All 81 crepant resolutions of a finite quotient singularity are hyperpolygon spaces
Gwyn Bellamy, Alastair Craw, Steven Rayan, Travis Schedler, Hartmut Weiss
TL;DR
We establish a Poisson isomorphism between the exceptional four-dimensional quotient $C^4/G$ with $G=Q_8\times_{Z_2}D_8$ and the affine hyperpolygon space $X_5(0)$, providing a uniform quiver-variety framework that realizes all $81$ projective crepant resolutions of $C^4/G$ as hyperpolygon spaces $X_5(\\theta)$. By exploiting the star-shaped quiver description, we completely describe the birational geometry of $X_n(\\theta)$, including the movable cone and Namikawa Weyl group action, and show that the number of resolutions equals the number of chambers in a fixed chamber in stability space, yielding $81$ for $n=5$ and $1684$ for $n=6$. We prove that for $n>5$ the affine hyperpolygon spaces $X_n(0)$ are not finite quotient singularities, situating the $n=4$–$5$ cases within the broader Hamiltonian-reduction/quiver-variety paradigm and highlighting why this phenomenon does not extend to higher dimensions. The work also analyzes the hyperkähler geometry, proving the $n=5$ case should be QALE in the generic metric class, while the higher-$n$ geometries fail to be QALE, and discusses connections to the McKay correspondence in this higher-rank setting.
Abstract
We demonstrate that the linear quotient singularity for the exceptional subgroup G in Sp(4,C) of order 32 is isomorphic to an affine quiver variety for a 5-pointed star-shaped quiver. This allows us to construct uniformly all 81 projective crepant resolutions of the quotient singularity C4/G as hyperpolygon spaces by variation of GIT quotient, and we describe both the movable cone and the Namikawa Weyl group action via an explicit hyperplane arrangement. More generally, for the n-pointed star shaped quiver, we describe completely the birational geometry for the corresponding hyperpolygon spaces in dimension 2n - 6; for example, we show that there are 1684 projective crepant resolutions when n = 6. We also prove that the resulting affine cones are not quotient singularities for n >= 6.
