Welschinger invariants are a way of counting real solutions to geometric problems, specifically counting real rational curves (a type of algebraic curve) passing through a given set of points on a surface. Unlike their complex counterparts, these counts can change depending on how you arrange the real structure of the problem, which makes them less useful as true invariants. The challenge this paper addresses is finding combinations of these Welschinger numbers that stay constant no matter how you deform or change the real structure, giving you something genuinely invariant and therefore more meaningful.
The authors previously found such stable combinations for a special class of surfaces called del Pezzo surfaces. In this paper they extend and refine that work to cover all rational surfaces, which is a much broader class. The key technical ingredient is choosing an auxiliary geometric structure called a conic bundle along with an additional piece of data called a Pin-structure. A conic bundle is roughly a way of fibering the surface with families of conics (degree-two curves), and the Pin-structure is a subtle topological choice associated with it. The main result is that even though these auxiliary choices seem like they could introduce ambiguity, the invariants you construct do not actually depend on them, making the construction well-defined.
The paper also establishes recursive formulas that allow you to compute these new invariants step by step, breaking down complicated curve-counting problems into simpler ones. This is important practically, because recursive relations are often the only feasible way to calculate such invariants explicitly. Overall, the work gives the real enumerative geometry community a more powerful and flexible toolkit for extracting meaningful, stable numerical information from real algebraic surfaces, extending the reach of previous results significantly.