Why should a transverse intersection give intersection multiplicity one?

357 Views Asked by At

My setting is that of Hartshorne chapter I ("classical varieties" over an algebraically closed field, which are irreducible [not schemes]).

Background

I've been working on problem 7.7(a) and I've hit a bit of a wall with part (a):

Problem 7.7. Let $Y$ be a variety of dimension $r$ and degree $d>1$ in $\Bbb P^n$. Let $P\in Y$ be a nonsingular point. Define $X$ to be the closure of the union of all lines $PQ$, where $Q\in Y$, $Q\neq P$.

(a) Show that $X$ is a variety of dimension $r+1$.

Clearly $X$ contains $Y$, so $X$ must be of dimension at least $r$. On the other hand, we can define a map $f$ from $(Y\setminus\{P\})\times\Bbb P^1$ to $X$ by sending $(Q,[a:b])\mapsto aQ+bP$ which surjects on to the collection of points in $X$ which are on a line $PQ$ with $Q\in Y$ and $Q\neq P$. Since $(Y\setminus\{P\})\times\Bbb P^1$ is irreducible, this shows that $X$ is irreducible, and the map is dominant, so we get an extension of function fields $k(X)\to K(Y\times\Bbb P^1)$. This shows that $\dim X \leq r+1$. So all I have to do finish the problem is to show that $X$ contains one point which isn't in $Y$. This is turning out to be harder than I expected!

Here's how I'd want to solve this problem: Pick a hyperplane $H$ through $P$ which is transverse to $Y$ (we can do this because $r<n$ by the degree assumption plus the fact that $P$ is a smooth point). Then the algebraic set $Y\cap H$ is smooth at $P$ (I can prove this via the Jacobian criteria), so there's a unique irreducible component $Z\subset Y\cap H$ through $P$. If I could prove that the intersection multiplicity along $Z$ is equal to one, I'd win: either $Y\cap H$ is reducible and I can connect $P$ with a point $Q$ on some other irreducible component and find a point in $X\setminus Y$ on the line $PQ$, or $Y\cap H=Z$, a variety of dimension $r-1$ and degree $d$, and eventually I can get it down to the case of a curve and I understand what to do in that case.

Question

Suppose $Y\subset \Bbb P^n$ is a variety of dimension $r$ and degree $d>1$, and $P$ a smooth point on $Y$. Suppose $H$ is a hyperplane through $P$ so that $Y\cap H$ is smooth at $P$ and $Y\cap H$ is irreducible. Why should $i(Y,H;Y\cap H)=1$?

My thoughts

The problem is that I feel very unsure about why or how I should be able to get this result. I'm supposed to compute the length of $(S/(I_Y+h))_\mathfrak{p}$ over the local ring $S_\mathfrak{p}$, where $\mathfrak{p}$ is the homogeneous prime ideal corresponding to the irreducible subvariety $Y\cap H$ and $h$ is a generator for the ideal of $H$. I can show that this the same as what you'd want in the affine case - that is, if we work in some standard affine patch $U$ containing $P$, I can show that the intersection multiplicity is the same as the length of $(k[Y\cap U]/(\widetilde{h}))_{\mathfrak{p}'}$ over $k[Y\cap U]_{\mathfrak{p}'}$ where $\mathfrak{p}'$ is the ideal corresponding to $Y\cap H\cap U$, but I don't see how I'm supposed to conclude this is $1$ if I know smoothness at $P$.

Edit: I've made a little progress - I can show that $f$ is in the maximal ideal of $P$ in $Y$ but not it's square, so $f$ is in the ideal of $Y\cap H$ but not it's square. I still don't see how this gives that the intersection multiplicity is one - I would need to know something along the lines of $\mathcal{O}_{Y,Y\cap H}$ having principal maximal ideal, which would mean it's a DVR? Still pretty stuck, though.

Or, if you think I'm barking up the wrong tree, is there some other way I'm supposed to finish problem 7.7(a)?

1

There are 1 best solutions below

2
On BEST ANSWER

In the language of the post, the local ring of $Y\cap H\cap U$ at $P$ is $\mathcal{O}_{Y,P}/\sqrt{(f)}=k[x_1,\cdots,x_n]_{(x_1,\cdots,x_n)}/\sqrt{(I_{Y\cap U}+(f))}_{(x_1,\cdots,x_n)}$, which is the reduction of $k[x_1,\cdots,x_n]_{(x_1,\cdots,x_n)}/(I_{Y\cap U}+(f))_{(x_1,\cdots,x_n)}$. But Hartshorne's theorem I.5.1 says that this last ring is actually a regular local ring by the Jacobian criterion and therefore reduced. So this is an equality, and $\mathcal{O}_{Y\cap U,P}/\sqrt{(f)}=\mathcal{O}_{Y\cap U,P}/(f)$. By the transitivity of localization, we have that $\mathcal{O}_{Y\cap U,Z\cap U}=(\mathcal{O}_{Y\cap U,P})_{I_{Z\cap U,P}}$, so the multiplicity of $k[U]/(I_{Y\cap U}+(f))$ is one. Since this is equal to the multiplicity of the homogeneous version, we have that the intersection multiplicity is one.