sequence of elements of AA such that H1(x,A)=0H_1(\mathfrak{x},A)=0; then the kernel of the surjective homomorphism u:An→xu:A^n\to\mathfrak{x} such that

u(∑aiei)=∑aixi u\left(\sum a_i e_i\right)=\sum a_i x_i
is generated by the elements Xjei−XiejX_j e_i-X_i e_j; consequently, the AA-algebra A[X1,…,Xn]/dA[X_1,\ldots,X_n]/\mathfrak{d} is isomorphic to the symmetric algebra SA(x)S_A(\mathfrak{x}) (III, p. 69, prop. 4). It therefore follows from Theorem 1 that the algebra homomorphism
SA(x)⟶⨁nxn S_A(\mathfrak{x})\longrightarrow\bigoplus_n \mathfrak{x}^n
induced by the canonical injection of x\mathfrak{x} into ⨁nxn\bigoplus_n\mathfrak{x}^n is an isomorphism. The same holds for the homomorphism
SA(x/x2)⟶⨁nxn/xn+1. S_A(\mathfrak{x}/\mathfrak{x}^2)\longrightarrow\bigoplus_n\mathfrak{x}^n/\mathfrak{x}^{n+1}.

8. Proof of Theorem 1: first part

The implication (i) ⇒\Rightarrow (ii) has already been proved (X, p. 157, prop. 5). The implication (ii) ⇒\Rightarrow (iii) is obvious; the same is true of (iv) ⇒\Rightarrow (v), since βMx\beta_M^x is identified with αMx⊗1A/x\alpha_M^x\otimes 1_{A/\mathfrak{x}}.

To show that (iv) implies (iii), consider the homomorphism (αMx)1(\alpha_M^x)_1 induced on the components of degree 1. Let EE denote the graded AA-module A[X1,…,Xn]A[X_1,\ldots,X_n]; the AA-module E1E_1 is free with basis X1,…,XnX_1,\ldots,X_n and d1\mathfrak{d}_1 is the sub-AA-module of E1E_1 generated by the elements (xiXj−xjXi)(x_iX_j-x_jX_i) for 1≤i<j≤n1\leq i<j\leq n. Consequently ((E/d)⊗AM)1((E/\mathfrak{d})\otimes_A M)_1 is identified with K1(x,M)/B1(K∙(x,M))K_1(\mathfrak{x},M)/B_1(K_\bullet(\mathfrak{x},M)), the homomorphism (αMx)1(\alpha_M^x)_1 being identified with the mapping of K1(x,M)/B1(K∙(x,M))K_1(\mathfrak{x},M)/B_1(K_\bullet(\mathfrak{x},M)) into B0(K∙(x,M))B_0(K_\bullet(\mathfrak{x},M)) induced by d1d_1. Thus the vanishing of H1(x,M)H_1(\mathfrak{x},M) is equivalent to the injectivity of (αMx)1(\alpha_M^x)_1, whence the implication (iv) ⇒\Rightarrow (iii).

To prove that (iii) implies (iv), we shall use the following lemma:

Lemma 4

Let A0A_0 be the ring Z[T1,…,Tn]\mathbf{Z}[T_1,\ldots,T_n], and let u:A0[X1,…,Xn]→A0[U]u:A_0[X_1,\ldots,X_n]\to A_0[U] be the homomorphism of A0A_0-algebras such that u(Xi)=TiUu(X_i)=T_iU. The kernel of uu is the ideal d0\mathfrak{d}_0 of A0[X1,…,Xn]A_0[X_1,\ldots,X_n] generated by the elements (TiXj−TjXi)(T_iX_j-T_jX_i) for 1≤i<j≤n1\leq i<j\leq n. If tt is the ideal of A0A_0 generated by (T1,…,Tn)(T_1,\ldots,T_n), uu induces an isomorphism

uˉ:A0[X1,…,Xn]/d0⟶⨁r≥0tr. \bar{u}:A_0[X_1,\ldots,X_n]/\mathfrak{d}_0\longrightarrow\bigoplus_{r\geq0}t^r.

