A note on the rational cuspidal curves (Q2926665)

Let \(C\) be a rational cuspidal curve of degree \(d\), with finitely many singular points \(p_1, \dots, p_n\). In [Proc. Lond. Math. Soc. (3) 92, No. 1, 99--138 (2006; Zbl 1115.14021)], \textit{J. Fernández De Bobadilla} et al. proposed the following conjecture: for \(i=1, \dots, n\) let \(\Delta_i(t)\) be the characteristic polynomial associated to the germ \((C,p_i)\), and let \(\Delta(t)\) be the product of the \(\Delta_i's\); let \(Q(t)\) be defined by the equation \(\Delta(t) = 1 + (d-1)(d-2)(t-1)/2 + (t-1)^2 Q(t)\); for \(j = 0, \dots, d-3\), let \(c_j\) be the coefficient of \(t^{(d-3-j)d}\) in \(Q(t)\); then for each \(j\) the inequality \(c_j \leq (j +1)(j+2)/2\) holds. In this paper the Author prove, by an elementary combinatorial argument, that this conjecture follows from a result of \textit{M. Borodzik} and \textit{C. Livingston} [Forum Math. Sigma 2, Article ID e28, 23 p. (2014; Zbl 1325.14047)] in the case \(n=2\). Recently \textit{J. Bodnár} and \textit{A. Némethi} [``Lattice cohomology and rational cuspidal curves'', \url{arXiv:1405.0437}] showed that the conjecture is false for \(n \geq 3\).