Restriction of rational functions to closed subvarieties

204 Views Asked by At

I'm confused about a basic point regarding the definition of the ring of regular functions on a closed subvariety. Let $X=\text{Spec}(A)$ be an affine variety (I'm thinking of varieties as integral, separated schemes of finite over an algebraically closed field but I don't think it matters here). Let $\mathfrak{p}$ be a prime ideal in A; we have a closed subscheme $Y=\text{Spec}(A/\mathfrak{p})\cong V(\mathfrak{p})\subset X$. Therefore the regular functions on $Y$ are the restrictions of regular functions on $X$, given by the quotient map $A\to A/\mathfrak{p}$. In particular I'm thinking about $A = \mathbb{C}[x,y]$, with $\mathfrak{p} = (f)$ for some irreducible polynomial $f$, so the regular functions on $V(f)$ are just polynomials in $x$ and $y$, identified if they agree on $V(f)$.

However, on an open set $U\subset X$, there are additional regular functions other than the restrictions of regular functions on $X$; in particular, we allow rational functions with poles outside $U$. For concreteness, let $h\in A$ be irreducible and suppose that $U=D(h)$, so we have the distinguished inclusion $U\cong \text{Spec}(A_h)\hookrightarrow \text{Spec}(A)$. The regular functions on $U$ then look like $\frac{g(x,y)}{h(x,y)^n}$ for $g\in A$ and $n\geq 0$.

Question: So what happens if $Y\subset U \subset X$? (That is, $V(f)\cap V(h)=\emptyset$.) Thinking of $Y$ as a closed subvariety of $U$, its regular functions are restrictions of regular functions on $U$ (and thus rational functions on $X$), so they are elements of $A_h/(f)$. It's not clear to me how to show in general (if it's even true!) that $A_h/(f)\cong A/(f)$, and so these are in fact the same scheme structure on $Y$.

I suspect that these should in fact be the same structure, and so any rational function like $\frac{g(x,y)}{h(x,y)^n}$ restricted to $V(f)$ should be the same as the restriction of some polynomial $q(x,y)$ to $V(f)$ (again, assuming that $h$ vanishes nowhere on $V(f)$). The only cases that I can come up in $\mathbb{A}_{\mathbb{C}}^2$ of two nonintersecting hypersurfaces are two parallel lines, and there it is clear that, if $f=0$ and $h=0$ define these lines, then $h$ is constant on $V(f)$ so these rational functions agree with polynomials. But I have no idea how one would deal with this in general cases where the intersections or lack thereof are much less obvious.

1

There are 1 best solutions below

0
On

These two rings are in fact the same, so we can consider the subvariety $Y$ directly in $X$, or $Y$ in $U$ in $X$, and and the resulting structure sheaf is the same. Note that $h$ is already invertible in $A/(f)$: since $h$ doesn't vanish at any point of $V(f)$, this mean it isn't contained in any maximal ideal of $A/(f)$, so it must be a unit. Therefore $(A/(f))_h = A/(f)$. Now we use the fact that localization commutes with quotients, so $A_h/(f)\cong (A/(f))_h$. Putting these facts together, we have $A_h/(f)\cong A/(f)$. Therefore the regular functions are in fact the same.