It is obviously enough to prove the first assertion. For every sequence of natural integers α=(α1,…,αn)\alpha=(\alpha_1,\ldots,\alpha_n) and every integer k∈[0,n]k\in[0,n], let Pα,kP_{\alpha,k} denote the monomial

T1α1⋯TkαkXk+1αk+1⋯Xnαn; T_1^{\alpha_1}\cdots T_k^{\alpha_k}X_{k+1}^{\alpha_{k+1}}\cdots X_n^{\alpha_n};

let NN be the sub-Z\mathbf{Z}-module of A0[X1,...,Xn]A_0[X_1, ..., X_n] generated by the Pα,kP_{\alpha, k} for α∈Nn\alpha \in \mathbf{N}^n and 0≤k≤n0 \leq k \leq n. We shall prove that the restriction of uu to NN is injective and that A0[X1,...,Xn]=d0+NA_0[X_1, ..., X_n] = \mathfrak{d}_0 + N; since we have d0⊂Ker u\mathfrak{d}_0 \subset \mathrm{Ker}\,u, this will entail the lemma.

Observe that NN is generated by the set SS formed by P0,0=1P_{0,0} = 1 and by those of the Pα,kP_{\alpha, k} for which αk≠0\alpha_k \neq 0. To prove the injectivity of the restriction of uu to NN, it suffices to show that two distinct elements of SS have as their images by uu distinct monomials in A0[U]A_0[U]. Now we have u(Pα,k)=TαUi≥k∑αiu(P_{\alpha, k}) = T^\alpha U^{\sum \alpha_i}_{i \geq k}, so that the equality u(Pα,k)=u(Pα′,k′)u(P_{\alpha, k}) = u(P_{\alpha', k'}) entails α=α′\alpha = \alpha' and ∑i≥kαi=∑i≥k′αi\sum_{i \geq k} \alpha_i = \sum_{i \geq k'} \alpha_i. Suppose that Pα,kP_{\alpha, k} and Pα′,k′P_{\alpha', k'} belong to SS. If α=0\alpha = 0, then we have k=k′=0k = k' = 0; if α≠0\alpha \neq 0, we obtain k=k′k = k' since αk\alpha_k and αk′\alpha_{k'} are different from zero, whence the result.

Show that every monomial TαXβ∈A0[X1,...,Xn]T^\alpha X^\beta \in A_0[X_1, ..., X_n] is congruent modulo d0\mathfrak{d}_0 to a Pη,kP_{\eta, k}. For every sequence λ∈Nn\lambda \in \mathbf{N}^n, we shall denote by i(λ)i(\lambda) (resp. j(λ)j(\lambda)) the smallest (resp. the largest) integer k∈[1,n]k \in [1, n] such that λk≠0\lambda_k \neq 0. Among the monomials TγXδT^\gamma X^\delta which are congruent to TαXβT^\alpha X^\beta modulo d0\mathfrak{d}_0, choose one for which the rational integer j(γ)−i(δ)j(\gamma) - i(\delta) is minimal; let us show that we then have j(γ)−i(δ)<0j(\gamma) - i(\delta) < 0. Suppose that we have j(γ)≥i(δ)j(\gamma) \geq i(\delta); put j=j(γ),i=i(δ)j = j(\gamma), i = i(\delta), and let ε=inf⁡(γj,δi)\varepsilon = \inf (\gamma_j, \delta_i). Since (TjεXiε−TiεXjε)(T_j^\varepsilon X_i^\varepsilon - T_i^\varepsilon X_j^\varepsilon) is divisible by (TjXi−TiXj)(T_j X_i - T_i X_j), and therefore belongs to d0\mathfrak{d}_0, we see that TγXδT^\gamma X^\delta is congruent modulo d0\mathfrak{d}_0 to Tγ′Xδ′T^{\gamma'} X^{\delta'}, where γi′=γi+ε,γj′=γj−ε,γk′=γk\gamma'_i = \gamma_i + \varepsilon, \gamma'_j = \gamma_j - \varepsilon, \gamma'_k = \gamma_k for k≠i,jk \neq i, j, and δi′=δi−ε,δj′=δj+ε,δk′=δk\delta'_i = \delta_i - \varepsilon, \delta'_j = \delta_j + \varepsilon, \delta'_k = \delta_k for k≠i,jk \neq i, j. Since γj′\gamma'_j or δi′\delta'_i is zero, we have j(γ′)−i(δ′)<j(γ)−i(δ)j(\gamma') - i(\delta') < j(\gamma) - i(\delta) which contradicts the minimal character of j(γ)−i(δ)j(\gamma) - i(\delta).

