The Stacks project

45.8 Cycles over non-closed fields

Some lemmas which will help us in our study of motives over base fields which are not algebraically closed.

Lemma 45.8.1. Let $k$ be a field. Let $X$ be a smooth projective scheme over $k$. Then $\mathop{\mathrm{CH}}\nolimits _0(X)$ is generated by classes of closed points whose residue fields are separable over $k$.

Proof. The lemma is immediate if $k$ has characteristic $0$ or is perfect. Thus we may assume $k$ is an infinite field of characteristic $p > 0$.

We may assume $X$ is irreducible of dimension $d$. Then $k' = H^0(X, \mathcal{O}_ X)$ is a finite separable field extension of $k$ and that $X$ is geometrically integral over $k'$. See Varieties, Lemmas 33.25.4, 33.9.3, and 33.9.4. We may and do replace $k$ by $k'$ and assume that $X$ is geometrically integral.

Let $x \in X$ be a closed point. To prove the lemma we are going to show that $[x] \in \mathop{\mathrm{CH}}\nolimits _0(X)$ is rationally equivalent to an integer linear combination of classes of closed points whose residue fields are separable over $k$. Choose an ample invertible $\mathcal{O}_ X$-module $\mathcal{L}$. Set

\[ V = \{ s \in H^0(X, \mathcal{L}) \mid s(x) = 0 \} \]

After replacing $\mathcal{L}$ by a power we may assume (a) $\mathcal{L}$ is very ample, (b) $V$ generates $\mathcal{L}$ over $X \setminus x$, (c) the morphism $X \setminus x \to \mathbf{P}(V)$ is an immersion, (d) the map $V \to \mathfrak m_ x\mathcal{L}_ x/\mathfrak m_ x^2\mathcal{L}_ x$ is surjective, see Morphisms, Lemma 29.39.5, Varieties, Lemma 33.46.1, and Properties, Proposition 28.26.13. Consider the set

\[ V^ d \supset U = \{ (s_1, \ldots , s_ d) \in V^ d \mid s_1, \ldots , s_ d \text{ generate } \mathfrak m_ x\mathcal{L}_ x/\mathfrak m_ x^2\mathcal{L}_ x \text{ over }\kappa (x) \} \]

Since $\mathcal{O}_{X, x}$ is a regular local ring of dimension $d$ we have $\dim _{\kappa (x)}(\mathfrak m_ x/\mathfrak m_ x^2) = d$ and hence we see that $U$ is a nonempty (Zariski) open of $V^ d$. For $(s_1, \ldots , s_ d) \in U$ set $H_ i = Z(s_ i)$. Since $s_1, \ldots , s_ d$ generate $\mathfrak m_ x\mathcal{L}_ x$ we see that

\[ H_1 \cap \ldots \cap H_ d = x \amalg Z \]

scheme theoretically for some closed subscheme $Z \subset X$. By Bertini (in the form of Varieties, Lemma 33.46.3) for a general element $s_1 \in V$ the scheme $H_1 \cap (X \setminus x)$ is smooth over $k$ of dimension $d - 1$. Having chosen $s_1$, for a general element $s_2 \in V$ the scheme $H_1 \cap H_2 \cap (X \setminus x)$ is smooth over $k$ of dimension $d - 2$. And so on. We conclude that for sufficiently general $(s_1, \ldots , s_ d) \in U$ the scheme $Z$ is étale over $\mathop{\mathrm{Spec}}(k)$. In particular $H_1 \cap \ldots \cap H_ d$ has dimension $0$ and hence

\[ [H_1] \cdot \ldots \cdot [H_ d] = [x] + [Z] \]

in $\mathop{\mathrm{CH}}\nolimits _0(X)$ by repeated application of Chow Homology, Lemma 42.61.5 (details omitted). This finishes the proof as it shows that $[x] \sim _{rat} - [Z] + [Z']$ where $Z' = H'_1 \cap \ldots \cap H'_ d$ is a general complete intersection of vanishing loci of sufficiently general sections of $\mathcal{L}$ which will be étale over $k$ by the same argument as before. $\square$

Lemma 45.8.2. Let $K/k$ be an algebraic field extension. Let $X$ be a finite type scheme over $k$. Then $\mathop{\mathrm{CH}}\nolimits _ i(X_ K) = \mathop{\mathrm{colim}}\nolimits \mathop{\mathrm{CH}}\nolimits _ i(X_{k'})$ where the colimit is over the subextensions $K/k'/k$ with $k'/k$ finite.

