Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let F be a quasi-coherent sheaf on a scheme X. To check that F vanishes it suffices to check that all the stalks of F vanish. I would like to know whether it suffices to check that all the fibers of F vanish.

(I think I am using standard terms: the tensor product over O_X with the local ring at x is the stalk, and the tensor product over O_X with the residue field at x is the fiber.)

It suffices to answer the question on an affine scheme. Let R be a commutative ring. For each prime ideal p of R let k(p) be the residue field of the local ring R_p. Let M be an R-module, and suppose that Tor_i(k(p),M) = 0 for all i and all p. (Tor taken in the category of R-modules). Does it follow that M = 0?

If M is finitely generated, the answer is yes. In that case M vanishes even when Tor_0(k(p),M) = 0 for all maximal p, by Nakayama's lemma. My question is whether there is a good replacement for Nakayama's lemma when M is not finitely generated.

share|improve this question
    
I believe something similar will hold if M is simple. –  Harry Gindi Jan 26 '10 at 0:44

2 Answers 2

up vote 2 down vote accepted

If the scheme is locally noetherian, this is true and can be proved by noetherian induction. In fact, you can even replace $M$ with an object of bounded derived quasi-coherent category, if you are interested in such things.

The proof is relatively straightforward: For a complex of modules $M$ over the ring $R$, we may assume that any non-zero $f\in R$ acts by a quasi-isomorphism $f:M\to M$ (by the induction hypothesis), and then the cohomology of $M$ are defined over the field of fractions of $M$.

share|improve this answer
    
Wonderful. Is this a famous argument, or off the top of your head? –  David Treumann Jan 26 '10 at 1:19
    
I suspect that this is well known. If so, I would like references; the only one I know is Lemma 10 in Arinkin, Orthogonality of natural sheaves on moduli stacks of SL(2)-bundles with connections on P^1 minus 4 points. –  t3suji Jan 26 '10 at 2:24

What about $R=\mathbb Z$ and $M=\mathbb Q/\mathbb Z$ ? All fibers are zero.

share|improve this answer
    
You'll notice he has derived fibers. Those are non-zero in your example. –  t3suji Jan 26 '10 at 1:02
    
Yes, you are right. I only answered the first part of the question. –  Qing Liu Jan 26 '10 at 1:06

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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