Consequently we have j(γ)<i(δ)j(\gamma) < i(\delta), whence TγXδ∈NT^\gamma X^\delta \in N, which completes the proof of the lemma.

Let us prove that (iii) implies (iv). Consider the ring A0=Z[T1,...,Tn]A_0 = \mathbf{Z}[T_1, ..., T_n] and the ideal t of A0A_0 generated by T1,...,TnT_1, ..., T_n. Endow MM with the structure of A0A_0-module for which Tim=ximT_i m = x_i m for m∈M,1≤i≤nm \in M, 1 \leq i \leq n. According to X, p. 155, Hi(x,M)H_i(x, M) is identified canonically with Hi(T,M)H_i(T, M).

Since the sequence TT is regular for A0A_0, it follows from the implication (i) ⇒\Rightarrow (ii) that the complex K.(T,A0)K.(T, A_0) defines a free resolution of the A0A_0-module A0/tA_0/t. Let us remark that the latter is identified with Z\mathbf{Z}, endowed with the structure of A0A_0-module such that TiZ=0T_i \mathbf{Z} = 0 for 1≤i≤n1 \leq i \leq n. Thus the condition (iii) is equivalent to Tor1A0(M,Z)=0\mathrm{Tor}_1^{A_0}(M, \mathbf{Z}) = 0.

Let us show that (iii) implies Tor1A0(M,A0/tr)=0\mathrm{Tor}_1^{A_0}(M, A_0/t^r) = 0 for all r≥1r \geq 1. This results from what precedes for r=1r = 1. In the general case, consider the exact sequence

0→tr/tr+1→A0/tr+1→A0/tr→0, 0 \to t^r/t^{r+1} \to A_0/t^{r+1} \to A_0/t^r \to 0,

The A0A_0-module tr/tr+1t^r/t^{r+1} is isomorphic to a finite product of copies of Z\mathbf{Z}; we therefore have Tor1A0(M,tr/tr+1)=0\mathrm{Tor}_1^{A_0}(M, t^r/t^{r+1}) = 0. It follows, by induction on rr, that

Tor1A0(M,A0/tr)=0for all r, \mathrm{Tor}_1^{A_0}(M, A_0/t^r) = 0 \quad \text{for all } r,

The exact sequence (16) therefore provides, by tensor product with MM, an exact sequence

0→(tr/tr+1)⊗A0M→M/xr+1M→M/xrM→0 0 \to (t^r/t^{r+1}) \otimes_{A_0} M \to M/x^{r+1} M \to M/x^r M \to 0

whence an isomorphism of (t′/t′r+1)⊗A0M(t'/{t'}^{r+1}) \otimes_{A_0} M onto x′M/x′r+1Mx' M/{x'}^{r+1} M. By considering the exact sequence 0→t′r+1→t′→t′/t′r+1→00 \to {t'}^{r+1} \to t' \to t'/{t'}^{r+1} \to 0, one then sees by induction on rr that the mapping mr:t′⊗A0M→x′Mm_r : t' \otimes_{A_0} M \to x' M, induced by the operation of A0A_0 in MM, is an isomorphism.

To prove (iv), it remains to observe that the diagram

(A0[X1,...,Xn]/D0)⊗A0M→uˉ⊗1M⨁r≥0(t′⊗A0M)↓e↓⊕mr(A[X1,...,Xn]/D)⊗AM→αMx⨁r≥0x′M \begin{array}{ccc} (A_0[X_1, ..., X_n]/\mathscr{D}_0) \otimes_{A_0} M & \xrightarrow{\bar{u} \otimes 1_M} & \bigoplus_{r \geq 0} (t' \otimes_{A_0} M) \\ \downarrow e & & \downarrow \oplus m_r \\ (A[X_1, ..., X_n]/\mathscr{D}) \otimes_A M & \xrightarrow{\alpha_M^x} & \bigoplus_{r \geq 0} x' M \end{array}

where ee is the canonical isomorphism of extension of scalars (II, p. 85, prop. 6), is commutative, and to apply Lemma 4.