Proof. This is a special case of Chow Homology, Lemma 42.66.10. $\square$

Lemma 45.8.3. Let $k$ be a field. Let $X$ be a geometrically irreducible smooth projective scheme over $k$. Let $x, x' \in X$ be $k$-rational points. Let $n$ be an integer invertible in $k$. Then there exists a finite separable extension $k'/k$ such that the pullback of $[x] - [x']$ to $X_{k'}$ is divisible by $n$ in $\mathop{\mathrm{CH}}\nolimits _0(X_{k'})$.

Proof. Let $k'$ be a separable algebraic closure of $k$. Suppose that we can show the the pullback of $[x] - [x']$ to $X_{k'}$ is divisible by $n$ in $\mathop{\mathrm{CH}}\nolimits _0(X_{k'})$. Then we conclude by Lemma 45.8.2. Thus we may and do assume $k$ is separably algebraically closed.

Suppose $\dim (X) > 1$. Let $\mathcal{L}$ be an ample invertible sheaf on $X$. Set

\[ V = \{ s \in H^0(X, \mathcal{L}) \mid s(x) = 0\text{ and }s(x') = 0 \} \]

After replacing $\mathcal{L}$ by a power we see that for a general $v \in V$ the corresponding divisor $H_ v \subset X$ is smooth away from $x$ and $x'$, see Varieties, Lemmas 33.46.1 and 33.46.3. To find $v$ we use that $k$ is infinite (being separably algebraically closed). If we choose $s$ general, then the image of $s$ in $\mathfrak m_ x\mathcal{L}_ x/\mathfrak m_ x^2\mathcal{L}_ x$ will be nonzero, which implies that $H_ v$ is smooth at $x$ (details omitted). Similarly for $x'$. Thus $H_ v$ is smooth. By Varieties, Lemma 33.47.3 (applied to the base change of everything to the algebraic closure of $k$) we see that $H_ v$ is geometrically connected. It suffices to prove the result for $[x] - [x']$ seen as an element of $\mathop{\mathrm{CH}}\nolimits _0(H_ v)$. In this way we reduce to the case of a curve.

Assume $X$ is a curve. Then we see that $\mathcal{O}_ X(x - x')$ defines a $k$-rational point $g$ of $J = \underline{\mathop{\mathrm{Pic}}\nolimits }^0_{X/k}$, see Picard Schemes of Curves, Lemma 44.6.7. Recall that $J$ is a proper smooth variety over $k$ which is also a group scheme over $k$ (same reference). Hence $J$ is geometrically integral (see Varieties, Lemma 33.7.13 and 33.25.4). In other words, $J$ is an abelian variety, see Groupoids, Definition 39.9.1. Thus $[n] : J \to J$ is finite étale by Groupoids, Proposition 39.9.11 (this is where we use $n$ is invertible in $k$). Since $k$ is separably closed we conclude that $g = [n](g')$ for some $g' \in J(k)$. If $\mathcal{L}$ is the degree $0$ invertible module on $X$ corresponding to $g'$, then we conclude that $\mathcal{O}_ X(x - x') \cong \mathcal{L}^{\otimes n}$ as desired. $\square$

Lemma 45.8.4. Let $K/k$ be an algebraic extension of fields. Let $X$ be a finite type scheme over $k$. The kernel of the map $\mathop{\mathrm{CH}}\nolimits _ i(X) \to \mathop{\mathrm{CH}}\nolimits _ i(X_ K)$ constructed in Lemma 45.8.2 is torsion.

