On \(\mathbb C\)-fibrations over projective curves (Q1001951)

In this interesting paper, the authors introduce a variant of the Makar-Limanov invariant for rings, the Generalized Makar-Limanov invariant. The \(ML\)-invariant of a ring \(R\) is the intersection of the kernels of all locally nilpotent derivations (or equivalently, the invariants of all \(G_a\)-actions). The \(GML\)-invariant allows a larger class of derivations. Let \(f\in R\), and suppose \(D\) is a locally nilpotent derivation of \(F(R)\), the fraction field of \(R\), such that \(D(f)=0\) and \(D\) is locally nilpotent on \(R[f]\). The set of such derivations is called \(GLND(R)\), the set of generalized locally nilpotent derivations. The invariant \(GML(R)\), the generalized Makar-Limanov invariant, is defined as the intersection of all kernels \(R^D\) where \(D\in GLND(R)\). It turns out that this invariant is really different, though related to the \(ML\) invariant. They compute an example of a smooth surface \(S\) for which \(ML(S)=S\) but \(GML(S)\) is different. Reviewer's remark: The authors don't specifically say so, but they probably mean to assume that \(R\) is a domain (as they speak about \(F(R)\) etc.). It might be possible to define \(GML\) for non-domains, though.