-1
$\begingroup$

I am trying to learn how to compute the projective bundle $\mathbb{P}(\mathcal{O}(a_1)\oplus \cdots \mathcal{O}(a_k))$ over some projective space using relative proj. How can I find a presentation for the ideal $I$ giving the closed subscheme of $\mathbb{P}^n\times\mathbb{P}^m$? For example, I want to understand how I can find an algebra representing the Hirzebruch surface $\mathbb{P}(\mathcal{O}\oplus \mathcal{O}(1))$ over $\mathbb{P}^1_{s,t}$.


For my example, I have to twist by $\mathcal{O}(1)$ to get a vanishing $h^1$. Then, there is a map $$ \mathcal{O}^{\oplus 5} \xrightarrow{ \begin{bmatrix} s \\ t \\ s^2 \\ st \\ t^2 \end{bmatrix}} \mathcal{O}(1)\oplus\mathcal{O}(2) \to 0 $$ All that's left is finding the generators of the ideal $I \subseteq \mathbb{C}[s,t][x_0,\ldots, x_4]$, which I think has the presentation $$ (tx_0 - sx_1, x_2x_4 - x_3^2, t^2x_2 - s^2x_4) $$

$\endgroup$
9
  • $\begingroup$ For any locally free sheaf $\mathscr E$, $\mathbb P(\mathscr E)=\mathrm {Proj}\ \mathrm{Sym} (\mathscr E)$. Is this what you are asking? $\endgroup$ Commented Feb 13, 2017 at 0:48
  • $\begingroup$ No, I want to get an actual algebra presentation for this construction. $\endgroup$
    – 54321user
    Commented Feb 13, 2017 at 0:52
  • 1
    $\begingroup$ But this is projective, so it will not be a Spec and I gave you a presentation as a Proj. I still don't understand what it is that you want. Could you give a projective example? $\endgroup$ Commented Feb 13, 2017 at 2:22
  • 1
    $\begingroup$ Aha, so your question is that you want an absolute Proj and not a relative Proj? $\endgroup$ Commented Feb 13, 2017 at 2:47
  • 1
    $\begingroup$ For all of the downvoters, can you please explain why you gave me these votes? It seems like many Hartshorne level questions go above math.stackexchange and are more appropriately written here. $\endgroup$
    – 54321user
    Commented May 6, 2017 at 21:09

1 Answer 1

8
$\begingroup$

OK, I guess I will write it here as it might need more room.

You know that $\mathbb P(\mathscr E)$ remains the same if you twist it by a line bundle. So, choose a sufficiently ample line bundle $\mathscr L$ such that $\mathscr E\otimes \mathscr L$ is generated by global sections and switch $\mathscr E$ with $\mathscr E\otimes \mathscr L$. In other words, you may assume that $\mathscr E$ is generated by global sections.

Now you have a surjective morphism $$ \mathscr O^{\oplus (m+1)}_X \to \mathscr E. $$ which using the properties of projective bundles gives you an embedding $$ \mathbb P(\mathscr E) \to X\times \mathbb P^m. $$

Now, you can probably write down an absolute Proj for $\mathbb P^2\times \mathbb P^m$ and then your original projective bundle is a closed subscheme in there, so you just have to figure out its ideal.

Ta-da.

$\endgroup$
4
  • $\begingroup$ Does $\mathbb{V}(\mathcal{E})$ remain the same if I twist by a line bundle? $\endgroup$
    – 54321user
    Commented Jul 26, 2017 at 4:56
  • $\begingroup$ No. The sheaf of sections of $\mathbb V(\mathscr E)$ is the dual of $\mathscr E$. This is the same phenomenon as with "standard" $\mathrm{Spec}$ and $\mathrm{Proj}$; $\mathrm{Spec}(A)\simeq \mathrm{Spec} B$ if and only if $A\simeq B$, but the analogous statement fails for $\mathrm{Proj}$. The twisting with the line bundle corresponds to a $d$-uple type embedding of the projective scheme/bundle. They are abstractly isomorphic, but have a different embedding. $\endgroup$ Commented Jul 26, 2017 at 17:00
  • $\begingroup$ I created a bounty for a question related to my comment: math.stackexchange.com/questions/2364401/… $\endgroup$
    – 54321user
    Commented Aug 26, 2017 at 22:20
  • $\begingroup$ Also, why can the Hirzebruch surface have such a simpler presentation, as seen here: mathoverflow.net/questions/122952/on-a-hirzebruch-surface ? $\endgroup$
    – 54321user
    Commented Sep 4, 2017 at 5:22

You must log in to answer this question.

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