Weakly-exceptional quotient singularities (Q424141)

Let \(P \in X\) be a Kawamata log terminal singularity. \(P \in X\) is called exceptional if and only if for any effective \(\mathbb{Q}\)-divisor \(D\) such that the pair \((X,D)\) has log canonical singularities, there exists at most one exceptional divisor \(E\) with discrepancy \(a(E,X,D)=-1\). \(P \in X\) is called weakly exceptional if and only if there exists a unique birational map \(f : Y \rightarrow X\) such that the exceptional set of \(f\) is an irreducible divisor \(E\), \((Y,E)\) has purely log terminal singularities, \(-E\) is \(f\)-ample and \(P \in f(E)\). Minimal model theory shows that every exceptional singularity is weakly exceptional. This paper classifies finite subgroups \(G < \mathrm{GL}(n,\mathbb{C})\), \(n \leq 4\), such that the quotient \(X=\mathbb{C}^n/G\) has weakly exceptional singularities. The case \(n=2\) was treated by V. Shokurov who showed that the two dimensional exceptional singularities are exactly the DuVal singularities of type \(D_n, E_6, E_7, E_8\). The classification method used by the author is the following. Let \(\bar{G}\) be the image of \(G\) in \(\mathrm{PGL}(n-1,\mathbb{C})\). Then according to a result of I. Cheltsov and C. Shramov, \(\mathbb{C}^n/G\) has weakly exceptional singularities if and only if \(\mathrm{lct}(\mathbb{P}^{n-1},\bar{G})\geq 1\), where \(\mathrm{lct}(\mathbb{P}^{n-1},\bar{G})\) is the global \(\bar{G}\)-invariant log canonical threshold of \(\mathbb{P}^{n-1}\) [\textit{I. Cheltsov} and \textit{C. Shramov}, Geom. Topol. 15, No. 4, 1843--1882 (2011; Zbl 1232.14001)]. Then by lifting \(\bar{G}\) to \(\mathrm{SL}(n,\mathbb{C})\), one can assume that \(G\) is a subgroup of \(\mathrm{SL}(n,\mathbb{C})\). Then according to a result of I. Cheltsov and C. Shramov, \(\mathrm{lct}(\mathbb{P}^{n-1},\bar{G})\geq 1\) if and only if \(G\) does not have semi invariants of degree at most \(n-1\) for its action on \(\mathbb{C}^n\), \(n \leq 4\). Hence if \(\mathbb{C}^n/G\) does not have weakly exceptional singularities, \(G\) fixes a degree at most \(n-1\), \(n \leq 4\), hypersurface of \(\mathbb{P}^{n-1}\) and therefore it is a subgroup of the automorphisms of either a conic, a quadric surface or a cubic surface. Then the classification follows from the explicit classification of finite subgroups of \(\mathrm{SL}(n,\mathbb{C})\) and the automorphism groups of cubic surfaces, quadrics and conics.