Proof. It clearly suffices to show that the kernel of flat pullback $\mathop{\mathrm{CH}}\nolimits _ i(X) \to \mathop{\mathrm{CH}}\nolimits _ i(X_{k'})$ by $\pi : X_{k'} \to X$ is torsion for any finite extension $k'/k$. This is clear because $\pi _* \pi ^* \alpha = [k' : k] \alpha $ by Chow Homology, Lemma 42.15.2. $\square$

reference

Lemma 45.8.5 (Voevodsky). Let $k$ be a field. Let $X$ be a geometrically irreducible smooth projective scheme over $k$. Let $x, x' \in X$ be $k$-rational points. For $n$ large enough the class of the zero cycle

\[ ([x] - [x']) \times \ldots \times ([x] - [x']) \in \mathop{\mathrm{CH}}\nolimits _0(X^ n) \]

is torsion.

Proof. If we can show this after base change to the algebraic closure of $k$, then the result follows over $k$ because the kernel of pullback is torsion by Lemma 45.8.4. Hence we may and do assume $k$ is algebraically closed.

Using Bertini we can choose a smooth curve $C \subset X$ passing through $x$ and $x'$. See proof of Lemma 45.8.3. Hence we may assume $X$ is a curve.

Assume $X$ is a curve and $k$ is algebraically closed. Write $S^ n(X) = \underline{\mathrm{Hilb}}^ n_{X/k}$ with notation as in Picard Schemes of Curves, Sections 44.2 and 44.3. There is a canonical morphism

\[ \pi : X^ n \longrightarrow S^ n(X) \]

which sends the $k$-rational point $(x_1, \ldots , x_ n)$ to the $k$-rational point corresponding to the divisor $[x_1] + \ldots + [x_ n]$ on $X$. There is a faithful action of the symmetric group $S_ n$ on $X^ n$. The morphism $\pi $ is $S_ n$-invariant and the fibres of $\pi $ are $S_ n$-orbits (set theoretically). Finally, $\pi $ is finite flat of degree $n!$, see Picard Schemes of Curves, Lemma 44.3.4.

Let $\alpha _ n$ be the zero cycle on $X^ n$ given by the formula in the statement of the lemma. Let $\mathcal{L} = \mathcal{O}_ X(x - x')$. Then $c_1(\mathcal{L}) \cap [X] = [x] - [x']$. Thus

\[ \alpha _ n = c_1(\mathcal{L}_1) \cap \ldots \cap c_1(\mathcal{L}_ n) \cap [X^ n] \]

where $\mathcal{L}_ i = \text{pr}_ i^*\mathcal{L}$ and $\text{pr}_ i : X^ n \to X$ is the $i$th projection. By either Divisors, Lemma 31.17.6 or Divisors, Lemma 31.17.7 there is a norm for $\pi $. Set $\mathcal{N} = \text{Norm}_\pi (\mathcal{L}_1)$, see Divisors, Lemma 31.17.2. We have

\[ \pi ^*\mathcal{N} = (\mathcal{L}_1 \otimes \ldots \otimes \mathcal{L}_ n)^{\otimes (n - 1)!} \]

in $\mathop{\mathrm{Pic}}\nolimits (X^ n)$ by a calculation. Deails omitted; hint: this follows from the fact that $\text{Norm}_\pi : \pi _*\mathcal{O}_{X^ n} \to \mathcal{O}_{S^ n(X)}$ composed with the natural map $\pi _*\mathcal{O}_{S^ n(X)} \to \mathcal{O}_{X^ n}$ is equal to the product over all $\sigma \in S_ n$ of the action of $\sigma $ on $\pi _*\mathcal{O}_{X^ n}$. Consider

\[ \beta _ n = c_1(\mathcal{N})^ n \cap [S^ n(X)] \]

in $\mathop{\mathrm{CH}}\nolimits _0(S^ n(X))$. Observe that $c_1(\mathcal{L}_ i) \cap c_1(\mathcal{L}_ i) = 0$ because $\mathcal{L}_ i$ is pulled back from a curve, see Chow Homology, Lemma 42.33.6. Thus we see that

\begin{align*} \pi ^*\beta _ n & = ((n - 1)!)^ n (\sum \nolimits _{i = 1, \ldots , n} c_1(\mathcal{L}_ i))^ n \cap [X^ n] \\ & = ((n - 1)!)^ n n^ n c_1(\mathcal{L}_1) \cap \ldots \cap c_1(\mathcal{L}_ n) \cap [X^ n] \\ & = (n!)^ n \alpha _ n \end{align*}

Thus it suffices to show that $\beta _ n$ is torsion.

