4
$\begingroup$

Let $f:\tilde{X} \to X$ be a normalization of projective variety. Let $L$ be a very ample line bundle on $X$. Is $f^*L$ a very ample line bundle on $\tilde{X}$? If not true in general, is there any dimension restriction on $X$ for which this is true?

$\endgroup$
5
  • 1
    $\begingroup$ Not in general, because the sections of $f^*L$ need not separate points. E.g. take a nodal curve in the plane, and then normalize. $\endgroup$ Apr 19, 2014 at 12:55
  • $\begingroup$ Never, sort of definitionally, unless $X$ is normal . It does stay ample. $\endgroup$
    – meh
    Apr 19, 2014 at 12:56
  • $\begingroup$ I don't understand the two comments above. Take $\mathcal O(1)$ on $\mathbb P^2$ and restrict to a nodal cubic $X$. Then $f^* L$ is degree $3$ on $\tilde X$, hence very ample, no? $\endgroup$ Apr 19, 2014 at 14:37
  • $\begingroup$ OK, yes, my 2nd sentence was not carefully formulated. I was thinking of a scenario along the lines of abx 2). $\endgroup$ Apr 19, 2014 at 14:56
  • $\begingroup$ Mine too :) . I think abx took the time to think out the question more thoroughly. $\endgroup$
    – meh
    Apr 20, 2014 at 12:46

1 Answer 1

6
$\begingroup$

1) There are certainly many cases where $f^*L$ is still very ample: e.g. if $X$ is an irreducible curve, any line bundle on $X$ of degree $\geq 2g(\tilde{X} )+1$ will have this property.

2) A typical situation where $f^*L$ is not very ample is when $H^0(\tilde{X},f^*L )\cong H^0(X,L)$. For instance, take for $X$ a curve with one ordinary double point $s$, and $L=\omega _X(-p)$, $p$ a smooth point of $X$. Then $f^*L=\omega _{\tilde{X}}(s'+s''-\tilde{p} )$, where $s',s''$ are the 2 points of $\tilde{X} $ above $s$ and $\tilde{p} $ the point above $p$. Riemann-Roch gives $h^0(f^*L)= g(\tilde{X} )=h^0(L)$ (I choose $p$ so that $h^0(s'+s''-\tilde{p} )=0$).

I think there is no simple criterion allowing to decide which situation occurs, you have to work.

$\endgroup$
2
  • $\begingroup$ Thanks for the answer. Could you indicate how to proceed in the case of irreducible surfaces with at most rational singularities (at isolated points). $\endgroup$
    – user46578
    Apr 24, 2014 at 12:06
  • $\begingroup$ But aren't these surfaces already normal? $\endgroup$
    – abx
    May 5, 2014 at 5:31

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.