9. Proof of Theorem 1: second part

Let us again consider the exact sequence

(16)

0⟶t′/t′r+1→irA0/t′r+1→prA0/t′⟶0 0 \longrightarrow t'/{t'}^{r+1} \xrightarrow{i_r} A_0/{t'}^{r+1} \xrightarrow{p_r} A_0/t' \longrightarrow 0

and the associated exact sequence of torsion modules

(18)

Tor1A0(A0/t′r+1,M)⟶Tor1A0(A0/t′,M)⟶(t′/t′r+1)⊗A0M→ir⊗1M(A0/t′r+1)⊗A0M→pr⊗1M(A0/t′)⊗A0M⟶0. \begin{array}{cccccc} \mathrm{Tor}_1^{A_0}(A_0/{t'}^{r+1}, M) & \longrightarrow & \mathrm{Tor}_1^{A_0}(A_0/t', M) \\ \longrightarrow (t'/{t'}^{r+1}) \otimes_{A_0} M & \xrightarrow{i_r \otimes 1_M} & (A_0/{t'}^{r+1}) \otimes_{A_0} M & \xrightarrow{p_r \otimes 1_M} & (A_0/t') \otimes_{A_0} M & \longrightarrow 0 . \end{array}

The kernel of pr⊗1Mp_r \otimes 1_M is identified with x′M/x′r+1Mx' M/{x'}^{r+1} M; moreover t′/t′r+1t'/{t'}^{r+1} is annihilated by the TiT_i and is identified with the homogeneous component of degree rr of A0A_0, so that the homomorphism (t′/t′r+1)⊗A0M→x′M/x′r+1M(t'/{t'}^{r+1}) \otimes_{A_0} M \to x' M/{x'}^{r+1} M deduced from ir⊗1Mi_r \otimes 1_M is identified with the homogeneous component of degree rr of the homomorphism βMx\beta_M^x. It follows therefore from the exact sequence (18) that condition (v) is equivalent to

(v') : the homomorphism Tor1A0(pr,1M)\mathrm{Tor}_1^{A_0}(p_r, 1_M) : Tor1A0(A0/t′r+1,M)→Tor1A0(A0/t′,M)\mathrm{Tor}_1^{A_0}(A_0/{t'}^{r+1}, M) \to \mathrm{Tor}_1^{A_0}(A_0/t', M) is surjective for all r≥1r \geq 1.

It remains for us to prove the implication (v) ⇒\Rightarrow (i) when the modules

M/(x1M+⋯+xi−1M) M/(x_1 M + \cdots + x_{i-1} M)
are separated for the xx-adic topology (1≤i≤n1 \leq i \leq n). Let M‾\overline{M} denote the AA-module M/x1MM/x_1 M. By definition, the sequence xx is regular for MM if and only if (x1)M(x_1)_M is injective and the sequence x′=(x2,...,xn)x' = (x_2, ..., x_n) is regular for M‾\overline{M}. Reasoning by induction on nn, it therefore suffices to prove that, if MM is separated for the xx-adic topology and if βMx\beta_M^x is bijective, then (x1)M(x_1)_M is injective and βMx′\beta_M^{x'} is bijective. Now the bijectivity of βMx\beta_M^x implies in particular that the homothety with ratio x1x_1 in ⨁rx′M/x′r+1M\bigoplus r x' M/{x'}^{r+1} M is injective, hence that Ker (x1)M⊂⋂ixiM\mathrm{Ker}\,(x_1)_M \subset \bigcap_i x^i M and consequently that (x1)M(x_1)_M is injective if the xx-adic topology on MM is separated.

We are thus reduced to proving that if (x1)M(x_1)_M is injective and if MM satisfies condition (v'), then M‾\overline{M} satisfies condition (v') relative to the sequence x′x'.

By assumption, we have an exact sequence

0⟶M→(x1)MM⟶M‾⟶0; 0 \longrightarrow M \xrightarrow{(x_1)_M} M \longrightarrow \overline{M} \longrightarrow 0;
let A‾0=A0/T1A0, t‾=tA‾0\overline{A}_0 = A_0 / T_1 A_0,\ \overline{t} = t \overline{A}_0. Let q:L→Mq : L \to M be a free resolution of the A0A_0-module MM; since the homothety with ratio T1T_1 is injective in A0A_0, it is injective in LL, and we have an exact sequence of complexes
0⟶L→(x1)LL⟶L‾⟶0 0 \longrightarrow L \xrightarrow{(x_1)_L} L \longrightarrow \overline{L} \longrightarrow 0
with L‾=L/x1L\overline{L} = L / x_1 L, and a commutative diagram
0⟶L→(x1)LL⟶L‾⟶0↓q↓q↓q‾0⟶M→(x1)MM⟶M‾⟶0. \begin{array}{ccc} 0 & \longrightarrow & L \xrightarrow{(x_1)_L} L \longrightarrow \overline{L} \longrightarrow 0 \\ & & \downarrow q \downarrow q \downarrow \overline{q} \\ 0 & \longrightarrow & M \xrightarrow{(x_1)_M} M \longrightarrow \overline{M} \longrightarrow 0 . \end{array}

