The paper tackles several long-standing problems about how well certain mathematical measurements can distinguish between different types of singularities, which are points where a smooth surface or shape develops a sharp corner, cusp, or self-intersection. One central question, posed by the mathematician Vladimir Arnold in 1975, asks whether a quantity called the "corank" of a function is preserved when two functions are considered topologically equivalent, meaning one can be continuously deformed into the other without tearing. The corank is a basic measure derived from the second-order behavior of a function near a singular point, and the paper proves that at minimum, the corank modulo 2 (that is, whether it is even or odd) is always preserved under such deformations. This gives a new and more general proof of a result that was only recently established for the specific case of three-dimensional spaces.
The paper also addresses a conjecture from 2022 by a group of mathematicians about what happens to certain geometric objects called Milnor fibers when two functions are topologically equivalent. The Milnor fiber is a smooth surface naturally associated with a singularity, and the conjecture predicts that if two functions are equivalent in a strong topological sense, the Milnor fibers associated with their "leading term" approximations should be essentially the same shape. The author proves that a weaker version of this, asking only that these Milnor fibers share the same Euler characteristic (a basic count of holes and connected pieces), turns out to be logically equivalent to another famous open problem called the Zariski multiplicity conjecture. This conjecture, dating back to Oscar Zariski, asks whether the "multiplicity" of a singularity, a measure of how complicated it is, is always preserved under topological equivalence.
Beyond these main results, the paper establishes that for a broad family of geometric objects called generalized Brieskorn-Pham hypersurfaces, having a multiplicity greater than two is itself a topological invariant. The paper also shows that in continuously varying families of functions where a numerical quantity called the Milnor number stays constant, the corank cannot jump around either. Taken together, these results push forward the understanding of which geometric and algebraic properties of singularities are truly topological in nature, meaning they depend only on the shape of the object and not on the specific formula used to describe it.