The paper addresses a question in the geometry of surface singularities, which are points where a geometric surface fails to be smooth. One particularly well-behaved type of singularity can be "smoothed," meaning you can deform the singular surface into a nearby smooth one. The paper focuses on a special kind of smoothing called a rational homology disk smoothing, where the smooth surface that replaces the singularity has the simplest possible topology in a specific algebraic sense. Understanding which singularities admit such smoothings has been an open problem for decades.
To study singularities, mathematicians use a tool called the resolution graph, which encodes how the singular point can be replaced by a collection of curves in a controlled way. The graph has nodes representing these curves, and edges representing how they intersect. The main result of the paper is that if this resolution graph has two or more nodes (meaning the singularity is sufficiently complex), then a rational homology disk smoothing is impossible. This rules out a huge class of singularities from admitting such smoothings.
Combined with earlier work by other researchers who had classified simpler cases and provided explicit analytic descriptions of the singularities that do qualify, this new result completes the full picture. Together, these works confirm a conjecture made by Jonathan Wahl, settling a long-standing open problem by giving a definitive list of exactly which normal surface singularities can be smoothed in this special rational homology disk sense. The paper also proves a related statement known as Wahl's Rational Conjecture as a consequence of the same core argument.