The Stacks project

\begin{equation*} \DeclareMathOperator\Coim{Coim} \DeclareMathOperator\Coker{Coker} \DeclareMathOperator\Ext{Ext} \DeclareMathOperator\Hom{Hom} \DeclareMathOperator\Im{Im} \DeclareMathOperator\Ker{Ker} \DeclareMathOperator\Mor{Mor} \DeclareMathOperator\Ob{Ob} \DeclareMathOperator\Sh{Sh} \DeclareMathOperator\SheafExt{\mathcal{E}\mathit{xt}} \DeclareMathOperator\SheafHom{\mathcal{H}\mathit{om}} \DeclareMathOperator\Spec{Spec} \newcommand\colim{\mathop{\mathrm{colim}}\nolimits} \newcommand\lim{\mathop{\mathrm{lim}}\nolimits} \newcommand\Qcoh{\mathit{Qcoh}} \newcommand\Sch{\mathit{Sch}} \newcommand\QCohstack{\mathcal{QC}\!\mathit{oh}} \newcommand\Cohstack{\mathcal{C}\!\mathit{oh}} \newcommand\Spacesstack{\mathcal{S}\!\mathit{paces}} \newcommand\Quotfunctor{\mathrm{Quot}} \newcommand\Hilbfunctor{\mathrm{Hilb}} \newcommand\Curvesstack{\mathcal{C}\!\mathit{urves}} \newcommand\Polarizedstack{\mathcal{P}\!\mathit{olarized}} \newcommand\Complexesstack{\mathcal{C}\!\mathit{omplexes}} \newcommand\Pic{\mathop{\mathrm{Pic}}\nolimits} \newcommand\Picardstack{\mathcal{P}\!\mathit{ic}} \newcommand\Picardfunctor{\mathrm{Pic}} \newcommand\Deformationcategory{\mathcal{D}\!\mathit{ef}} \end{equation*}

10.120 Orders of vanishing

Lemma 10.120.1. Let $R$ be a semi-local Noetherian ring of dimension $1$. If $a, b \in R$ are nonzerodivisors then

\[ \text{length}_ R(R/(ab)) = \text{length}_ R(R/(a)) + \text{length}_ R(R/(b)) \]

and these lengths are finite.

Proof. We saw the finiteness in Lemma 10.118.11. Additivity holds since there is a short exact sequence $0 \to R/(a) \to R/(ab) \to R/(b) \to 0$ where the first map is given by multiplication by $b$. (Use length is additive, see Lemma 10.51.3.) $\square$

Definition 10.120.2. Suppose that $K$ is a field, and $R \subset K$ is a local1 Noetherian subring of dimension $1$ with fraction field $K$. In this case we define the order of vanishing along $R$

\[ \text{ord}_ R : K^* \longrightarrow \mathbf{Z} \]

by the rule

\[ \text{ord}_ R(x) = \text{length}_ R(R/(x)) \]

if $x \in R$ and we set $\text{ord}_ R(x/y) = \text{ord}_ R(x) - \text{ord}_ R(y)$ for $x, y \in R$ both nonzero.

We can use the order of vanishing to compare lattices in a vector space. Here is the definition.

Definition 10.120.3. Let $R$ be a Noetherian local domain of dimension $1$ with fraction field $K$. Let $V$ be a finite dimensional $K$-vector space. A lattice in $V$ is a finite $R$-submodule $M \subset V$ such that $V = K \otimes _ R M$.

The condition $V = K \otimes _ R M$ signifies that $M$ contains a basis for the vector space $K$. We remark that in many places in the literature the notion of a lattice may be defined only in case the ring $R$ is a discrete valuation ring. If $R$ is a discrete valuation ring then any lattice is a free $R$-module, and this may not be the case in general.

Lemma 10.120.4. Let $R$ be a Noetherian local domain of dimension $1$ with fraction field $K$. Let $V$ be a finite dimensional $K$-vector space.

  1. If $M$ is a lattice in $V$ and $M \subset M' \subset V$ is an $R$-submodule of $V$ containing $M$ then the following are equivalent

    1. $M'$ is a lattice,

    2. $\text{length}_ R(M'/M)$ is finite, and

    3. $M'$ is finitely generated.

  2. If $M$ is a lattice in $V$ and $M' \subset M$ is an $R$-submodule of $M$ then $M'$ is a lattice if and only if $\text{length}_ R(M/M')$ is finite.

