Extension theorems for differential forms on low-dimensional git quotients (Q2175521)

In several branches of geometry, spaces are analyzed by looking at the differential forms they carry. More specifically, in algebraic geometry one considers the sheaf of Kähler differentials on a given variety. If that variety is smooth, this is a useful notion. However, also singular varieties may appear and then it is better to look at \textit{reflexive} differential forms. A reflexive differential form on \(X\) is simply a differential form on the smooth locus of \(X\), at least if \(X\) is normal. An important way in which singularities can arise is by using Geometric Invariant Theory (GIT) to take the quotient of a smooth variety by the action of a reductive algebraic group. This is the subject of the paper under review. Given a reflexive form on \(X\), one can pull it back to a resolution of singularities \(\widetilde X \to X\), but in general it will acquire poles along the exceptional divisor. There is quite a tradition of proving statements of the form ``if \(X\) has sufficiently nice singularities, then reflexive differential forms extend to \(\widetilde X\) without any poles''. The main result of the present paper is the following: Theorem 1. Let \(X\) be a good quotient of dimension \(n \le 4\) and \(\eta \colon \widetilde X \to X\) a resolution of singularities. If \(\sigma \in H^0 \big( X, \Omega_X^{[p]} \big)\) is a reflexive \(p\)-form (\(0 \le p \le n\) an integer), then \[ \eta^* \sigma \in H^0 \big( \widetilde X, \Omega_{\widetilde X}^p \big) \] is a regular \(p\)-form on \(\widetilde X\). Theorem 1 follows immediately from two more general results: Theorem 2 says that \(p\)-forms on good quotients always extend if \(p \le 2\). Theorem 3 says that \((n - 1)\)-forms on good quotients of dimension \(n\) extend as soon as the analog of Theorem 1 in dimension less than \(n\) is known. Since good quotients have rational singularities, the above results are a special case of the Extension Theorem of \textit{S. Kebekus} and \textit{C. Schnell} [``Extending holomorphic forms from the regular locus of a complex space to a resolution of singularities'', Preprint, \url{arXiv:1811.03644}], which was posted later. The proofs presented here are completely different, and also more concrete.