4
$\begingroup$

I cannot understand the notes that my teachers use for class.

Theorem: let $S$ be a scheme and $E$ a quasicoherent $O_S$-module. The functor $$Sch_S^o\rightarrow Set$$ $$(\phi :X\rightarrow S)\mapsto \{[\phi ^*E\rightarrow L]\}$$ is representable by an $S$-scheme $\mathbb{P}(E)\rightarrow S$. To elaborate on the meaning on the RHS of $\mapsto$ an $S$-scheme $X$ is mapped to the set of maps from $\phi ^*E$ to $L$, where $L$ is invertible, modulo isomorphisms i.e. two maps $a$ and $b$ are same iff $\exists s:L_1\rightarrow L_2$ such that $b=sa$.

For a ring $A$, we denote $\mathbb{P}^n _A :=\mathbb{P} ((A^{(n+1)})^*)$. Firstly how is this seen as a quasicoherent sheaf on $\mathrm{Spec}(A)$ As an example they take $\mathbb{P} ^n _\mathbb{C}(\mathbb{C}) _{\mathbb{C}}$ which I don't understand. Firstly, can anyone guess the meaning of the last $\mathbb{C}$? I have given my guess below. The one in the brackets is talking about $\mathbb{C}$ points of the scheme. Apparently, the following sequence of equalities hold $\mathbb{P}^n _\mathbb{C}(\mathbb{C})_\mathbb{C}=\{\mathrm{Spec}(\mathbb{C})\mathrm{-morphisms.of. }\mathrm{Spec}(\mathbb{C})\mathrm{to.}\mathbb{P}^n _\mathbb{C}\}$ $=\{\mathrm{invertible.quotients.of.rank.1.of.}(\mathbb{C}^{n+1})^*\}=\{\mathrm{subspaces.of.rank.1.of.}\mathbb{C}^{n+1}\}$ which is indeed the projective space. Can anyoune explain meaning of the first and second $=$? I think the first is just definition. What would be obtained if we would take $\mathbb{P} ^n _\mathbb{C} (\mathbb{C})$ ?

How actually to work with this (in my view) cumbersome definition? How is $\mathrm{Proj}$ related to this?

$\endgroup$
  • $\begingroup$ Do you have a link to the notes in question? $\endgroup$ – Daniel McLaury Dec 25 '16 at 17:02
  • $\begingroup$ I have, but it is in French. However the part on $\mathbb{P}^n _{\mathbb{C}}$ is not visible to the general public. $\endgroup$ – ralleee Dec 25 '16 at 17:16
3
$\begingroup$

If we define the functor $\mathbb{P}$ by its representable functor property, then $\mathbb{P}(\mathcal{E})$ is by definition the $S$-scheme such that, for any $S$-scheme $\pi : X \to S$, we have $$\operatorname{Hom}_S(X, \mathbb{P}(\mathcal{E})) = \{ \text{invertible quotients of } \pi^\ast\mathcal{E} \}$$ for all quasicoherent $\mathcal{E}$ (and such that the right thing happens for morphisms). You've written this as a theorem above, so you're working with a different definition, but at any rate this is a true statement we can use.

Now $(A^{n+1})^\ast$ is an $A$-module, so the sheaf on $\operatorname{Spec} A$ associated to it is quasicoherent by definition. So we can apply the above result to it with no trouble.

For the example you give, we want to consider $\mathbb{P}^n_{\mathbb{C}} = \mathbb{P}((\mathbb{C}^{n+1})^\ast)$. By definition its $\mathbb{C}$-points are given by $$\operatorname{Hom}_\mathbb{C}(\operatorname{Spec} \mathbb{C}, \mathbb{P}^n_\mathbb{C})$$ so let's take $S = \operatorname{Spec} \mathbb{C}$ and $X = \operatorname{Spec} \mathbb{C}$ and $\mathcal{E} = (\mathbb{C}^{n+1})^\ast$. Here $\mathbb{C}$ denotes the trivial line bundle on $S$, which is of course just a copy of the vector space $\mathbb{C}$ since $S$ consists of a single point. From the defining property of $\mathbb{P}(\mathcal{E})$ above we have $$\operatorname{Hom}_{\operatorname{Spec} \mathbb{C}}(\operatorname{Spec} \mathbb{C}, \mathbb{P}^n_\mathbb{C}) = \{\text{invertible quotients of } \pi^\ast (\mathbb{C}^{n+1})^\ast\}$$ Since $X$ consists of a single point, $\pi^\ast (\mathbb{C}^{n+1})^\ast$ is just a single copy of $(\mathbb{C}^{n+1})^\ast$. And the rest is now linear algebra.

I'm not a hundred percent sure what they were going for with the notation. There's at least one error -- a $\mathbb{P}^n$ where a $\mathbb{P}$ is meant -- so it could be a mistake. It could also mean that $\mathbb{P}^n_\mathbb{C}$ is taken as a scheme per se, whereas the other thing denotes the same space regarded as a $\mathbb{C}$-scheme.

$\endgroup$
  • $\begingroup$ Awesome. Thanks! Can you just add me one more thing, if it is not too difficult, what is relation with the $\mathrm{Proj}$ functor? $\endgroup$ – ralleee Dec 25 '16 at 21:15

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.