There is a canonical morphism

\[ f : S^ n(X) \longrightarrow \underline{\mathrm{Pic}}^ n_{X/k} \]

See Picard Schemes of Curves, Lemma 44.6.7. For $n \geq 2g - 1$ this morphism is a projective space bundle (details omitted; compare with the proof of Picard Schemes of Curves, Lemma 44.6.7). The invertible sheaf $\mathcal{N}$ is trivial on the fibres of $f$, see below. Thus by the projective space bundle formula (Chow Homology, Lemma 42.35.2) we see that $\mathcal{N} = f^*\mathcal{M}$ for some invertible module $\mathcal{M}$ on $\underline{\mathrm{Pic}}^ n_{X/k}$. Of course, then we see that

\[ c_1(\mathcal{N})^ n = f^*(c_1(\mathcal{M})^ n) \]

is zero because $n > g = \dim (\underline{\mathrm{Pic}}^ n_{X/k})$ and we can use Chow Homology, Lemma 42.33.6 as before.

We still have to show that $\mathcal{N}$ is trivial on a fibre $F$ of $f$. Since the fibres of $f$ are projective spaces and since $\mathop{\mathrm{Pic}}\nolimits (\mathbf{P}^ m_ k) = \mathbf{Z}$ (Divisors, Lemma 31.28.5), this can be shown by computing the degree of $\mathcal{N}$ on a line contained in the fibre. Instead we will prove it by proving that $\mathcal{N}$ is algebraically equivalent to zero. First we claim there is a connected finite type scheme $T$ over $k$, an invertible module $\mathcal{L}'$ on $T \times X$ and $k$-rational points $p, q \in T$ such that $\mathcal{M}_ p \cong \mathcal{O}_ X$ and $\mathcal{M}_ q = \mathcal{L}$. Namely, since $\mathcal{L} = \mathcal{O}_ X(x - x')$ we can take $T = X$, $p = x'$, $q = x$, and $\mathcal{L}' = \mathcal{O}_{X \times X}(\Delta ) \otimes \text{pr}_2^*\mathcal{O}_ X(-x')$. Then we let $\mathcal{L}'_ i$ on $T \times X^ n$ for $i = 1, \ldots , n$ be the pullback of $\mathcal{L}'$ by $\text{id}_ T \times \text{pr}_ i : T \times X^ n \to T \times X$. Finally, we let $\mathcal{N}' = \text{Norm}_{\text{id}_ T \times \pi }(\mathcal{L}'_1)$ on $T \times S^ n(X)$. By construction we have $\mathcal{N}'_ p = \mathcal{O}_{S^ n(X)}$ and $\mathcal{N}'_ q = \mathcal{N}$. We conclude that

\[ \mathcal{N}'|_{T \times F} \]

is an invertible module on $T \times F \cong T \times \mathbf{P}^ m_ k$ whose fibre over $p$ is the trivial invertible module and whose fibre over $q$ is $\mathcal{N}|_ F$. Since the euler characteristic of the trivial bundle is $1$ and since this euler characteristic is locally constant in families (Derived Categories of Schemes, Lemma 36.32.2) we conclude $\chi (F, \mathcal{N}^{\otimes s}|_ F) = 1$ for all $s \in \mathbf{Z}$. This can happen only if $\mathcal{N}|_ F \cong \mathcal{O}_ F$ (see Cohomology of Schemes, Lemma 30.8.1) and the proof is complete. Some details omitted. $\square$


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 0FH4. Beware of the difference between the letter 'O' and the digit '0'.