Bourbaki

Definition 1

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

Algebra, chapter X, § 2. Complexes de A-modules

A differential complex of AA-modules is a pair (C,d)(C,d) formed by a graded AA-module CC and an endomorphism d:C→Cd:C\to C graded of descending degree −1-1 and such that d∘d=0d\circ d=0.

It is also called a complex of AA-modules, or an AA-complex, or a complex. One often writes C in place of (C, d); the endomorphism d is called the differential of the complex (C, d), or by abuse of language of C.

If C_n (resp. C^n) is the homogeneous component of descending degree (resp. ascending degree) n of C, the datum of d is equivalent to that of the sequence of homomorphisms

(1)

⋯⟶Cn+1→dn+1Cn→dnCn−1⟶⋯ \cdots \longrightarrow C_{n+1} \xrightarrow{d_{n+1}} C_n \xrightarrow{d_n} C_{n-1} \longrightarrow \cdots
resp.

(1')

⋯⟶Cn−1→dn−1Cn→dnCn+1⟶⋯ \cdots \longrightarrow C^{n-1} \xrightarrow{d^{n-1}} C^n \xrightarrow{d^n} C^{n+1} \longrightarrow \cdots
such that dn∘dn+1=0d_n \circ d_{n+1} = 0 for all n∈Zn \in \mathbf{Z} (resp. dn∘dn−1=0d^n \circ d^{n-1} = 0 for all n∈Zn \in \mathbf{Z}). By abuse of language, one will also call a complex the datum of such a sequence of AA-modules and homomorphisms.

One will remark, as a mnemonic device, that when one “follows the direction of the arrows” in the diagrams (1) and (1’), the descending degree decreases and the ascending degree increases.

Every graded AA-module will tacitly be considered as a complex by endowing it with the zero differential; the complexes thus obtained will be called complexes with zero differential. In particular, every AA-module M will be endowed with the unique structure of AA-complex such that M0=M0=MM_0 = M^0 = M. The complex (C, d) is said to be zero if C is reduced to 0. In the sequel, we denote by 0 a zero complex, chosen once and for all.

Adjoin to the ordered set Z\mathbf{Z} two elements denoted −∞-\infty and +∞+\infty; denote by Z‾\overline{\mathbf{Z}} the set obtained, and endow it with the order relation extending that of Z\mathbf{Z} and such that −∞<n<+∞-\infty < n < +\infty for all n∈Zn \in \mathbf{Z}; every subset of Z‾\overline{\mathbf{Z}} possesses a greatest lower bound and a greatest upper bound.

Let C be a complex; one calls the right and left bounds 1^1 of C the elements bd(C)b_d(C) and bg(C)b_g(C) of Z‾\overline{\mathbf{Z}} defined by

bd(C)=inf⁡{n∈Z,Cn≠0},bg(C)=sup⁡{n∈Z,Cn≠0}. b_d(C) = \inf \{ n \in \mathbf{Z}, C_n \neq 0 \}, \quad b_g(C) = \sup \{ n \in \mathbf{Z}, C_n \neq 0 \}.
One says that C is zero on the right if bd(C)⩾0b_d(C) \geqslant 0, bounded on the right if bd(C)≠−∞b_d(C) \neq -\infty, zero on the left if bg(C)⩽0b_g(C) \leqslant 0, bounded on the left if bg(C)≠+∞b_g(C) \neq +\infty; one says that C is bounded if
bd(C)≠−∞,bg(C)≠+∞. b_d(C) \neq -\infty, \quad b_g(C) \neq +\infty.
One calls the length 2^2 of C and denotes by l(C)l(C) the element of Z‾\overline{\mathbf{Z}} defined as follows: if C is zero, l(C)=−∞l(C) = -\infty; if C is bounded and nonzero l(C)=bg(C)−bd(C)l(C) = b_g(C) - b_d(C); if C

1^1 The words right and left are relative to the description of C by means of the diagrams (1) and (1’). 2^2 One must not confuse the notion of length of the complex (C, d) with that of length of the module C (II, p. 21).

is not bounded, l(C)=+∞l(C) = + \infty. \* With the conventions of TG, IV, p. 13-17, one always has l(C)=bg(C)−bd(C)l(C) = b_g(C) - b_d(C). \*

For example, if kk consecutive components of CC are non-zero, the others being zero, one has l(C)=k−1l(C) = k - 1 if k>0k > 0, l(C)=−∞l(C) = - \infty if k=0k = 0.

One says that the complex (C,d)(C, d) is free, projective, flat, injective, if each of the modules CnC_n is so. It will be noted that the complex (C,d)(C, d) is projective or flat if and only if the module CC is so (II, p. 39, prop. 3 and X, p. 8, prop. 4), but that CC can be free without the complex (C,d)(C, d) being so (since a direct factor of a free module is not always free), analogously (C,d)(C, d) can be injective without CC being so (X, p. 170, exercise 21).

Let (C,d)(C, d) be a complex. Put Z(C,d)=Ker (d)Z(C, d) = \mathrm{Ker}\,(d), B(C,d)=Im (d)B(C, d) = \mathrm{Im}\,(d); these are graded submodules of CC, called respectively the module of cycles and the module of boundaries of (C,d)(C, d); the homogeneous components of Z(C,d)Z(C, d) and B(C,d)B(C, d) are denoted Zn(C,d)=Z−n(C,d)Z_n(C, d) = Z^{-n}(C, d), Bn(C,d)=B−n(C,d)B_n(C, d) = B^{-n}(C, d); one has Zn(C,d)=Ker (dn)Z_n(C, d) = \mathrm{Ker}\,(d_n), Bn(C,d)=Im (dn+1)B_n(C, d) = \mathrm{Im}\,(d_{n+1}), Zn(C,d)=Ker (dn)Z^n(C, d) = \mathrm{Ker}\,(d^n), Bn(C,d)=Im (dn−1)B^n(C, d) = \mathrm{Im}\,(d^{n-1}).

Since d∘d=0d \circ d = 0, one has B(C)⊂Z(C)B(C) \subset Z(C); two cycles are said to be homologous if their difference is a boundary; the graded quotient module H(C,d)=Z(C,d)/B(C,d)H(C, d) = Z(C, d)/B(C, d) is called the homology module of (C,d)(C, d); its elements are the homology classes; its homogeneous components are denoted Hn(C,d)=H−n(C,d)H_n(C, d) = H^{-n}(C, d).

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.