The paper develops a rigorous mathematical framework for working with multidimensional delta functions, which are generalizations of the classical Dirac delta function. The Dirac delta is a standard tool in physics and mathematics that can be thought of as a spike concentrated at a single point, useful for selecting specific values from integrals. The authors extend this idea by defining delta functions on higher-dimensional surfaces and curves by a process called "pulling back," which essentially means transferring the delta function from a simple setting onto more complicated geometric objects. This gives a clean and consistent way to define these objects rather than relying on informal or case-by-case treatments.
A central result of the paper is a local version of the coarea formula. The coarea formula is a classical result in geometric measure theory that relates integrals over a space to integrals over the level sets of a function, somewhat like a generalization of Fubini's theorem for slicing a space in a geometric way. The authors prove a localized version of this, meaning it applies in smaller regions rather than globally, using the language of delta functions as generalized functions. This makes the result more flexible and applicable in a wider range of situations.
The paper also applies this machinery to probability theory, specifically to computing marginal distributions on spheres. A marginal distribution is what you get when you take a joint probability distribution over several variables and integrate out some of them to focus on the rest. Working on a sphere introduces geometric complications that the delta function framework handles naturally. The results provide concrete computational tools for probability problems involving spherical geometry, with potential uses in statistics, physics, and data analysis where data naturally lives on spherical or curved spaces.