Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

I'm trying to understand Example 1.7.4 in Fulton's "Intersection Theory", asserting that if a morphism of schemes (of finite type over a field) $f:X\to Y$ is flat and finite of degree $d$, then for every subvariety $V\subset Y$ one has $f_\ast f^\ast[V]=d[V]$. My calculation leads to the equality \begin{equation} f_\ast f^\ast[V]=\sum_{W\subset f^{-1}(V)}n_W\deg(W/f(W))[f(W)], \end{equation} where the sum is on the irreducible components $W$ of $f^{-1}(V)$ and $n_W=\ell(\mathscr O_{f^{-1}(V),W})$. Because $f^{-1}(V)\to V$ is flat and $V$ is irreducible, every $W$ dominates $V$. But as $f$ is finite (hence closed) we get $f(W)=V$ for all $W$, so we can write \begin{equation} f_\ast f^\ast[V]=\sum_{W\subset f^{-1}(V)}n_W\deg(W/V)[V]. \end{equation} Now, we are left to show that $\sum_Wn_W\deg(W/V)=d$. I know that if I have a local ring $A$ and an $A$-algebra $B\cong A^d$, then $d=\sum_{\mathfrak n\in\textrm{Spm}\,B}[B_\mathfrak n/\mathfrak nB_\mathfrak n:A/\mathfrak m_A]\ell_{B_\mathfrak n}(B_\mathfrak n/\mathfrak m_AB_\mathfrak n)$. Hence, to conclude I'd like to take $A=\mathscr O_{Y,V}$, but $\textbf{what}$ $B$ $\textbf{do I have to choose?}$ I'd like to interpret $n_W$ as the length appearing in the sum (so I need a correspondence $\mathfrak n\leftrightarrow W$) and to recover $\deg(W/V):=[R(W):R(V)]$ as the degree $[B_\mathfrak n/\mathfrak nB_\mathfrak n:A/\mathfrak m_A]$.

Also, can you show me an $\textbf{example}$ of this result?

Thank you in advance.

share|cite|improve this question
A nice example to work out is to consider a ramified cover of Riemann surfaces (curves over $\mathbb{C}$), $\pi: C_1\to C_2$. The reason to pick something ramified is to see what is going on in the formula. For all the unramified points $p\in C_2$, when you pull back you get $d$ distinct points so everything is fine. At ramified points the pullback gives you less than $d$ distinct points set-theoretically, but because you remember more information scheme theoretically the points that "came together" will have multiplicity and hence give the right formula again. – Matt Sep 29 '12 at 16:48
OK @Matt, I see. So then, if $\pi$ is unramified then I can take any $y\in C_2$ and get $d=\sum_{x\in \pi^{-1}y}[k(x):k(y)]=|\pi^{-1}(y)|$. And if $\pi$ ramifies (necessarily at a finite set $S\subset C_2$) I get, for all $y\in S$, $d=\sum_{x\in \pi^{-1}y}e_{x/y}$, right? Question: when we say $\pi$ ramifies at $y\in C_2$, does it mean that for $\textbf{every}$ $x\in\pi^{-1}(y)$ one has $e_{x/y}>1$? Or may this hold just for some $x$? – Brenin Sep 30 '12 at 9:21
Just fir some $x$'s : take $C_1=C_2=\mathbb C$ and $\pi(z)=z(z-1)^2$. Then above the critical value $y=0$ the point $x=1$ is a ramification point but $x=0$ is not. – Georges Elencwajg Sep 30 '12 at 16:54
@atricolf My terminology wasn't quite standard. I should have called $p\in C_2$ for which some $\pi^{-1}(p)$ are ramified a branch point. The pullback $\pi^*(p)=\sum_{x\in \pi^{-1}p} e_{x/p}x$ and only some of the $x$ will necessarily have $e_{x/y}>1$. It is true that $d=\sum e_{x/y}$. – Matt Sep 30 '12 at 21:16
up vote 2 down vote accepted

Let $\xi$ be the generic point of $V$. Then $O_{Y,V}=O_{Y,\xi}$. Let $U$ be an affine open neighborhood of $\xi$. Then take $B=A\otimes_{O_Y(U)} O_X(f^{-1}(U))$.

share|cite|improve this answer
Thank you, @QiL! I see that $B\cong A^d$, but: $\textbf{(1)}$ I don't see why a maximal ideal $\mathfrak n\subset B$ should correspond to an irreducible component $W$ of $f^{-1}(V)$. $\textbf{(2)}$ Let us suppose that $\mathfrak n\leftrightarrow W$ (so that the local ring of $W$ in $f^{-1}(V)$ is $B_{\mathfrak n}$). The multiplicity $n_W$ is defined as $\ell_{B_{\mathfrak n}}(B_{\mathfrak n})$, but the corresponding term in the sum is $\ell_{B_{\mathfrak n}}(B_{\mathfrak n}/\mathfrak m_AB_{\mathfrak n})=:e_{B_{\mathfrak n}/A}$. Is there any problem here? That is, are these numbers the same? – Brenin Oct 1 '12 at 19:48
@atricolf: (1) The maximal ideals in $B$ are over the maximal ideal $A$ by the finiteness of $B$ over $A$. So the corresponding points are over the generic points of $V$. As $f$ is flat, $f^{-1}(V)\to V$ is flat, so the irreducible components of $f^{-1}(V)$ are the Zariski closure of the points of the generic fiber. (2) non the multiplicity is not the length of $B_{\mathfrak n}$ (which is not of finite length!), but the length of $B_{\mathfrak n}/(\mathfrak m_A)$. – user18119 Oct 1 '12 at 22:36

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.