The paper studies curves with special types of singular points called cusps, which are points where a curve crosses itself in a pinched or pointed way rather than forming a clean crossing. The specific curves under investigation live on smooth surfaces and are "rational," meaning they have a relatively simple overall shape despite their singularities. The central goal is to figure out how "twisted" or "self-intersecting" such a curve can be within its surrounding surface, with self-intersection being a precise numerical measure of how a curve sits inside the surface around it.
The main result proves a sharp upper bound on this self-intersection number, confirming a conjecture made by Weimin Chen, who encountered the question while studying contact geometry and symplectic fillings, which are structures from modern geometry and topology. The authors use a sophisticated set of tools called the log minimal model program, which is a powerful framework from algebraic geometry for systematically simplifying geometric spaces. By running this program, they can constrain what kinds of singular curves are geometrically possible and compare their purely algebraic bounds with previously known bounds coming from symplectic geometry, finding that the two approaches give complementary information.
The paper also resolves a specific open question about whether a particular type of plane curve can exist: a degree-102 curve of genus 10 with a very specific kind of cusp described by the pair of numbers (36, 289). The answer is no, such a curve cannot exist. This question arose from a completely different direction involving a combinatorial and geometric object called the McDuff-Schlenk embedding staircase, which tracks how balls can be packed into ellipses in symplectic geometry. The fact that this algebraic non-existence result connects back to packing problems illustrates how different areas of mathematics are surprisingly intertwined.