Nearly Gorenstein rational surface singularities
Kyosuke Maeda, Tomohiro Okuma, Kei-ichi Watanabe, Ken-ichi Yoshida
TL;DR
The paper develops a practical, resolution-based framework to understand when two-dimensional rational surface singularities are nearly Gorenstein by computing the canonical trace ideal Tr_A(K_A) via anti-nef cycles on a minimal resolution. It proves that Tr_A(K_A) is represented by the minimal anti-nef cycle F with F+K_X anti-nef and that F=Z_f characterizes nearly Gorenstein rational singularities, yielding concrete combinatorial criteria. It then classifies nearly Gorenstein rational singularities in two key cases: almost reduced fundamental cycles and quotient singularities, providing explicit graph-theoretic characterizations and families. The results provide both computational tools and structural insights, linking trace ideals to resolution graphs and expanding the landscape of nearly Gorenstein rational and quotient surface singularities.
Abstract
In this paper, we show that for any rational surface singularity $A$, the canonical trace ideal $\mathrm{Tr}_A(K_A)$ is integrally closed ideal which is represented by the minimal anti-nef cycle $F$ on the minimal resolution of singularities so that $K_X+F$ is anti-nef. Then $F \ge \mathbb Z$ if $A$ is not Gorenstein, where $\mathbb Z$ is the fundamental cycle. As a result, we give a criterion for rational surface singularity $A$ to be nearly Gorenstein. Moreover, we classify all nearly Gorenstein rational singularities in terms of resolution of singularities in the following cases: (a) the fundamental cycle $\mathbb Z$ is almost reduced; (b) quotient singularity.