  3. If $M$, $M'$ are lattices in $V$, then so are $M \cap M'$ and $M + M'$.

  4. If $M \subset M' \subset M'' \subset V$ are lattices in $V$ then

    \[ \text{length}_ R(M''/M) = \text{length}_ R(M'/M) + \text{length}_ R(M''/M'). \]
  5. If $M$, $M'$, $N$, $N'$ are lattices in $V$ and $N \subset M \cap M'$, $M + M' \subset N'$, then we have

    \begin{eqnarray*} & & \text{length}_ R(M/M \cap M') - \text{length}_ R(M'/M \cap M')\\ & = & \text{length}_ R(M/N) - \text{length}_ R(M'/N) \\ & = & \text{length}_ R(M + M' / M') - \text{length}_ R(M + M'/M) \\ & = & \text{length}_ R(N' / M') - \text{length}_ R(N'/M) \end{eqnarray*}

Proof. Proof of (1). Assume (1)(a). Say $y_1, \ldots , y_ m$ generate $M'$. Then each $y_ i = x_ i/f_ i$ for some $x_ i \in M$ and nonzero $f_ i \in R$. Hence we see that $f_1 \ldots f_ m M' \subset M$. Since $R$ is Noetherian local of dimension $1$ we see that $\mathfrak m^ n \subset (f_1 \ldots f_ m)$ for some $n$ (for example combine Lemmas 10.59.12 and Proposition 10.59.6 or combine Lemmas 10.118.9 and 10.51.4). In other words $\mathfrak m^ nM' \subset M$ for some $n$ Hence $\text{length}(M'/M) < \infty $ by Lemma 10.51.8, in other words (1)(b) holds. Assume (1)(b). Then $M'/M$ is a finite $R$-module (see Lemma 10.51.2). Hence $M'$ is a finite $R$-module as an extension of finite $R$-modules. Hence (1)(c). The implication (1)(c) $\Rightarrow $ (1)(a) follows from the remark following Definition 10.120.3.

Proof of (2). Suppose $M$ is a lattice in $V$ and $M' \subset M$ is an $R$-submodule. We have seen in (1) that if $M'$ is a lattice, then $\text{length}_ R(M/M') < \infty $. Conversely, assume that $\text{length}_ R(M/M') < \infty $. Then $M'$ is finitely generated as $R$ is Noetherian and for some $n$ we have $\mathfrak m^ n M \subset M'$ (Lemma 10.51.4). Hence it follows that $M'$ contains a basis for $V$, and $M'$ is a lattice.

Proof of (3). Assume $M$, $M'$ are lattices in $V$. Since $R$ is Noetherian the submodule $M \cap M'$ of $M$ is finite. As $M$ is a lattice we can find $x_1, \ldots , x_ n \in M$ which form a $K$-basis for $V$. Because $M'$ is a lattice we can write $x_ i = y_ i/f_ i$ with $y_ i \in M'$ and $f_ i \in R$. Hence $f_ ix_ i \in M \cap M'$. Hence $M \cap M'$ is a lattice also. The fact that $M + M'$ is a lattice follows from part (1).

Part (4) follows from additivity of lengths (Lemma 10.51.3) and the exact sequence

\[ 0 \to M'/M \to M''/M \to M''/M' \to 0 \]

Part (5) follows from repeatedly applying part (4). $\square$

Definition 10.120.5. Let $R$ be a Noetherian local domain of dimension $1$ with fraction field $K$. Let $V$ be a finite dimensional $K$-vector space. Let $M$, $M'$ be two lattices in $V$. The distance between $M$ and $M'$ is the integer

\[ d(M, M') = \text{length}_ R(M/M \cap M') - \text{length}_ R(M'/M \cap M') \]

of Lemma 10.120.4 part (5).

In particular, if $M' \subset M$, then $d(M, M') = \text{length}_ R(M/M')$.

Lemma 10.120.6. Let $R$ be a Noetherian local domain of dimension $1$ with fraction field $K$. Let $V$ be a finite dimensional $K$-vector space. This distance function has the property that

\[ d(M, M'') = d(M, M') + d(M', M'') \]

whenever given three lattices $M$, $M'$, $M''$ of $V$. In particular we have $d(M, M') = - d(M', M)$.