Since qq is a homomorphism of complexes inducing isomorphisms in homology, q‾:L‾→M‾\overline{q} : \overline{L} \to \overline{M} is a free resolution of the A‾0\overline{A}_0-module M‾\overline{M} (cor. 1, p. 30). For every A‾0\overline{A}_0-module PP, there is a canonical isomorphism

P⊗A0L→P⊗A‾0L‾, P \otimes_{A_0} L \to P \otimes_{\overline{A}_0} \overline{L},
whence by passing to homology an isomorphism
φP:Tor1A0(P,M)→Tor1A‾0(P,M‾). \varphi_P : \mathrm{Tor}_1^{A_0}(P, M) \to \mathrm{Tor}_1^{\overline{A}_0}(P, \overline{M}) .
If u:P→P′u : P \to P' is a A‾\overline{A}-homomorphism, then
φP′∘Tor1A0(u,1M)=Tor1A‾0(u,1M)∘φP. \varphi_{P'} \circ \mathrm{Tor}_1^{A_0}(u, 1_M) = \mathrm{Tor}_1^{\overline{A}_0}(u, 1_M) \circ \varphi_P .
That being so, suppose that condition (v') is satisfied for MM, and let us prove that it is true for M‾\overline{M}. Let rr be an integer ⩾1\geqslant 1; we have a commutative diagram with exact rows
0⟶A0/tr→T1A0/tr+1⟶A‾0/t‾r+1⟶0↓pr−1↓pr↓p‾r0⟶A0/tr−1→T1A0/tr⟶A‾0/t‾r⟶0 \begin{array}{ccccccccc} 0 & \longrightarrow & A_0 / t^r & \xrightarrow{T_1} & A_0 / t^{r+1} & \longrightarrow & \overline{A}_0 / \overline{t}^{r+1} & \longrightarrow & 0 \\ & & \downarrow p_{r-1} & & \downarrow p_r & & \downarrow \overline{p}_r & & \\ 0 & \longrightarrow & A_0 / t^{r-1} & \xrightarrow{T_1} & A_0 / t^r & \longrightarrow & \overline{A}_0 / \overline{t}^r & \longrightarrow & 0 \end{array}
from which one deduces a commutative diagram with exact rows
Tor1A0(A0/tr+1,M)⟶Tor1A0(A‾0/t‾r+1,M)⟶M/xrM→x1M/xr+1MTor1(pr,r′)↓Tor1(p‾r,1)↓↓Tor1(A0/tr,M)⟶Tor1A0(A‾0/t‾r,M)⟶M/xr−1M→x1M/xrM. \begin{array}{ccccccccccc} \mathrm{Tor}_1^{A_0}(A_0 / t^{r+1}, M) & \longrightarrow & \mathrm{Tor}_1^{A_0}(\overline{A}_0 / \overline{t}^{r+1}, M) & \longrightarrow & M / x^r M & \xrightarrow{x_1} & M / x^{r+1} M \\ \mathrm{Tor}_1(p_{r, r'}) & \downarrow & \mathrm{Tor}_1(\overline{p}_{r, 1}) & \downarrow & & & \downarrow \\ \mathrm{Tor}_1(A_0 / t^r, M) & \longrightarrow & \mathrm{Tor}_1^{A_0}(\overline{A}_0 / \overline{t}^r, M) & \longrightarrow & M / x^{r-1} M & \xrightarrow{x_1} & M / x^r M . \end{array}
Now multiplication by x1x_1 defines an injection of M/xrMM / x^r M into M/xr+1MM / x^{r+1} M: this follows immediately, by induction on rr, from the exact sequence
0→xr−1M/xrM→M/xrM→M/xr−1M→0 0 \to x^{r-1} M / x^r M \to M / x^r M \to M / x^{r-1} M \to 0

