2
$\begingroup$

I am working on Exercise 6.3.N. on Ravi Vakil's notes (on morphism from $S$ scheme to $\mathbb{P}_B^n$) and I would like some assistance. The exercise states: Let $B$ be a ring. If $X$ is a $B$-scheme, and $f_0, ..., f_n$ are $n+1$ functions on $X$ with no common zeros, then show that $[f_0, ..., f_n]$ gives a morphism of $B$-schemes $X \rightarrow \mathbb{P}_B^n$.

I would appreciate any hint, comments, etc. Thank you very much!


Edit (Not by OP) (10/06/2021). In the 2017 version of the notes, this is exercise 6.3.M.

$\endgroup$
1
  • 1
    $\begingroup$ Now in the 2022 Aug version, it's exercise 7.3.O $\endgroup$
    – onriv
    Dec 16, 2022 at 3:15

2 Answers 2

5
$\begingroup$

If I denote the homogeneous coordinates on $\mathbf{P}^n_B$ by $x_0, \dots, x_n$ then $X_{f_i}$ should map into the affine open $D(x_i) \subset \mathbf{P}^n_B$. To specify this restriction you just need to give a $B$-homomorphism $B[x_0/x_i, \dots, x_n/x_i] \to \Gamma(X_{f_i}, \mathscr{O}_X)$. What's the right formula? Then check that these agree on overlaps.

$\endgroup$
8
  • $\begingroup$ What does the notation $X_{f_i}$ mean here? Thank you! $\endgroup$
    – Oga
    Apr 5, 2015 at 21:31
  • $\begingroup$ @Oga The open set of points where $f_i$ does not vanish, in the sense that its image in the residue field is not zero. I'm sure he defines something like this at some point. $\endgroup$
    – Hoot
    Apr 6, 2015 at 2:15
  • 1
    $\begingroup$ Thank you for the answer. I figured it is the map that sends $x_l/x_i to f_l |_{X_{f_i}}$. Could you possibly explain me why the morphisms $[f_0, ..., f_n]$ and $[gf_0, ..., gf_n]$ to $\mathbb{P}_B^n$ are the same, if $g$ does not vanish anywhere on $X$? This is part b of the question. I can see that the map of topological spaces are the same, but they seem to have (a slightly) different maps of sheaves... Thank you! $\endgroup$
    – Oga
    Apr 8, 2015 at 18:51
  • $\begingroup$ @Oga I should have included more of an indication of what's going on. I think you want to send $x_l/x_i$ to $f_l/f_i$. This should make part (b) more obvious! $\endgroup$
    – Hoot
    Apr 9, 2015 at 3:35
  • $\begingroup$ oops, I guess I had a wrong map! Thank you for the clarification! $\endgroup$
    – Oga
    Apr 9, 2015 at 20:34
0
$\begingroup$

I would suggest another solution. We have done most of the gluing work already in what is now (December 2022 version) exercise 7.3.F. There we construct a natural map $\mathbb{A}_B^{n+1} \setminus \{0\} \to \mathbb{P}_B^n$. (Technically we've only done it for fields $k$, but the construction of course works generally for rings $B$.) Hence it suffices to produce morphism $X \to \mathbb{A}_B^n \setminus \{0\}$, which also more or less directly solves subtask (b). But this is easily constructed: A morphism $\varphi: X \to \mathbb{A}_B^n$ is induced by the ring morphism $$ \begin{align} B[x_0,...,x_n] &\to \Gamma(X,\mathcal{O}_X) \\ x_i &\mapsto f_i \end{align} $$ and we are left to show that this morphism doesn't hit $0 \in \mathbb{A}_B^n$. Note that the pullback of the function $x_i \in \Gamma(\mathbb{A}_B^n, \mathcal{O}_{\mathbb{A}_B^n})$ is $f_i \in \Gamma(X,\mathcal{O}_X)$. Since we have a morphism of locally ringed spaces, $x_i$ vanishing at $p \in \mathbb{A}_B^n$ implies $f_i$ vanishes at all $q \in \varphi^{-1}(p) \subseteq X$. But at $0 \in \mathbb{A}_B^n$, all $x_i$ vanish. Thus all $f_i$ vanish at all $q \in \varphi^{-1}(0)$. By assumption such a point $q$ can't exist, showing that $\operatorname{im} \varphi \subseteq \mathbb{A}_B^n \setminus \{0\}$.

$\endgroup$

You must log in to answer this question.

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