0
$\begingroup$

This question is motivated by Shafarevich: Basic algebraic geometry in projective space 1, Chapter II, Section 1, Exercise 6.

Let $k$ be an algebraically closed field, and let $X$ be the union of $n$ one-dimensional subspaces $L_1,\dots,L_n\subseteq k^2$. Let $\mathcal{O}_0$ be the local ring of $k$ at $0$. It seems that the local ring $\mathcal{O}_{X, (0,0)}$ of $X$ at $(0,0)$ should be the subring $\mathcal{O}\subseteq \mathcal{O}_0^{\times n}$ consisting of those $(f_1,\dots, f_n)$ for which $f_1(0)=\dots=f_n(0)$. The isomorphism seems to be $f \mapsto (f|_{L_1},\dots,f|_{L_n})$ in one direction and in the other direction you glue together $n$ functions to get a rational function on $X$ regular at $(0,0)$.

... Is this correct? Am I sane?

$\endgroup$
5
  • $\begingroup$ Dear Georges, what I meant by "the local ring of $k$ at $0$" is the localization of $k[t]$ at $\langle t \rangle$. I hope this makes sense, or I am missing something obvious. $\endgroup$
    – Ben
    Commented Apr 21, 2020 at 16:31
  • $\begingroup$ Why do you think that $f$ or its restrictions to the lines $L_i$ should vanish at $0$ ? Do you realize that $1\in \mathcal O_{X,(0,0)}$ (as is the case in all rings) ? $\endgroup$ Commented Apr 21, 2020 at 17:30
  • $\begingroup$ Sorry, I don't understand your question. $f \in \mathcal{O}_{X, (0,0)}$ so it won't vanish at $(0,0)$ by definition. Meanwhile, the $f_i \in \mathcal{O}_{\mathbb{A}^1,(0)}$ so those don't vanish at $(0)$ by definition. $f|_{L_i}$ doesn't vanish at $(0)$ because $f$ doesn't vanish at $(0,0)$. $\endgroup$
    – Ben
    Commented Apr 21, 2020 at 17:50
  • $\begingroup$ You seem not to know what $\mathcal O_{X,(0,0)}$ means. That ring has a $1$, like all rings, and that $1$ does not vanish at $(0,0)$. Read the definition in Shafarevich, page 83, or in any other book for that matter. $\endgroup$ Commented Apr 21, 2020 at 17:58
  • $\begingroup$ Sorry, my previous comment was very dumb. Every time I said "don't vanish at $0$" should be replaced by "is defined at $0$". Of course there are elements of $\mathcal{O}_{X,(0,0)}$ that vanish at $(0,0)$ (e.g. $0$), and elements that don't (e.g. $1$). Despite my stupid comment, I still don't see how you gleaned from my post that I think every element of $\mathcal{O}_{X,(0,0)}$ vanishes at $(0,0)$. $\endgroup$
    – Ben
    Commented Apr 21, 2020 at 18:04

1 Answer 1

0
$\begingroup$

The map $f\mapsto (f|_{L_1},\ldots ,f|_{L_n})$ is not surjective if $n>2$, which means that in order to glue $n$ functions together you need more conditions than just $f_1(0)=\cdots = f_n(0)$.$\newcommand{\m}{\mathfrak m} \newcommand{\O}{\mathcal O}$ For example, if $x$ is the coordinate of $L$, you can see that $(x,0,0,\ldots )$ is not in the image of your map.

A more geometric reason comes from looking at the tangent spaces. Let $\m$ be the maximal ideal in $\O$, the ideal $\{(f_1,\ldots ,f_n): f_1(0) = \cdots f_n(0) = 0\}$. $\m/\m^2$ is $n$-dimensional: you can check that a basis is $\left\{(0,\ldots ,0,\overset{(i)}x,0,\ldots ,0):1\le i\le n\right\}$, where $x$ is the coordinate of the line.

Meanwhile, if $\m'$ is the maximal ideal of $\O_{X,(0,0)}$, the surjection $ \O_{k^2,(0,0)} \to \O_{X,(0,0)} $ maps $(x,y)/(x,y)^2$ onto $\m'/\m'^2$. Therefore, the dimension of $\m'/\m'^2$ is at most $2$.

I think the ring $\O_0$ you've described is the local ring of the union of the coordinate axes of $k^n$.

$\endgroup$
2
  • $\begingroup$ Thanks (now I feel foolish). So I'm back to trying to figure out the exercise in Shafarevich, which asks what the local ring of V(xy(x-y))=V(x)UV(y)UV(x-y) is at (0,0) in k^2. I can write down the definitions but this isn't very enlightening. Any ideas for this one? $\endgroup$
    – Ben
    Commented Apr 19, 2020 at 20:50
  • 2
    $\begingroup$ Don't feel foolish! I've answered this question because I believed your claim was true for several weeks before I read a paper that told me otherwise... I am not sure what Shafarevich means by "determine the local ring". I would think that saying it's the ring $k[x,y]/(xy(x-y))$ localized at $(0,0)$ would make it "determined". $\endgroup$
    – Moisés
    Commented Apr 19, 2020 at 20:54

You must log in to answer this question.

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