4
$\begingroup$

Let $V\subset \mathbf{P}^n$ be a variety. Let $f:\mathbb{P}^n\times \mathbb{P}^{n-1}\to \mathbb{P}^n$ be a blow-up of a point $P$ on $V$. Write $\widetilde{V}$ for the strict transform of $V$ under $f$, i.e., the Zariski closure of $f^{-1}(V\setminus \{P\})$.

It is tempting to guess that there ought to be an ample divisor $D$ on $\mathbb{P}^n\times \mathbb{P}^{n-1}$ such that, for $\deg_D$ the associated degree, $$\deg_D \widetilde{V} = \deg V.$$ However, as is shown in a comment, this need not be the case. What is the minimal $c_n$ such that, for every $V$, $$\deg_D \widetilde{V} \leq c_n \deg V$$ for some divisor $D$?

$\endgroup$
5
  • $\begingroup$ An easy example is $V = \mathbf P^n$ itself for $n = 2$. Then the strict transform $\tilde V$ is given by the equation $x_0y_1=x_1y_0$, corresponding to the divisor $\mathcal O(1,1)$. The cohomology ring of $\mathbf P^2 \times \mathbf P^1$ is $\mathbf Q[h_1,h_2]/(h_1^3,h_2^2)$, and $\mathcal O(1,1)$ gives cohomology class $h_1+h_2$. If $D = ah_1+bh_2$, we get $\deg \tilde V = (h_1+h_2)(ah_1+bh_2)^2 = a^2+2ab$ (the term in $h_1^2h_2$), which is always bigger than $1 = \deg V$ when $a,b > 0$. $\endgroup$ Commented Feb 2, 2021 at 3:13
  • $\begingroup$ Also a comment: a blowup in a point gives a morphism $f \colon \mathbf P^n \times \mathbf P^{n-1} \to \mathbf P^n$ such that $f^{-1}(V)$ splits up into the exceptional divisor and $\tilde V$. It's not the image of some morphism $\mathbf P^n \to \mathbf P^n \times \mathbf P^{n-1}$ (which in fact do not exist as every map $\mathbf P^n \to \mathbf P^{n-1}$ is undefined somewhere). $\endgroup$ Commented Feb 2, 2021 at 3:21
  • $\begingroup$ @R. van Dobben de Bruyn. You are of course right about $f$ - changed the wording. And the example is nice (though I don't see why it's not symmetrical in $a$ and $b$). $\endgroup$ Commented Feb 2, 2021 at 7:56
  • $\begingroup$ Why should the previous example be symmetrical in $a, \, b$? There is no involution of $\mathbf{P}^2 \times \mathbf{P}^1$ exchanging the two factors. $\endgroup$ Commented Feb 2, 2021 at 8:46
  • $\begingroup$ True - I had another reason in my head, but I guess it doesn't work. $\endgroup$ Commented Feb 2, 2021 at 9:02

1 Answer 1

3
$\begingroup$

First, the ample divisors on $\mathbf P^n \times \mathbf P^{n-1}$ are of the form $aH_1+bH_2$ for $a,b>0$, which means that every ample divisor is of the form $H_1+H_2+N$ for some nef divisor $N$. This shows that $\operatorname{deg}_{H_1+H_2}\widetilde{V} \leq \operatorname{deg}_D\widetilde{V}$ for every other ample divisor $D$.

Next, the restriction of $H_1+H_2$ to the blowup $\operatorname{Bl}_p\mathbf P^n \subset \mathbf P^n \times \mathbf P^{n-1}$ is represented by the strict transform $\widetilde{Q}$ of any quadric $Q$ smooth at $p$.

For a variety $V \subset \mathbf P^n$ of dimension $d$, we can find $d$ such quadrics $Q_1,\ldots,Q_d$ such that the projectivised tangent cones of $V$ and $Q_1 \cap \cdots \cap Q_d$ at $p$ are disjoint, and away from $p$ the intersection $V \cap Q_1 \cap \cdots \cap Q_d$ is a finite set.

Hence $$\operatorname{deg}_D \widetilde{V} = \widetilde{V} \cdot \widetilde{Q_1} \cdots \widetilde{Q_d}= V \cdot Q_1 \cdots Q_d -\operatorname{mult}_pV =2^d \operatorname{deg} V -\operatorname{mult}_p V.$$

In particular taking $V=\mathbf P^n$ as in Remy van Dobben de Bruyn's comment, we get

$$\operatorname{deg}_D \widetilde{V} = (2^n-1)\operatorname{deg} V$$

while for any other $V$ we have

$$\operatorname{deg}_D \widetilde{V} \leq 2^d \operatorname{deg} V<(2^n-1) \operatorname{deg} V.$$

So $c_n=2^n-1$.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .