Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This is a homework problem for me so please do not post a full solution. I would very much appreciate a hint to move me past the point at which I'm stuck. Here is my work so far:

The setting is that $X$ is an affine algebraic variety over an algebraically closed field $k$ and $U$ is an open affine subvariety. At first I thought that $k[U]$ would necessarily be a localization of $k[X]$ and then it would automatically be flat. Then I learned (e.g. see Hailong Dao's example here) that this doesn't have to be so because a function on $k[U]$ could be locally a quotient of functions in $k[X]$ at every point of $U$ without being a quotient globally on all of $U$.

However, flatness is a very local property, so this shouldn't be a terminal problem. At each point of $U$, $k[U]$ is a localization of $k[X]$. Here's what came out when I tried to make this thought precise: $k[U]$ is flat over $k[X]$ if $k[U]_\mathfrak{m}$ is flat over $k[X]_\mathfrak{m}$ for all maximal ideals $\mathfrak{m}$ of $k[X]$. Now if $\mathfrak{m}$ corresponds to a point $p$ of $X$ that is in $U$, then $k[U]_\mathfrak{m}=k[X]_\mathfrak{m}=\mathcal{O}_{X,p}$. Since any ring is flat over itself, for $\mathfrak{m}$ corresponding to $p\in U$, we have what we need.

This also seems to be the argument given by James Milne in his notes (see p. 146, footnote 1). However, I am unsatisfied. To apply the theorem that flatness is local, I need to localize at every maximal ideal of the base ring, which in this case is $k[X]$. I localized at all $\mathfrak{m}$ corresponding to $p\in U$; but what about the maximal ideals of $k[X]$ corresponding to points outside $U$? Is it obvious that $k[U]_\mathfrak{m}$ is flat over $k[X]_\mathfrak{m}$ in these cases?

I have gotten as far as trying to describe the rings $k[U]_\mathfrak{m}$ and $k[X]_\mathfrak{m}$ to myself in these cases. $k[U]_\mathfrak{m}$ is functions regular on $U$ divided by functions nonzero at $p\notin U$. $k[X]_\mathfrak{m}$ is functions regular on all of $X$ divided by functions nonzero at $p$. This can now be a smaller ring, so it is no longer any more clear to me that $k[U]_\mathfrak{m}$ is flat over $k[X]_\mathfrak{m}$ than it was that $k[U]$ was flat over $k[X]$ in the first place. This is where I'm stuck.

Further thoughts: although I haven't seen so far how to make this local proof work, the basic logic of working locally seems inescapable to me: after all, $k[U]$ consists of quotients of functions from $k[X]$. Shouldn't it be flat for "essentially the same reason" that localizations are flat?

At any rate, a hint to push me past this stuck point would be much appreciated. Thanks in advance.

share|cite|improve this question
@Andrew - Oops! I meant but forgot to stipulate that $U$ is open. I've edited the post. – Ben Blum-Smith Nov 20 '12 at 22:15
up vote 1 down vote accepted

As Andrew said in the comment, I think your $U$ here really means an affine open set.

One thing you may want to know is that in fact, for a ring extension $A \to B$, to check that it is flat it suffices to show the flatness for $A_p \to B_{P}$, where $P$ is any prime ideal (or even better, any maximal ideal) of $B$ and $p$ is the pullback to $A$. (See, eg Matsumura's "Commutative Ring Theory", Theorem 7.1) So in fact you don't need to worry about those points lying outside $U$.

share|cite|improve this answer
Hmmm, yeah, that does it completely. I looked up the Matsumura reference. The moral of the proof seems to be that because tensoring an $A$-module with $B$ makes it a $B$-module, injectivity of $M\otimes_A B\rightarrow N\otimes_A B$ follows from injectivity at the localizations with respect to maximal ideals of $B$. Thanks. – Ben Blum-Smith Nov 21 '12 at 18:22

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.