3
$\begingroup$

Let $\mathcal M_C$ denote the moduli space of rank two vector bundles over a smooth proper curve $C$ over an algebraically closed field $k$ of characteristic zero. Let $k(t)$ denote the function field of $\mathbb P^1_k$, and let $\overline{k(t)}$ denote a fixed algebraic closure of $k(t)$.

I would like to consider the direct limit $$\mathcal M:=\varinjlim_{k(t)\subseteq k(C)\subseteq\overline{k(t)}}\mathcal M_C$$ where the transition maps are the obvious pullback maps $\mathcal M_{C'}\to\mathcal M_C$ associated to maps of curves $C\to C'$ for $k(C')\subseteq k(C)$. Note that it doesn't matter that we started with $\mathbb P^1_k$; we could just as well have taken the direct limit over $k(C_0)\subseteq k(C)\subseteq\overline{k(C_0)}$ for any fixed curve $C_0/k$ and fixed algebraic closure of its function field, and it would give us the same moduli space $\mathcal M$.

Has this moduli space $\mathcal M$ been studied before? Can anything nontrivial/interesting be said about it?

I have been deliberately vague about in which sense I mean "moduli space of rank two vector bundles" (i.e. whether I mean a stack, or the GIT construction, or whether we impose stability conditions on our vector bundles, etc.), because I would be happy with an answer which says something interesting about any of these possible settings.

EDIT: I know a reasonable amount about each individual $\mathcal M_C$, and that others know even more. The point of this question is to say something nontrivial about the direct limit $\mathcal M$.

EDIT2: I am happy to restrict to bundles of trivial determinant, or, what is almost the same, to consider bundles $E$ and $E\otimes\mathcal L$ to be equivalent for any line bundle $\mathcal L$.

$\endgroup$
12
  • $\begingroup$ Sorry, your question is not clear,at least put some additional words with more explanation, even for those who can not answer be able to follow your question $\endgroup$
    – user21574
    Commented Jan 18, 2017 at 22:19
  • $\begingroup$ I know that such moduli spaces studied by Narasimhan in 60 and he charactrized them. They are Fano when genus is greater than 1 .May you add some reference or you want to consider quasi-projectivity of them. So certainly it correspond to Lelong number of their canonical metric as Weil-Petersson metric(if you take K-stability notion related to kahler-Einstein metric). $\endgroup$
    – user21574
    Commented Jan 18, 2017 at 22:31
  • $\begingroup$ For every (say, non constant) map $u:C \to C'$ you get a "pullback" morphism $\mathcal{M}_{C'}\to \mathcal{M}_{C}, \; E\mapsto u^* E$. Then you want to form the limit $lim_{\to}\mathcal{M}_C$ w.r.t. the directed set $\{C\to C'\}^{\mathrm{op}}$. Is it correct? In which category are you expecting such a limit to exist? $\endgroup$
    – Qfwfq
    Commented Jan 18, 2017 at 22:47
  • $\begingroup$ @Qfwfq: For the direct limit to be well-behaved, one can't consider simply all curves $C$ and all maps $C\to C'$, rather one needs to "rigidify" things appropriately. This is the reason for only considering (the smooth proper curves corresponding to) function fields over $k(t)$ contained in a fixed algebraic closure $\overline{k(t)}$. Now this is a directed set. The direct limit is pretty trivially an ind-(algebraic stack) over $k$. $\endgroup$ Commented Jan 19, 2017 at 0:31
  • 1
    $\begingroup$ John -- perhaps you can tell us what sort of thing you have in mind for rank 1 bundles? Here we get some crazy direct limit of Jacobians x Z; I don't know if one can say much more than that. Of course the direct limit you have in mind fibers over this (by taking the determinant); so let's say we look at bundles with trivial determinant. This is still probably pretty crazy, but I suppose one can compute things like the cohomology of this ind-stack... $\endgroup$ Commented Jan 19, 2017 at 4:03

0

You must log in to answer this question.