The Stacks project

Lemma 82.13.2. In Situation 82.2.1 let $f : X \to Y$ be an étale morphism of good algebraic spaces over $B$. Assume $Y$ is integral. Let $g \in R(Y)^*$. As cycles on $X$ we have

\[ f^*(\text{div}_ Y(g)) = \sum \nolimits _{X'} (X' \to X)_*\text{div}_{X'}(g \circ f|_{X'}) \]

where the sum is over the irreducible components of $X$ (Remark 82.5.1).

Proof. The map $|X| \to |Y|$ is open. The set of irreducible components of $|X|$ is locally finite in $|X|$. We conclude that $f|_{X'} : X' \to Y$ is dominant for every irreducible component $X' \subset X$. Thus $g \circ f|_{X'}$ is defined (Morphisms of Spaces, Section 67.47), hence $\text{div}_{X'}(g \circ f|_{X'})$ is defined. Moreover, the sum is locally finite and we find that the right hand side indeed is a cycle on $X$. The left hand side is defined by Definition 82.10.1 and the fact that an étale morphism is flat of relative dimension $0$.

Since $f$ is étale we see that $\delta _ X(x) = \delta _ y(f(x))$ for all $x \in |X|$. Thus if $\dim _\delta (Y) = n$, then $\dim _\delta (X') = n$ for every irreducible component $X'$ of $X$ (since generic points of $X$ map to the generic point of $Y$, see above). Thus both left and right hand side are $(n - 1)$-cycles.

Let $Z \subset X$ be an integral closed subspace with $\dim _\delta (Z) = n - 1$. To prove the equality, we need to show that the coefficients of $Z$ are the same. Let $Z' \subset Y$ be the integral closed subspace constructed in Lemma 82.7.1. Then $\dim _\delta (Z') = n - 1$ too. Let $\xi \in |Z|$ be the generic point. Then $\xi ' = f(\xi ) \in |Z'|$ is the generic point. Consider the commutative diagram

\[ \xymatrix{ \mathop{\mathrm{Spec}}(\mathcal{O}_{X, \xi }^ h) \ar[r] \ar[d] & X \ar[d] \\ \mathop{\mathrm{Spec}}(\mathcal{O}_{Y, \xi '}^ h) \ar[r] & Y } \]

of Decent Spaces, Remark 68.11.11. We have to be slightly careful as the reduced Noetherian local rings $\mathcal{O}_{X, \xi }^ h$ and $\mathcal{O}_{Y, \xi '}^ h$ need not be domains. Thus we work with total rings of fractions $Q(-)$ rather than fraction fields. By definition, to get the coefficient of $Z'$ in $\text{div}_ Y(g)$ we write the image of $g$ in $Q(\mathcal{O}_{Y, \xi '}^ h)$ as $a/b$ with $a, b \in \mathcal{O}_{Y, \xi '}^ h$ nonzerodivisors and we take

\[ \text{ord}_{Z'}(g) = \text{length}_{\mathcal{O}_{Y, \xi '}^ h} (\mathcal{O}_{Y, \xi '}^ h/a \mathcal{O}_{Y, \xi '}^ h) - \text{length}_{\mathcal{O}_{Y, \xi '}^ h} (\mathcal{O}_{Y, \xi '}^ h/b \mathcal{O}_{Y, \xi '}^ h) \]

Observe that the coefficient of $Z$ in $f^*\text{div}_ Y(G)$ is the same integer, see Lemma 82.10.3. Suppose that $\xi \in X'$. Then we can consider the maps

\[ \mathcal{O}_{Y, \xi '}^ h \to \mathcal{O}_{X, \xi }^ h \to \mathcal{O}_{X', \xi }^ h \]

The first arrow is flat and the second arrow is a surjective map of reduced local Noetherian rings of dimension $1$. Therefore both these maps send nonzerodivisors to nonzerodivisors and we conclude the coefficient of $Z'$ in $\text{div}_{X'}(g \circ f|_{X'})$ is

\[ \text{ord}_ Z(g \circ f|_{X'}) = \text{length}_{\mathcal{O}_{X', \xi }^ h} (\mathcal{O}_{X', \xi }^ h/a \mathcal{O}_{X', \xi }^ h) - \text{length}_{\mathcal{O}_{Y, \xi '}^ h} (\mathcal{O}_{X', \xi }^ h/b \mathcal{O}_{X', \xi }^ h) \]

by the same prescription as above. Thus it suffices to show

\[ \text{length}_{\mathcal{O}_{Y, \xi '}^ h} (\mathcal{O}_{Y, \xi '}^ h/a \mathcal{O}_{Y, \xi '}^ h) = \sum \nolimits _{\xi \in |X'|} \text{length}_{\mathcal{O}_{X', \xi }^ h} (\mathcal{O}_{X', \xi }^ h/a \mathcal{O}_{X', \xi }^ h) \]

First, since the ring map $\mathcal{O}_{Y, \xi '}^ h \to \mathcal{O}_{X, \xi }^ h$ is flat and unramified, we have

\[ \text{length}_{\mathcal{O}_{Y, \xi '}^ h} (\mathcal{O}_{Y, \xi '}^ h/a \mathcal{O}_{Y, \xi '}^ h) = \text{length}_{\mathcal{O}_{X, \xi }^ h} (\mathcal{O}_{X, \xi }^ h/a \mathcal{O}_{X, \xi }^ h) \]

by Algebra, Lemma 10.52.13. Let $\mathfrak q_1, \ldots , \mathfrak q_ t$ be the nonmaximal primes of $\mathcal{O}_{X, \xi }^ h$ and set $R_ j = \mathcal{O}_{X, \xi }^ h/\mathfrak q_ j$. For $X'$ as above, denote $J(X') \subset \{ 1, \ldots , t\} $ the set of indices such that $\mathfrak q_ j$ corresponds to a point of $X'$, i.e., such that under the surjection $\mathcal{O}_{X, \xi }^ h \to \mathcal{O}_{X', \xi }$ the prime $\mathfrak q_ j$ corresponds to a prime of $\mathcal{O}_{X', \xi }$. By Chow Homology, Lemma 42.3.2 we get

\[ \text{length}_{\mathcal{O}_{X, \xi }^ h} (\mathcal{O}_{X, \xi }^ h/a \mathcal{O}_{X, \xi }^ h) = \sum \nolimits _ j \text{length}_{R_ j}(R_ j/a R_ j) \]

and

\[ \text{length}_{\mathcal{O}_{X', \xi }^ h} (\mathcal{O}_{X', \xi }^ h/a \mathcal{O}_{X', \xi }^ h) = \sum \nolimits _{j \in J(X')} \text{length}_{R_ j}(R_ j/a R_ j) \]

Thus the result of the lemma holds because $\{ 1, \ldots , t\} $ is the disjoint union of the sets $J(X')$: each point of codimension $0$ on $X$ lies on a unique $X'$. $\square$


Comments (0)


Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.




In order to prevent bots from posting comments, we would like you to prove that you are human. You can do this by filling in the name of the current tag in the following input field. As a reminder, this is tag 0EQ2. Beware of the difference between the letter 'O' and the digit '0'.