Let $C$ be a curve (=smooth projective curve) of genus $g$ over an algebraic closed field $\mathbb{k}$. It is well known that the Grothendieck group $K_0(\operatorname{coh} C)$ of the category of coherent sheaves on $C$ is the following (see, for example, Exercise 6.11 of Hartshorne): $$ K_0(\operatorname{coh} C)\simeq \operatorname{Pic} C \times \mathbb{Z} \simeq \operatorname{Pic}_0 C \times \mathbb{Z}^2, $$ where $\operatorname{Pic} C$ is the Picard group of $C$ and $\operatorname{Pic}_0 C$ is the kernel of the degree map $\mathrm{deg} : \operatorname{Pic} C\to \mathbb{Z}$. In the case $\mathbb{k} = \mathbb{C}$, we also know that $$ \operatorname{Pic}_0 C \simeq \mathbb{C}^g/\Lambda, $$ where $\Lambda$ is a $\mathbb{Z}$-submodule in $\mathbb{C}^g$ of rank $2g$.

My Questions:

  1. For two curves $C_1$ and $C_2$, if $K_0(\operatorname{coh} C_1) \simeq K_0(\operatorname{coh} C_2)$ as abstract groups, then do we conclude that $\operatorname{Pic}_0 C_1 \simeq \operatorname{Pic}_0 C_2$ (or $\operatorname{Pic} C_1 \simeq \operatorname{Pic} C_2$) as abstract groups ? Here abstract groups means groups without additional structure, such as structure of topology, variety and so on.
  2. Does an isomorphism $\operatorname{Pic}_0 C_1 \simeq \operatorname{Pic}_0 C_2$ of abstract groups contain some geometric information about $C_1$ and $C_2$?
  • 1
    $\begingroup$ As an abstract group $Pic^0(C)$ is the same for every genus $g$ curve, because all complex tori of the same dimension are diffeomorphic? On the other hand, genus is recovered, as dimension of the $p$-torsion of this group (over $\mathbb{F}_p$) equals $2g$. $\endgroup$ Jun 1, 2022 at 8:37

1 Answer 1


I don't know about 2., but the answer to 1. is yes.

More generally, suppose $(S^1)^g\times\mathbb Z^k \cong (S^1)^h \times \mathbb Z^m$ as abstract groups, then $g=h, k=m$ (in fact, you only need $g=h$ for question 1., because the abstract isomorphism type of $\mathbb C^g/\Lambda$ only depends on $g$, and because you already know $k=m=2$)

(Note that this gives part of the answer for 2., namely that $C_1$ and $C_2$ have the same genus)

Let's prove this statement.

First, if there is an isomorphism between the two groups, then their torsion parts are isomorphic. Their torsion parts are $(\mathbb{Q/Z})^g$ and $(\mathbb{Q/Z})^h$ respectively. But you can recover $g$ from $(\mathbb{Q/Z})^g$ in the following way : take the $2$-torsion (not $2$-power torsion, $2$-torsion) part as an $\mathbb F_2$-vector space, this has dimension $g$. (I chose $2$ but could have chosen any $p$, in fact for any integer $n\geq 2$, the $n$-torsion part is a free $\mathbb Z/n$-module of rank $g$, and these rings have the invariant basis property).

Therefore $g=h$ - this is all that's relevant to your question, but let me still prove $k=m$.

Note that to get $k=m$, we cannot rationalize and take dimensions because $(S^1)^g\otimes \mathbb Q$ is an uncounatbly dimensional rational vector space.

We now observe that we can mod out the divisible part - I don't know the standard name, but let me define it : $\mathrm{Div}(A) = \{a\in A\mid \forall n \neq 0, \exists b \in A, nb = a\}$. This is clearly a subgroup of $A$, and $\mathrm{Div}(A\times B)\cong \mathrm{Div}(A)\times \mathrm{Div}(B)$. In particular $\mathrm{Div}((S^1)^g\times \mathbb Z^k) = (S^1)^g\times \{0\}$, and so modding out $A/\mathrm{Div}(A)$ gives, in our case, an isomorphism $\mathbb Z^k\cong \mathbb Z^m$, and therefore $k=m$.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.