The paper tackles a classical question in algebraic geometry: when can you tell that two geometric objects with a singular point (a "kink" or "cusp" where the curve crosses itself or has a sharp corner) are essentially the same? The key idea is that a curve near a singular point can be described by a parameterization, much like describing a path by giving coordinates as functions of a single variable. Two parameterizations might look different on the surface but still trace out the same geometric shape, in which case they are called "equivalent." The central question is: if two parameterizations agree up to some finite level of approximation (technically, modulo a large power of the maximal ideal), does that guarantee they are truly equivalent?
The first main result shows that agreement up to a sufficiently high order of approximation is indeed enough to guarantee full equivalence of parameterizations. This improves on earlier work by Greuel and Pfister by providing sharper, more precise bounds on exactly how high that order of approximation needs to be. The second main result addresses a related but distinct question: if two curve singularities look identical up to some high finite order, are their complete (fully resolved, infinite-order) descriptions also identical? The paper proves that yes, the full completions must be isomorphic, and moreover the isomorphism at the full level is consistent with the original finite-order agreement. Here the paper improves a classical bound due to Hironaka.
The technical tool linking these results is the delta-invariant, a number that measures how complicated a singularity is by counting, roughly speaking, how far the singular curve is from being smooth. The paper uses careful estimates involving this invariant to pin down the precise thresholds needed for the equivalence results to hold. The upshot is a cleaner and more efficient criterion for deciding when two singular curves are geometrically the same, which is useful both theoretically and for practical classification of curve singularities.