Lattice cohomology and rational cuspidal curves

Abstract

We show a counterexample to a conjecture of de Bobadilla, Luengo, Melle-Hernandez and Nemethi on rational cuspidal projective plane curves. The counterexample is a tricuspidal curve of degree 8. On the other hand, we show that if the number of cusps is at most 2, then the original conjecture can be deduced from the recent results of Borodzik and Livingston and (lattice cohomology) computations of Nemethi and Roman. Then we formulate a `simplified' (slightly weaker) version, more in the spirit of the motivation of the original conjecture (comparing index type numerical invariants), and we prove it for all currently known rational cuspidal curves. We make all these identities and inequalities more transparent in the language of lattice cohomology of surgery 3-manifolds $S^3_{-d}(K)$, where $K$ is a connected sum of algebraic knots $\{K_i\}_i$. Finally, we prove that the zeroth lattice cohomology of this surgery manifold depends only on the multiset of multiplicities occurring in the multiplicity sequences describing the algebraic knots $K_i$. This result is closely related to the lattice-cohomological reformulation of the above mentioned theorems and conjectures, and provides new computational and comparison procedures.