The paper tackles a question in algebraic geometry about a specific type of curve called a rational unicuspidal curve. These are curves that have exactly one singular point, called a cusp, where the curve crosses or pinches itself in a complicated way. The shape of that singularity is described by a pair of numbers called a Puiseux pair. One key property of such a curve is its self-intersection number, which measures, loosely speaking, how the curve sits within the surrounding geometric space and how it would intersect a nearby copy of itself. A previous conjecture proposed an upper bound on how large this self-intersection number could be for curves with a given Puiseux pair.
The main result of the paper is a proof that this conjectured upper bound is actually achievable. The authors construct explicit curves that hit the maximum possible self-intersection number for every choice of Puiseux pair. They do this by studying two large families of related curves, called rational bicuspidal curves (which have two cusps instead of one), living in two different geometric spaces. By carefully analyzing these families, they show that one cusp can be simplified away, leaving behind optimal unicuspidal curves that realize the upper bound. This strengthens an earlier result that only worked in a more flexible geometric setting called symplectic geometry, extending it to the stricter algebraic geometry setting.
Along the way, the authors develop new computational tools. They derive recursive formulas that make the calculations more manageable, and in doing so they find a cleaner, more intuitive formula for the upper bound itself. They also discover a bonus result: a formula for the last term in a sequence of numbers, called the multiplicity sequence, that encodes detailed information about the structure of the singularity at the cusp. These side results are likely to be useful for future work on curve singularities and related problems.