Bourbaki

Theorem 1

Tag 051O, permanent. Cite it as [Bourbaki, Tag 051O].

Commutative Algebra, chapter VII, § 3. Factorial domains

Let AA be an integral domain. The following conditions are equivalent: (a) AA is factorial; (b) the ordered group of non-zero fractional principal ideals of AA is a direct sum of groups isomorphic to Z\mathbf{Z} (ordered by the product order); (c) condition (M) is satisfied and the intersection of two principal ideals of AA is a principal ideal; (d) condition (M) is satisfied and, for every extremal element pp of AA, the ideal ApAp is prime; (e) AA is a Krull domain and every prime ideal of height 1 is principal.

We shall denote by K the field of fractions of A and by P∗\mathscr{P}^* (or P∗(A)\mathscr{P}^*(A)) the ordered group of non-zero fractional principal ideals of A. The proof will be carried out by proving the following implications:

(a) ⇔ (b) ⇔ (c) ⇔ (d) ⇔ (e)

We show that (a) implies (b); if A is factorial, P∗\mathscr{P}^* is isomorphic to the group of divisors of A and hence to a direct sum of groups Z\mathbf{Z} (§1§ 1, no. 3, Theorem 2).

Note now that the relation "the intersection of two integral principal ideals of A is a principal ideal" means that every ordered pair of elements of A admits a lcmlcm, that is that P∗\mathscr{P}^* is a lattice-ordered group (Algebra, Chapter VI, §1§ 1, no. 9, Proposition 8). The fact that (b) implies (c) (and even is equivalent to it) therefore follows from Algebra, Chapter VI, §1§ 1, no. 13, Theorem 2. The fact that (c) implies (d) follows from Algebra, Chapter VI, §1§ 1, no. 13, Proposition 14 (DIV).

The fact that (d) implies (b) follows from Algebra, Chapter VI, §1§ 1, no. 13, Theorem 2 applied to the group S∗\mathscr{S}^*.

We show that (b) implies (e). If (b) holds, there is an isomorphism of P∗\mathscr{P}^* onto Z(l)\mathbf{Z}^{(l)}; let (vi(x))i∈I(v_i(x))_{i \in I} denote the element of Z(l)\mathbf{Z}^{(l)} corresponding to the ideal AxAx (x∈K∗x \in K^*). It is seen immediately that each viv_i is a discrete valuation on K, that A is the intersection of the rings of the viv_i and that, for x∈K∗x \in K^*, vi(x)=0v_i(x) = 0 except for a finite number of indices ii; hence A is a Krull domain. On the other hand, let q be a prime ideal of A of height 1; it contains a non-zero element a which is necessarily not invertible and hence also (by definition of a prime ideal) one of the extremal elements of A; as ApAp is prime and non zero, q=Apq = Ap, which proves that q is principal.

Finally we show that (e) implies (a). Let a be a divisorial ideal of A. There exist prime ideals pip_i of A of height 1 such that div⁡a=∑inidiv⁡pi\operatorname{div} a = \sum_i n_i \operatorname{div} p_i where ni∈Zn_i \in \mathbf{Z}. If (e) holds, pip_i is of the form ApiAp_i, whence div⁡a=div⁡(∏iApini)\operatorname{div} a = \operatorname{div} \left( \prod_i Ap_i^{n_i} \right) and hence a=∏iApinia = \prod_i Ap_i^{n_i} since a is divisorial.

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 volume. Not translated.