The Hilbert scheme of points on a threefold: broken Gorenstein structures and linkage
Joachim Jelisiejew, Ritvik Ramkumar, Alessio Sammartano
TL;DR
This work studies the Hilbert scheme of $d$ points on smooth threefolds, introducing broken Gorenstein structures as a sharp criterion for smoothness and conjecturing their exhaustiveness on the smoothable component. It provides a complete monomial-case classification, linking smoothness to the absence of singularizing triples and licci-ness, and develops the bicanonical module as a central invariant guiding deformation and tangent-space analysis. The authors establish a Pfaffian structural framework for broken Gorenstein algebras without flips, prove Hu’s singularity conjectures in several key cases (monomial, tripod, and Borel-fixed), and connect smoothness to linkage and nested Hilbert schemes. The paper also uncovers a torus-equivariant rank-$d$ bundle on Hilb^d(A^2) arising from the bicanonical module and highlights how linkage preserves smoothable tangents, offering a nuanced picture of singularities in codimension three Hilbert schemes with substantial implications for combinatorics and algebraic geometry.
Abstract
We investigate the Hilbert scheme of points on a smooth threefold. We introduce a notion of broken Gorenstein structure for finite schemes, and show that its existence guarantees smoothness on the Hilbert scheme. Moreover, we conjecture that it is exhaustive: every smooth point admits a broken Gorenstein structure. We give an explicit characterization of the smooth points on the Hilbert scheme of A^3 corresponding to monomial ideals. We investigate the nature of the singular points, and prove several conjectures by Hu. Along the way, we obtain a number of additional results, related to linkage classes, nested Hilbert schemes, and a bundle on the Hilbert scheme of a surface.
