Let $X\to S$ be a morphism of noetherian schemes such that, for all $s$ in $S$, the morphism $X_s\to $ Spec $k(s)$ is projective.

Then it doesn't follow that $X\to S$ is projective in general. In fact, if $X\to S$ is an open immersion then every fibre $X_s\to $ Spec $k(s)$ is projective (some of the fibres are presumable empty, and I am using that $\emptyset \to S$ is a projective morphism). So we at least need something like surjectivity.

Thus, assume $X\to S$ is surjective. Then, still it doesn't follow that $X\to S$ is proper as one can find non-projective proper threefolds $X$ over $\mathbf C$ that fibre over the projective line $S=\mathbf P^1_{\mathbf C}$ whose singularities on the fibres aren't too bad. This morphism $X\to S$ has projective fibres, but $X$ is not projective. Thus $X\to S$ is not projective.

I'm wondering what one can say about flat surjective morphisms $X\to S$ whose fibres are projective (resp. proper). Under which conditions is the morphism $X\to S$ projective (resp. proper)?

  • 4
    $\begingroup$ Tautologically, existence of a relatively ample line bundle. What else do you expect? $\endgroup$ – abx Dec 8 '13 at 13:04
  • $\begingroup$ @abx: the case of an open immersion is an obvious counterexample. $\endgroup$ – Laurent Moret-Bailly Dec 9 '13 at 8:22
  • 2
    $\begingroup$ @Kuipers: if $U\subset X$ is open (but not closed) then $U\coprod X\to X$ is surjective with projective fibers, but not proper. $\endgroup$ – Laurent Moret-Bailly Dec 9 '13 at 8:25

Let $f : X\to Y$ be a separated morphism of finite type, with $Y$ Noetherian. Suppose $f$ is faithfully flat and its fibers are geometrically connected and proper. Then $f$ is proper. See EGA IV.15.7.10 and 15.7.8. If you do not require fibers to be connected, counterexamples are not hard to find.

I am not sure about projectivity but this answer apparently says that if you have a proper flat morphism to a Noetherian regular scheme with fibers projective spaces (non-closed points are included), then the total space is projective. I think that Hartshorne exercise mentioned there only needs the fact that invertible sheaves can be extended from open sets and because of this lemma, I think that one can weaken regularity assumption to local factoriality.

| cite | improve this answer | |

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.