and from the injectivity of the homothety with ratio x1x_1 in ⨁r≥0(xrM/xr+1M)\bigoplus_{r \geq 0} (x^r M / x^{r+1} M). Condition (v) therefore implies that the homomorphism Tor⁡1A‾0(p‾r,1M)\operatorname{Tor}_{1}^{\overline{A}_0} (\overline{p}_r, 1_M) is surjective for every r≥1r \geq 1. By composition with the isomorphisms (φA‾0/t‾r+1)−1(\varphi_{\overline{A}_0/\overline{t}^{r+1}})^{-1} and φA‾0/t‾r\varphi_{\overline{A}_0/\overline{t}^r}, one deduces that the homomorphism

Tor⁡1A‾0(p‾r,1M):Tor⁡1A‾0(A‾0/t‾r+1,M)→Tor⁡1A‾0(A‾0/t‾r,M) \operatorname{Tor}_{1}^{\overline{A}_0} (\overline{p}_r, 1_M) : \operatorname{Tor}_{1}^{\overline{A}_0} (\overline{A}_0 / \overline{t}^{r+1}, M) \to \operatorname{Tor}_{1}^{\overline{A}_0} (\overline{A}_0 / \overline{t}^r, M)

is surjective for r≥1r \geq 1, whence condition (v') for M‾\overline{M}. This completes the proof of the theorem.

10. Class of extensions associated with a regular sequence

Let AA be a ring, MM an AA-module, x=(x1,...,xn)x = (x_1, ..., x_n) a sequence of elements of AA. Let MiM_i denote the AA-module M/(x1M+⋯+xi−1M)M / (x_1 M + \cdots + x_{i-1} M) for i=0,...,n+1i = 0, ..., n+1, so that M0M_0 and M1M_1 are identified with MM and that Mn+1=M/(x)MM_{n+1} = M / (x) M. Let

x‾i:Mi−1→Mi,i=1,...,n, \overline{x}_i : M_{i-1} \to M_i , \quad i = 1, ..., n ,

the AA-homomorphism composed of the homothety of Mi−1M_{i-1} with ratio xix_i and of the canonical projection of Mi−1M_{i-1} onto MiM_i. Finally, let us denote by p:Mn→M/(x)Mp : M_n \to M / (x) M the canonical projection. The diagram

(19)

0⟶M→x‾1M1→x‾2M2⟶⋯→x‾nMn→pM/(x)M⟶0 0 \longrightarrow M \xrightarrow{\overline{x}_1} M_1 \xrightarrow{\overline{x}_2} M_2 \longrightarrow \cdots \xrightarrow{\overline{x}_n} M_n \xrightarrow{p} M / (x) M \longrightarrow 0

is an exact sequence if and only if the sequence xx is MM-regular. Henceforth suppose that the sequence xx is regular for MM. The element θx∈Ext⁡An(M/(x)M,M)\theta_x \in \operatorname{Ext}_A^n (M / (x) M, M) associated with the exact sequence (19) is also said to be associated with the MM-regular sequence xx.

Let ii be an integer, 1≤i≤n1 \leq i \leq n. Note that the sequence (19) can be decomposed into the two exact sequences

(20)

0⟶M→x‾1M1→x‾2M2⟶⋯→x‾iMi⟶M/(x1M+⋯+xiM)⟶0 0 \longrightarrow M \xrightarrow{\overline{x}_1} M_1 \xrightarrow{\overline{x}_2} M_2 \longrightarrow \cdots \xrightarrow{\overline{x}_i} M_i \longrightarrow M / (x_1 M + \cdots + x_i M) \longrightarrow 0

(21)

0⟶M/(x1M+⋯+xiM)→x‾i+1Mi+1⟶⋯⟶Mn→pM/(x)M⟶0 0 \longrightarrow M / (x_1 M + \cdots + x_i M) \xrightarrow{\overline{x}_{i+1}} M_{i+1} \longrightarrow \cdots \longrightarrow M_n \xrightarrow{p} M / (x) M \longrightarrow 0

which are none other than the exact sequences associated with the sequence (x1,...,xi)(x_1, ..., x_i), which is regular for MM, and with the sequence (xi+1,...,xn)(x_{i+1}, ..., x_n), which is regular for M/(x1M+⋯+xiM)M / (x_1 M + \cdots + x_i M). Denoting by

