6
$\begingroup$

I say in advance that I am a novice in Intersection Theory, so forgive me if my question is trivial.

Let $X\subseteq\mathbb{P}^N$ be a smooth irreducible projective variety of dimension $n$ and $V, W\subseteq X$ be two irreducible subvarieties meeting properly in $X$. Choose a generic projection $f\colon X\longrightarrow \mathbb{P}^n$ and denote by $i(Z,V\cdot W;X)$ the intersection multiplicity of $V$ and $W$ at $Z$, in the ambient space $X$. As Fulton writes in his Intersection Theory (Example 8.2.6), it holds true the following formula $$i(Z, V\cdot W;X)=i(f(Z),f(V)\cdot f(W);\mathbb{P}^n).$$

This was one of Severi's methods for reducing the intersections on general varieties to intersections on the projective space. I would like to prove the formula above by using the Serre intersection formula.

I am pretty sure that there are results from Homological/Commutative Algebra (that I don't know) that would make the proof of the formula an easy exercise.
Any help is well accepted.

EDIT: If I did everything correctly, the following should be a proof of the claim above (feel free to point out possible mistakes).

Lemma. Let $f\colon A\longrightarrow B$ be an injective, flat ring homomorphism and $\mathfrak p\in\mathsf{Spec}(B), \mathfrak q:=f^{-1}(\mathfrak p)$. Then $$\mathfrak qB=\mathfrak p.$$

The ring map $B/\mathfrak qB\longrightarrow B/\mathfrak p$ is onto. We show that it is injective as well. As $f$ is flat and $A/\mathfrak q\longrightarrow B/\mathfrak p$ is injective, taking the tensor product by -$\otimes_AB$ we get an injective ring homomorphism $$\phi\colon B/\mathfrak qB=B\otimes_A A/\mathfrak q\longrightarrow B\otimes_AB/\mathfrak p.$$ Now notice that $\phi$ factorizes as $$B/\mathfrak qB\longrightarrow B/\mathfrak p \longrightarrow B\otimes_AB/\mathfrak p,$$ so the map on the left is injective.

Proof of Claim. We prove the claim for any finite surjective morphism of smooth, irreducible, projective varieties $f\colon X\longrightarrow Y$.

First of all, recall that if $f$ is as above, then it is flat. We may assume that $X$ and $Y$ are affine varieties, so we get an injective ring homomorphism $g\colon A=\mathcal O_{Y,f(Z)}\longrightarrow B=\mathcal O_{X,Z}$. Denote by $\mathfrak p,\mathfrak q\subseteq B$ the ideals corresponding to $V$ and $W$, respectively. The contractions $\mathfrak p'$ of $\mathfrak p$ and $\mathfrak q'$ of $\mathfrak q$ via $g$ are the prime ideals corresponding to $f(V)$ and $f(W)$. Thanks to Lemma we have $$\mathsf{Tor}_\bullet^B(B/\mathfrak p,B/\mathfrak q)=B\otimes_A \mathsf{Tor}_\bullet^A(A/\mathfrak p,A/\mathfrak q),$$ as it is shown here, property (10) at page 2. Thus by using standard properties of the length of modules (see here) we get $$\ell_B\mathsf{Tor}_\bullet^B(B/\mathfrak p,B/\mathfrak q)=\ell_B(B/\mathfrak m_AB)\cdot\ell_A \mathsf{Tor}_\bullet^A(A/\mathfrak p',A/\mathfrak q'),$$ where $\mathfrak m_A$ is the maximal ideal of $A$. Using our lemma again, and denoting by $\mathfrak m_B$ the maximal ideal of $B$, we get $$\ell_B(B/\mathfrak m_AB)=\ell_B(B/\mathfrak m_B)=1.$$ So we are done.

$\endgroup$
1
$\begingroup$

First, I assume that $V$ and $W$ are meeting at $Z$, and that's a typo.

Regarding your attempted proof:

  1. The Lemma is false. Consider for example $ A = k[x], B=k[y] $ and the injective flat map $ f:A \rightarrow B $ such that $ f(x) = y^2 $. Consider the ideal $ \mathfrak{p} = (y-y_0) \in Spec(B) $ with $ y_0 \ne 0 $. Then $ \mathfrak{q}= f^{-1}(\mathfrak{p})=(x-y_0^2)$ and $ \mathfrak{q}B = (y^2-y_0^2)=(y-y_0)(y+y_0) \ne \mathfrak{p}$.

  2. The reason it fails is that you assume that $A/ \mathfrak{q} \rightarrow B/ \mathfrak{p} $ is injective, which is clearly not the case.

However, the proof is still valid, since your map $f$ is a finite morphism between local rings - note that the fibers are finite, and hence locally (in a small enough neighbourhood) there will be only one ideal above $\mathfrak{p}$. (That is, you should change the assumptions in your Lemma, but the proof of the claim carries through)

$\endgroup$

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.