A stacky approach to identifying the semistable locus of bundles
Dario Weissmann, Xucheng Zhang
TL;DR
This work develops a stack-theoretic, moduli-space–driven lens for identifying the semistable locus of principal bundles over a curve. By translating stability into the existence of schematic or adequate moduli spaces and leveraging Alper’s framework, the authors show that the semistable locus $\mathscr{B}un_G^{ss}$ is the unique maximal open substack admitting a schematic (or adequate) moduli space in characteristic zero (and generally for adequate moduli spaces in arbitrary characteristic). They further analyze rank 2 vector bundles to prove a sharp maximality result, while constructing higher-rank counterexamples demonstrating that separated moduli spaces alone do not recover the classical semistability. The approach hinges on translating existence criteria into concrete stability notions ($\Theta$-reductivity, S-completeness, local reductivity) and employing central isogenies to reduce to almost simple cases, yielding both negative and positive results about when a given open substack admits a (separated) moduli space. Overall, the paper provides a robust stacky criterion for semistability and illustrates its limits across ranks, offering new perspective on when GIT-type constructions can produce moduli spaces and when they cannot.
Abstract
We show that the semistable locus is the unique maximal open substack of the moduli stack of principal bundles over a curve that admits a schematic moduli space. For rank $2$ vector bundles it coincides with the unique maximal open substack that admits a separated moduli space, but for higher rank there exist other open substacks that admit separated moduli spaces.
