Moduli spaces for $Θ$-strata and non-reductive quotients
Ludvig Modin
TL;DR
The work develops a unified framework linking Białynicki-Birula and Θ-strata via stacks of filtrations and gradings, enabling a general non-reductive GIT theory over any affine Noetherian base. Central to the approach are the stacks of filtrations $\operatorname{Filt}(\mathcal{X})$ and gradings $\operatorname{Grad}(\mathcal{X})$ and the contracting $\mathbb{G}_m$-actions on their fibers, which yield a contracting moduli-theoretic structure formalized in Theorem A. This leads to coarse and tame moduli spaces for non-split filtrations, rigidifications, and rigid centers, and culminates in a broad $\hat{U}$-theorem that provides geometric quotients for graded unipotent actions on projective schemes, with projective quotients for appropriate unstable loci. The results apply to families and arbitrary characteristics, extending NR-GIT techniques and enabling moduli constructions for unstable objects in broader moduli problems. Altogether, the paper advances non-reductive quotient theory by situating Θ-strata and filtrations within a stack-theoretic framework that yields explicit moduli spaces and quotients, with wide potential applications in moduli problems lacking a classic GIT presentation.
Abstract
We give a new proof of the $\hat{U}$-theorem of Bérczi, Doran, Hawes and Kirwan on the existence of geometric quotients for actions of graded unipotent groups in terms of stacks of filtrations and gradings introduced by Halpern-Leistner. Our proof works over any affine Noetherian base, in particular it simultaneously generalizes the previous results to arbitrary characteristic, actions in families and to general $Θ$-strata.
