Let A be a Krull domain and B a Krull domain containing A. Suppose that the following condition holds:
(PDE) For every prime ideal P of B of height 1, the prime ideal P∩A is zero or of height 1.
For p∈P(A) the P∈P(B) such that P∩A=p are finite in number; we write
i(p)=P∈P(B),P∩A=p∑e(P/p)P,
where e(P/p) denotes the ramification index of vp over vp (Chapter VI, § 8, no. 1). Then i defines, by linearity, an increasing homomorphism i of D(A) to D(B), which enjoys the following properties:
(a) for every non-zero element x of the field of fractions of A,
i(divA(x))=divB(x);
(b) for all D, D' in D(A),
i(sup(D,D′))=sup(i(D),i(D′));
Let p∈P(A); consider a non-zero element a of p; the P∈P(B) which contain a are finite in number (no. 6, Theorem 4); a fortiori the P∈P(B) such that P∩A=p are finite in number.
We now show (a). By additivity, it may be assumed that x∈A∗=A−{0}.
By definition, divB(x)=∑P∈P(B)vp(x)⋅P. For all P∈P(B) such that vp(x)>0, P∩A is non-zero (for x∈P) and is therefore of height 1 by (PDE); setting p=P∩A, by definition of the ramification index, vp(x)=e(P/p)vp(x) (since vp and vP are normed). As divA(x)=∑p∈P(A)vp(x)⋅p, and i(q)=0 for all q∈P(A) which is not of the form Q∩A where Q∈P(B), we deduce (a).
To prove (b) we write
D=p∈P(A)∑n(p)⋅pandD′=p∈P(A)∑n′(p)⋅p;
the coefficient of p in sup(D,D′) is sup(n(p),n′(p)). Let P be an element of P(B). If P∩A=(0), the coefficients of P in i(D) and i(D′) and hence also in sup(i(D),i(D′)) are zero; therefore the coefficient of P in i(sup(D,D′)) is zero. If P∩A=(0), it is a prime ideal p of height 1 (by (PDE)); writing e=e(P/p), the coefficients of P in i(D), i(D′) and i(sup(D,D′)) are respectively en(p),en′(p) and e⋅sup(n(p),n′(p)); that of sup(i(D),i(D′)) is
sup(e⋅n(p),e⋅n′(p))=e⋅sup(n(p),n′(p)).
This proves (b).
Under the hypotheses of Proposition 14, it follows from (a) that i defines, by taking quotients, a homomorphism iˉ, called canonical, of C(A) to C(B), which we shall also sometimes write as i, by an abuse of notation.
The condition (PDE) is fulfilled in the two following cases:
(1) B is integral over A; in this case, for the prime ideal P of B to be of height 1, it is necessary and sufficient that p=P∩A be of height 1. For (0) is the only prime ideal of B lying above the ideal (0) of A (Chapter V, § 2, no. 1, Corollary I to Proposition 1); if P is of height 1, then p=0; if p were not of height 1, there would exist a prime ideal p′ of A distinct from (0) and p and such that 0 ⊂ p' ⊂ p; but then, as B is an integral domain and A is an integrally closed domain, there would be a prime ideal p′ of B such that p′∩A=p′ and p′⊂p (Chapter V, § 2, no. 4, Theorem 3), contrary to the hypothesis. Conversely, if p is of height 1, there can exist no prime ideal p′ of B distinct from 0 and p and such that 0⊂p′⊂p, otherwise 0⊂p′∩A⊂p and p′∩A would be distinct from 0 and p by virtue of Chapter V, § 2, no. 1, Corollary 1 to Proposition 1.
(2) B is aflat A-module. More precisely:
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.