θ(x1,...,xi)∈Ext⁡Ai(M/(x1M+⋯+xiM),M) \theta_{(x_1, ..., x_i)} \in \operatorname{Ext}_A^i (M / (x_1 M + \cdots + x_i M), M)
θ(xi+1,...,xn)∈Ext⁡An−i(M/(x)M,M/(x1M+⋯+xiM)), \theta_{(x_{i+1}, ..., x_n)} \in \operatorname{Ext}_A^{n-i} (M / (x) M, M / (x_1 M + \cdots + x_i M)) ,

the extension classes associated with (20) and (21), one has, by X, p. 118, prop. 3,

(22)

θ(x1,…,xn)=θ(x1,…,xi)∘θ(xi+1,…,xn). \theta_{(x_1, \ldots, x_n)} = \theta_{(x_1, \ldots, x_i)} \circ \theta_{(x_{i+1}, \ldots, x_n)} .

Moreover, by prop. 5 (X, p. 157), the Koszul complex K.(x,M)\mathbf{K}.(x, M) is acyclic in degrees ≠n\neq n, whence an exact sequence

(23)

0⟶M→∂0K1(x,M)→∂1K2(x,M)→∂2⋯⟶Kn(x,M)→qM/(x)M⟶0, 0 \longrightarrow M \xrightarrow{\partial^0} \mathbf{K}^1(x, M) \xrightarrow{\partial^1} \mathbf{K}^2(x, M) \xrightarrow{\partial^2} \cdots \longrightarrow \mathbf{K}^n(x, M) \xrightarrow{q} M/(x)M \longrightarrow 0 ,
where K0(x,M)\mathbf{K}^0(x, M) has been identified with MM and where qq maps each element m∈Kn(x,M)m \in \mathbf{K}^n(x, M) to the class in M/(x)MM/(x)M of m(1,2,…,n)∈Mm(1, 2, \ldots, n) \in M.

Proposition 8

Suppose the sequence x is regular for M. The element of ExtAn(M/(x)M,M)\mathrm{Ext}_A^n(M/(x)M, M) associated with the exact sequence (23) is (−1)n(n+1)/2θx(-1)^{n(n+1)/2} \theta_x.* For i=0,1,…,ni = 0, 1, \ldots, n, define an A-linear mapping

ai:Ki(x,M)→Mi=M/(x1M+⋯+xi−1M) a^i : \mathbf{K}^i(x, M) \to M_i = M/(x_1 M + \cdots + x_{i-1} M)
as follows: if m∈Ki(x,M)m \in \mathbf{K}^i(x, M), ai(m)a^i(m) is the class in MiM_i of the element m(1,2,…,i)m(1, 2, \ldots, i) of MM. It is clear that a0a^0 is the identity mapping of MM and that p∘an=qp \circ a^n = q. Moreover ai+1∘∂i(m)a^{i+1} \circ \partial^i(m) is the image in Mi+1M_{i+1} of the element
∑k=1i+1(−1)k+1xkm(1,2,…,k−1,k+1,…,i+1). \sum_{k=1}^{i+1} (-1)^{k+1} x_k m(1, 2, \ldots, k-1, k+1, \ldots, i+1) .
Since xkx_k annihilates Mi+1M_{i+1} for k=1,…,ik = 1, \ldots, i, ai+1∘∂i(m)a^{i+1} \circ \partial^i(m) is also the image of
(−1)ixi+1m(1,2,…,i), (-1)^i x_{i+1} m(1, 2, \ldots, i) ,
hence
ai+1∘∂i=(−1)ix‾i+1∘ai. a^{i+1} \circ \partial^i = (-1)^i \overline{x}_{i+1} \circ a^i .
By X, p. 120, cor. 1 and 2, the element of ExtAn(M/(x)M,M)\mathrm{Ext}_A^n(M/(x)M, M) associated with (23) is equal to ∏i=1n(−1)i⋅θx\prod_{i=1}^n (-1)^i \cdot \theta_x, whence the assertion.

