For a homomorphism of rings $R \to S$, the following are equivalent:

a) $(-) \otimes_R S : \mathrm{Mod}(R) \to \mathrm{Mod}(S)$ reflects isomorphisms

b) $R \to S$ satisfies effective descent with respect to modules.

c) $R \to S$ is pure: For every $R$-module $M$ the natural map $M \to M \otimes_R S$ is a monomorphism.

For a reference, see "Descent Theory for Schemes" by Bachuki Mesablishvili; this work generalizes Grothendieck's descent theory for quasi-coherent modules.

Question. Let $R$ be a ring. Is there a noetherian ring $S$ together with a ring homomorphism $R \to S$ satisfying the equivalent properties above?

I don't know which one of (a) or (c) are more easy to verify. I've also included (b) to illustrate the strength of the condition, but also the geometric significance: Is every (affine) scheme "built up" out of noetherian schemes with respect to the "topology" of effective descent mophisms for quasi-coherent modules?

An affirmative answer would settle another problem in my work. On the other hand, I have no idea how to construct a reasonable noetherian ring $S$ over $R$ at all. If $R$ is a domain (which you may assume for convenience), then $S = \mathrm{Quot}(R)$ is an example, but of course it does not satisfy the condition.


1 Answer 1


I am afraid that this is not true for any non-noetherian ring.

Let $R \to S$ be a descent extension, and assume that $S$ is noetherian. Let $I$ be an ideal of $R$; there will be a finitely generated ideal $J \subseteq I$ of $R$ such that $JS = IS$. Hence $R/I$ and $R/J$ become isomorphic when tensored with $S$, so they are isomorphic are $R$-modules. So $I = J$. Hence $R$ is noetherian.

  • $\begingroup$ Thanks Angelo! I have no idea why I didn't see this obvious proof, especially after having written down the same for faithfully flat $R \to S$. $\endgroup$ Nov 27, 2011 at 17:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.