Proof. Omitted. $\square$

Lemma 10.120.7. Let $R$ be a Noetherian local domain of dimension $1$ with fraction field $K$. Let $V$ be a finite dimensional $K$-vector space. Let $\varphi : V \to V$ be a $K$-linear isomorphism. For any lattice $M \subset V$ we have

\[ d(M, \varphi (M)) = \text{ord}_ R(\det (\varphi )) \]

Proof. We can see that the integer $d(M, \varphi (M))$ does not depend on the lattice $M$ as follows. Suppose that $M'$ is a second such lattice. Then we see that

\begin{eqnarray*} d(M, \varphi (M)) & = & d(M, M') + d(M', \varphi (M)) \\ & = & d(M, M') + d(\varphi (M'), \varphi (M)) + d(M', \varphi (M')) \end{eqnarray*}

Since $\varphi $ is an isomorphism we see that $d(\varphi (M'), \varphi (M)) = d(M', M) = -d(M, M')$, and hence $d(M, \varphi (M)) = d(M', \varphi (M'))$. Moreover, both sides of the equation (of the lemma) are additive in $\varphi $, i.e.,

\[ \text{ord}_ R(\det (\varphi \circ \psi )) = \text{ord}_ R(\det (\varphi )) + \text{ord}_ R(\det (\psi )) \]

and also

\begin{eqnarray*} d(M, \varphi (\psi ((M))) & = & d(M, \psi (M)) + d(\psi (M), \varphi (\psi (M))) \\ & = & d(M, \psi (M)) + d(M, \varphi (M)) \end{eqnarray*}

by the independence shown above. Hence it suffices to prove the lemma for generators of $\text{GL}(V)$. Choose an isomorphism $K^{\oplus n} \cong V$. Then $\text{GL}(V) = \text{GL}_ n(K)$ is generated by elementary matrices $E$. The result is clear for $E$ equal to the identity matrix. If $E = E_{ij}(\lambda )$ with $i \not= j$, $\lambda \in K$, $\lambda \not= 0$, for example

\[ E_{12}(\lambda ) = \left( \begin{matrix} 1 & \lambda & \ldots \\ 0 & 1 & \ldots \\ \ldots & \ldots & \ldots \end{matrix} \right) \]

then with respect to a different basis we get $E_{12}(1)$. The result is clear for $E = E_{12}(1)$ by taking as lattice $R^{\oplus n} \subset K^{\oplus n}$. Finally, if $E = E_ i(a)$, with $a \in K^*$ for example

\[ E_1(a) = \left( \begin{matrix} a & 0 & \ldots \\ 0 & 1 & \ldots \\ \ldots & \ldots & \ldots \end{matrix} \right) \]

then $E_1(a)(R^{\oplus b}) = aR \oplus R^{\oplus n - 1}$ and it is clear that $d(R^{\oplus n}, aR \oplus R^{\oplus n - 1}) = \text{ord}_ R(a)$ as desired. $\square$

Lemma 10.120.8. Let $A \to B$ be a ring map. Assume

  1. $A$ is a Noetherian local domain of dimension $1$,

  2. $A \subset B$ is a finite extension of domains.

Let $L/K$ be the corresponding finite extension of fraction fields. Let $y \in L^*$ and $x = \text{Nm}_{L/K}(y)$. In this situation $B$ is semi-local. Let $\mathfrak m_ i$, $i = 1, \ldots , n$ be the maximal ideals of $B$. Then

\[ \text{ord}_ A(x) = \sum \nolimits _ i [\kappa (\mathfrak m_ i) : \kappa (\mathfrak m_ A)] \text{ord}_{B_{\mathfrak m_ i}}(y) \]

where $\text{ord}$ is defined as in Definition 10.120.2.

Proof. The ring $B$ is semi-local by Lemma 10.112.2. Write $y = b/b'$ for some $b, b' \in B$. By the additivity of $\text{ord}$ and multiplicativity of $\text{Nm}$ it suffices to prove the lemma for $y = b$ or $y = b'$. In other words we may assume $y \in B$. In this case the right hand side of the formula is

\[ \sum [\kappa (\mathfrak m_ i) : \kappa (\mathfrak m_ A)] \text{length}_{B_{\mathfrak m_ i}}((B/yB)_{\mathfrak m_ i}) \]

By Lemma 10.51.12 this is equal to $\text{length}_ A(B/yB)$. By Lemma 10.120.7 we have

\[ \text{length}_ A(B/yB) = d(B, yB) = \text{ord}_ A(\det \nolimits _ K(L \xrightarrow {y} L)). \]

Since $x = \text{Nm}_{L/K}(y) = \det \nolimits _ K(L \xrightarrow {y} L)$ by definition the lemma is proved. $\square$

[1] We could also define this when $R$ is only semi-local but this is probably never really what you want!

Comments (2)

Comment #3509 by Dmitrii on

In the proof of Lemma 02MJ it says "In this case the left hand side of the formula"; I think this should be replaced by the "right hand side of the formula".


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 02MB. Beware of the difference between the letter 'O' and the digit '0'.