Corollary

Suppose moreover that the modules M/(x1M+⋯+xi−1M)M/(x_1 M + \cdots + x_{i-1} M) are separated for the (x)-adic topology, and let (aij)∈GLn(A)(a_{ij}) \in \mathrm{GL}_n(A). Put*

yi=∑jaijxjandy=(y1,…,yn). y_i = \sum_j a_{ij} x_j \quad \text{and} \quad y = (y_1, \ldots, y_n) .
Then the sequence y is regular for M, and one has θy=det⁡(aij)−1θx\theta_y = \det(a_{ij})^{-1} \theta_x. Indeed, the sequence y is regular for M by Prop. 6 (X, p. 158) and Theorem 1 (X, p. 160); the last assertion follows from Prop. 8, and from Prop. 4 of X, p. 119.

Proposition 9

Suppose the sequence x regular for M. If N is an A-module such that (x)N=0(x)N = 0, one has ExtAi(N,M)=0\mathrm{Ext}_A^i(N, M) = 0 for i<ni < n, and the mapping α↦θx∘α\alpha \mapsto \theta_x \circ \alpha* from Hom_A (N, M/(x) M) into Ext^n_A (N, M) (which is also the iterated connecting homomorphism associated with (19), cf. X, p. 127, Corollary 3) is bijective.

It remains to prove that the homomorphism ψi:α↦θx∘α\psi^i : \alpha \mapsto \theta_x \circ \alpha from ExtAi−n(N,M/(x)M)\mathrm{Ext}_A^{i-n}(N, M/(x) M) into ExtAi(N,M)\mathrm{Ext}_A^i(N, M) is bijective for i≤ni \leq n. Let us reason by induction on nn, the assertion being trivial for n=0n = 0. Put M1=M/x1MM_1 = M/x_1 M, x′=(x2,...,xn)x' = (x_2, ..., x_n), so that x′x' is M1M_1-regular. By the induction hypothesis, the homomorphism

ψ‾i−1:α↦θx′∘α \overline{\psi}^{i-1} : \alpha \mapsto \theta_{x'} \circ \alpha

from ExtAi−n(N,M/(x)M)\mathrm{Ext}_A^{i-n}(N, M/(x) M) into ExtAi−1(N,M1)\mathrm{Ext}_A^{i-1}(N, M_1) is bijective for i<ni < n. On the other hand, consider the exact sequence

0⟶M→(x1)MM⟶M1⟶0; 0 \longrightarrow M \xrightarrow{(x_1)_M} M \longrightarrow M_1 \longrightarrow 0 ;

the connecting homomorphism ExtAi(N,M1)→ExtAi+1(N,M)\mathrm{Ext}_A^i(N, M_1) \to \mathrm{Ext}_A^{i+1}(N, M) associated with it is φi:β↦θx1∘β\varphi^i : \beta \mapsto \theta_{x_1} \circ \beta (X, p. 125, Prop. 5) ;

since one has Exti(1N,(x1)M)=Exti((x1)N,1M)=0\mathrm{Ext}^i(1_N, (x_1)_M) = \mathrm{Ext}^i((x_1)_N, 1_M) = 0, one deduces from this exact sequences

0⟶ExtAi(N,M)⟶ExtAi(N,M1)→φiExtAi+1(N,M)⟶0. 0 \longrightarrow \mathrm{Ext}_A^i(N, M) \longrightarrow \mathrm{Ext}_A^i(N, M_1) \xrightarrow{\varphi^i} \mathrm{Ext}_A^{i+1}(N, M) \longrightarrow 0 .

Since ExtAi(N,M1)=0\mathrm{Ext}_A^i(N, M_1) = 0 for i<n−1i < n - 1, it follows that ExtAi+1(N,M)=0\mathrm{Ext}_A^{i+1}(N, M) = 0 for i<n−1i < n - 1, that is to say for i+1<ni + 1 < n; it follows that φi\varphi^i is bijective for i<ni < n. Since ψi(α)=θx∘α=θx1∘θx′∘α=φi−1∘ψ‾i(α)\psi^i(\alpha) = \theta_x \circ \alpha = \theta_{x_1} \circ \theta_{x'} \circ \alpha = \varphi^{i-1} \circ \overline{\psi}^i(\alpha) for α∈ExtAi−n(N,M/(x)M)\alpha \in \mathrm{Ext}_A^{i-n}(N, M/(x) M), ψi\psi^i is indeed bijective for i≤ni \leq n.

EXERCISES

Exercises

See the exercises for § 9.

Machine translation of the English, by gpt-5-6, gpt-5.4, gpt-5-6-mini, against glossary version 34. Not checked by a person. That model is a cut down version of the one the rest of this language was written by.