The classical Skitovich-Darmois theorem says that if you take two linear combinations of independent random variables and those combinations turn out to be statistically independent of each other, then the original variables must have been normally (Gaussian) distributed. This is a beautiful characterization of the normal distribution through a purely algebraic condition. Over the decades, mathematicians extended this result in several directions, most notably to situations where the random variables are vectors rather than single numbers, and where you allow infinitely many of them rather than just a finite collection.
This paper pushes that idea into a much more abstract setting. Instead of random variables taking values in ordinary numbers or vectors, the authors consider random variables that live in objects called locally compact Abelian groups. These are mathematical spaces that have both a group structure (you can "add" elements together) and a topological structure (there is a sensible notion of closeness), with the two structures compatible. Examples include the real line, circles, and more exotic objects called solenoids, which are compact spaces built by wrapping lines around themselves in increasingly fine ways. The "linear forms" in this setting use symmetry transformations of the group rather than ordinary number multiplication.
The main results establish precise conditions under which the independence of two such generalized linear forms forces the underlying random variables to follow the group-theoretic analogue of a Gaussian distribution. The authors identify specific structural properties of the group, such as whether it contains compact subgroups or certain circular substructures, that determine which version of the theorem holds. They also work out the case of solenoids in detail. The paper thus unifies and significantly generalizes earlier results by Ghurye, Olkin, Zinger, and Ibragimov, showing that the deep connection between independence of linear forms and Gaussian behavior is not special to ordinary numbers but reflects something much more fundamental about symmetry and probability.