Lemma 10.164.7. Let R \to S be a ring map. Assume that
R \to S is smooth and surjective on spectra, and
S is a Nagata ring.
Then R is a Nagata ring.
Lemma 10.164.7. Let R \to S be a ring map. Assume that
R \to S is smooth and surjective on spectra, and
S is a Nagata ring.
Then R is a Nagata ring.
Proof. Recall that a Nagata ring is the same thing as a Noetherian universally Japanese ring (Proposition 10.162.15). We have already seen that R is Noetherian in Lemma 10.164.1. Let R \to A be a finite type ring map into a domain. According to Lemma 10.162.3 it suffices to check that A is N-1. It is clear that B = A \otimes _ R S is a finite type S-algebra and hence Nagata (Proposition 10.162.15). Since A \to B is smooth (Lemma 10.137.4) we see that B is reduced (Lemma 10.163.7). Since B is Noetherian it has only a finite number of minimal primes \mathfrak q_1, \ldots , \mathfrak q_ t (see Lemma 10.31.6). As A \to B is flat each of these lies over (0) \subset A (by going down, see Lemma 10.39.19) The total ring of fractions Q(B) is the product of the L_ i = \kappa (\mathfrak q_ i) (Lemmas 10.25.4 and 10.25.1). Moreover, the integral closure B' of B in Q(B) is the product of the integral closures B_ i' of the B/\mathfrak q_ i in the factors L_ i (compare with Lemma 10.37.16). Since B is universally Japanese the ring extensions B/\mathfrak q_ i \subset B_ i' are finite and we conclude that B' = \prod B_ i' is finite over B. Since A \to B is flat we see that any nonzerodivisor on A maps to a nonzerodivisor on B. The corresponding map
is injective (we used Lemma 10.12.15). Via this map A' maps into B'. This induces a map
which is injective (by the above and the flatness of A \to B). Since B' is a finite B-module and B is Noetherian we see that A' \otimes _ A B is a finite B-module. Hence there exist finitely many elements x_ i \in A' such that the elements x_ i \otimes 1 generate A' \otimes _ A B as a B-module. Finally, by faithful flatness of A \to B we conclude that the x_ i also generated A' as an A-module, and we win. \square
Comments (0)