I am looking for a very concrete and simple example of a line bundle $L$ (on a curve or a surface) which is ample, but not very ample. I would also like that $L^{\otimes k}$ is very ample for a small $k$, in the sense that I can do a very hands-on computation and show that, say, all degree $3$ monomials in certain global sections yield an immersion into projective space. Thanks a lot in advance!

  • 5
    $\begingroup$ You could take the canonical bundle on a hyperelliptic curve. Is that easy enough? $\endgroup$ Oct 28 '13 at 12:58
  • 7
    $\begingroup$ Another easy example: the line bundle corresponding to a point on an elliptic curve. $\endgroup$
    – user64687
    Oct 28 '13 at 13:11

(Let me collect the comments into a CW answer. Others can then edit the answer to add their own favourites.)

Here are a couple of simple examples, and one non-simple one. Note that any line bundle of degree $\geq 2g+1$ on a curve of genus $g$ is very ample, so any line bundle of positive degree on a curve is ample.

  1. The canonical bundle $K$ on a hyperelliptic curve of genus $\geq 2$. Sections of $K$ define a 2:1 cover, so $K$ is globally generated and ample, but not very ample. On the other hand $K^2$ is very ample: for $g \geq 3$ this is immediate by the above comment; for $g=2$ the argument is a little more involved (Hartshorne IV.3.1).

  2. Any bundle $L=O_C(p)$ where $p$ is a point on an elliptic curve $C$. Riemann—Roch shows that such a bundle has a 1-dimensional space of global sections, so is not very ample, or even globally generated, but it has positive degree, so is ample. On the other hand $L^3$ is very ample, and embeds $C$ into $\mathbf{P}^2$ as a smoooth cubic, with $p$ mapping to a flex.

  3. If $C$ is a curve of genus $2$, and $p,q,r$ are general points on $C$, then the bundle $L=\mathcal{O}_C(p+q-r)$$ is ample, but has no global sections at all.

  4. For a trickier example, one could consider a so-called Godeaux surface. This is a particular kind of surface of general type constructed as a quotient of a quintic surface in $\mathbf{P}^3$. It has the property that the canonical bundle $K_S$ is ample, but has no global sections. For more details, see the excellent answer of Clay Cordova here. Sadly, in this case I don't know what power of $K$ is needed to obtain a very ample bundle.

  • $\begingroup$ Unfortunately, it seems like only (3) really meets my requirements, because in the other cases the bundle isn't ample. I will have a look later. $\endgroup$ Oct 29 '13 at 9:05
  • $\begingroup$ @JeskoHüttenhain: which bundle isn't ample? As I said, any line bundle of positive degree on a curve is ample, so the bundles in 1 and 2 are ample. Is that what you meant? $\endgroup$
    – user64687
    Oct 29 '13 at 9:19
  • $\begingroup$ Hum, now I am confused. You wrote "Sections of K define a 2:1 cover, so K is globally generated but not ample." I think you simply mistyped then, because if that means it's a finite cover of degree $2$, then it is ample but not very ample. And that'd be what I want. In (2), you say it isn't globally generated and I thought that being globally generated is necessary for being ample. $\endgroup$ Oct 29 '13 at 9:23
  • $\begingroup$ @JeskoHüttenhain: yes, I just saw that I left out the word "very" in (1). Sorry for the confusion! For (2), no, an ample bundle need not be globally generated --- indeed as the example in (3) shows, it need not have any sections at all! $\endgroup$
    – user64687
    Oct 29 '13 at 9:24
  • 2
    $\begingroup$ Fascinating. This is presenting to be exactly the learning experience I had hoped it would be. I think I will have a close look at all of those ;). $\endgroup$ Oct 29 '13 at 9:27

Let $X$ be a nonsingular projective curve and $L$ be a line bundle on $X$. Then $L$ is ample if and only if deg$L> 0$. (see: Ample vector bundles, Hartshorne Proposition 7.1) It is not difficult to see that a line bundle $L$ over an elliptic curve is very ample iff and only if deg$L \geq 3$.


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.