7
$\begingroup$

Let us define smooth del Pezzo surfaces $dP_r$ as the blowup of $r$ generic points in $\mathbb{CP}_2$. One can show that if we request $dP_r$ to be Fano, then $r=0,...,8$. In theoretical physics literature is common to define also $dP_9$, which is the blowup of $9$ points in $\mathbb{CP}_2$ and refer to this as well as a del Pezzo surface, while as far as I understand mathematicians would call this a generic rational elliptically fibered surface.

Now, the question is why only $dP_r$ for $r=0,1,2,3$ are toric varieties? This statement is given for example in the appendix A.3 of http://inspirehep.net/record/1707635

For example, consider the fan of $\mathbb{CP}_2$. In order to perform the $i$-th blow-up, at the level of the fan, we add a new 1-dimensional cone which is representing the exceptional divisor $E_i$, associated to the new $\mathbb{CP}_1$. Which is the obstruction from keeping adding cones also for $r>3$? Why does the toric description fail?

$\endgroup$
4
  • 9
    $\begingroup$ You are only allowed to blow up torus-invariant points, and for $\mathbb{P}^2$ these lie on the intersections of the 3 torus-invariant divisors, which form a triangle. You can only blow up these 3 points if you want your surface to stay toric. $\endgroup$
    – pbelmans
    Mar 28, 2019 at 16:57
  • 3
    $\begingroup$ There is no obstruction to adding cones for r>3: you get some surface obtained from $\mathbb{P}^2$ by blowing up $r$ points but these $r$ points are not in generic position if $r>3$. $\endgroup$
    – user25309
    Mar 28, 2019 at 17:39
  • 1
    $\begingroup$ The last phrase of my comment should of course say "if you want your surface to stay toric and del Pezzo", as pointed out by @user25309. $\endgroup$
    – pbelmans
    Mar 28, 2019 at 17:54
  • $\begingroup$ I’ve heard physicists calling rational elliptic surface “half K3”. $\endgroup$ Apr 2, 2019 at 13:08

1 Answer 1

3
$\begingroup$

You have no problem for adding cones from a toric surface $\mathbb{P}^2$ or any surface obtained by blowing-up points on it.

However, there are only three toric points on $\mathbb{P}^2$, so once you have blown-up these $3$ points, the toric points that you see on your surface lie on $(-1)$-curves, and thus do not correspond anymore to points of $\mathbb{P}^2$. You can still blow-up them, and it corresponds to what you expect on the fan point of view, but you no longer get a del Pezzo surface, as there are $(-2)$-curves.

In general, if you blow-up more than $8$ points of $\mathbb{P}^2$ in (very) general points, you get a non-toric surface with only curve of self-intersection $\ge -1$ and such that negative curves should be $(-1)$-curves (it is in fact a conjecture, see https://arxiv.org/pdf/math/0512631.pdf ), but this is no longer del Pezzo as the anti-canonical divisor is no longer ample, only nef and not big.

$\endgroup$
5
  • 1
    $\begingroup$ Small clarification to the last line: if you blow up more than 9 points, the anticanonical is not even nef. $\endgroup$
    – Pop
    Apr 1, 2019 at 13:53
  • $\begingroup$ For $9$, it is certainly nef, but for $10$, why should it be not nef? Which curve is positive against the canonical divisor? $\endgroup$ Apr 4, 2019 at 12:42
  • 2
    $\begingroup$ Dear J\'er\'emy, a nef divisor on a surface always has nonnegative selfintersection. For 9+n points the anticanonical has selfintersection -n, so it cannot be nef. $\endgroup$
    – Pop
    Apr 4, 2019 at 21:18
  • 1
    $\begingroup$ To answer your second question: If I counted parameters correctly, given 10 points in the plane there is a curve of degree 9 with multiplicity 3 at 8 of the points and multiplicity 2 at the other 2. Its intersection with the anticanonical is $3 \cdot 9 - 8 \cdot 3 - 2 \cdot 2 =-1$. $\endgroup$
    – Pop
    Apr 5, 2019 at 6:35
  • $\begingroup$ Thanks for the answer. $\endgroup$ Apr 6, 2019 at 13:54

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.