2
$\begingroup$

It's true that the pushforward of a coherent sheaf is coherent via a proper morphism: but do proper morphisms preserve a finite presentation? Under some assumptions perhaps? Does it change if we are working with algebraic spaces instead of just schemes?

$\endgroup$
2
  • $\begingroup$ wait a minute: is the answer no, for trivial reasons? Take a non-finitely generated ideal $I$ in a ring $R$, then $R \to R/I$ wants to give a presentation of $R/I$ but it's not a finite presentation as $I$ is not finitely generated. $\endgroup$ Mar 28, 2012 at 17:42
  • $\begingroup$ and we know that if $R/I$ admits a finite presentation then the kernel of any surjection $R^n \to R/I$ will be finitely generated. mathoverflow.net/questions/1788/… $\endgroup$ Mar 28, 2012 at 17:43

1 Answer 1

1
$\begingroup$

Consider a ring $A$ and an ideal $I\subseteq A$, then $A/I$ is finitely presented as an $A/I$-module, but only finitely presented as an $A$-module if $I$ is finitely generated.

$\endgroup$
3
  • $\begingroup$ ah, good. so my comments made sense, cheers. $\endgroup$ Mar 28, 2012 at 17:47
  • $\begingroup$ Dear Y-S: The proper morphism in a-fortiori's example (= yours) is not finitely presented, so it doesn't seem reasonable to have expected an "operation" with such a morphism to preserve finite presentation. A more natural version of your question (to which the examples by you and a-fortiori are not applicable) is whether pushforward of quasi-coherent sheaves along a finitely presented proper morphism preserves finite presentation. Perhaps restate the question with this extra condition on the morphism? A counterexample (or proof!) for this refinement would be very interesting. Regards, q-c $\endgroup$
    – user22479
    Mar 29, 2012 at 4:48
  • $\begingroup$ I see, good point. $\endgroup$ Mar 29, 2012 at 11:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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