2
$\begingroup$

Assume that two smooth quasi-projective vareities $X,Y$ have the same class in the Grothendieck ring of varieties.

Do the symmetry products $X^{(n)}$ and $Y^{(n)}$ also have the same class?

Moreover, if there are some feasible criteria to judge if two varieties have the same class?

Edit: I leave this part of original the question as well as the answer as a complete closed post.

$\endgroup$
2
  • $\begingroup$ Please confer, "An exponential function on the set of varieties" by Gusein-Zade, Luengo, Melle-Hernandez: arxiv.org/pdf/math/0206279.pdf $\endgroup$ Apr 29, 2022 at 18:43
  • $\begingroup$ Actually, although the formulation by Gusein-Zade, et al., is novel, the result you ask seems to go back to Kapranov in his definition of the motivic zeta function. $\endgroup$ Apr 29, 2022 at 18:47

1 Answer 1

7
$\begingroup$

This is just a quick explanation. Let $k$ be an algebraically closed field. Let $\mathcal{Var}_k$ denote the set of (embedded) quasi-projective $k$-varieties. The Grothendieck group of varieties, $K_0(\mathcal{Var}_k)$, is the quotient of the free Abelian group on $\mathcal{Var}_k$ by all relations, $$[X] = [U]+[C],$$ if $C$ is $k$-isomorphic to a closed subvariety of $X$ whose open complement is $k$-isomorphic to $U$. This is actually a commutative ring under the product $[X]\cdot [Y] = [X\times_{\text{Spec}\ k} Y]$. The identity element is $1=[\text{Spec}\ k]$.

Inside the power series ring $K_0(\mathcal{Var}_k)[[t]]$, the subset of power series whose constant coefficient equals $1$ is an Abelian group under multiplication; call this group $G$.
For every $X$ in $\mathcal{Var}_k$, Kapranov defines an element in $G$, the motivic zeta function, $$\zeta_X(t) = 1 + \sum_{n\geq 1}[\text{Sym}^n(X)]t^n.$$ For every integer $n\geq 1$, there is a partition of $\text{Sym}^n(X)$ into locally closed subsets $Z_{n,m}$ for $m=0,\dots,n$, where $Z_{n,m}$ denotes the subvariety parameterizing length $n$ zero-cycles on $X$ whose intersection with $U$ has length precisely $m$. Then the "residual" zero-cycle is a length $n-m$ zero-cycle supported on $C$. In this way we get a $k$-isomorphism, $$Z_{n,m} \cong \text{Sym}^m(U)\times_{\text{Spec}\ k}\text{Sym}^{n-m}(C).$$ This is precisely the same as the identity, $$\zeta_X(t) = \zeta_U(t)\cdot \zeta_C(t).$$ Thus, the zeta function from $\mathcal{Var}_k$ to the Abelian group $G$ satisfies the relations to factor through a group homomorphism from $K_0(\mathcal{Var}_k)$, $$\zeta:K_0(\mathcal{Var}_k) \to (1+tK_0(\mathcal{Var}_k)[[t]])^\times.$$ In particular, the motivic zeta function $\zeta_X(t)$ only depends on the class of $X$ in $K_0(\mathcal{Var}_k)$. Consequently, for every integer $n\geq 1$, the class $[\text{Sym}^n(X)]$ in $K_0(\mathcal{Var}_k)$ only depends on the class of $X$ in $K_0(\mathcal{Var}_k)$.

$\endgroup$
2
  • 1
    $\begingroup$ "the set function" (after the fourth display) should be "the zeta function", I guess. $\endgroup$
    – Sasha
    Apr 30, 2022 at 11:59
  • $\begingroup$ Thanks @Sasha. I fixed it. $\endgroup$ Apr 30, 2022 at 13:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.