The paper develops a new framework connecting three mathematical worlds: algebraic geometry (studying shapes defined by polynomial equations over the complex numbers), tropical geometry (a simplified "shadow" version where complex geometry is replaced by combinatorial data involving minimum and maximum operations), and topology (the study of loops and how they can be continuously deformed). The central objects of interest are "algebraic loops," which are paths in a geometric space that arise from a specific algebraic construction involving formal power series. The key insight is that such loops, which live in a complicated complex geometric space, can often be tracked and classified using much simpler combinatorial data called tropical points, which are essentially integer-valued coordinates in a piecewise-linear shadow of the original space.
For a well-behaved class of spaces called "schon varieties," the paper proves that an algebraic loop is almost completely determined by its tropical shadow, with the remaining ambiguity captured by a connected component of an associated "initial variety." When these initial varieties happen to be connected, the tropical data alone is enough to completely classify all algebraic loops. The paper then applies this general theory to a concrete and important family of spaces: complements of hyperplane arrangements, which are spaces obtained by removing a collection of flat walls from a larger space. In this setting, the framework produces explicit geometric representatives for each algebraic loop, built from basic building blocks called "relative full twists."
The most concrete payoff appears in classical braid theory. When the hyperplane arrangement is of "type A" (the standard family), the paper's construction recovers the well-known pure algebraic braids, giving them a new tropical interpretation. For other families of arrangements called root arrangements, the construction connects to earlier work by Goresky, Kottwitz, and MacPherson on a combinatorial invariant called the "split root valuation." The paper closes with an application to braid varieties, proving that a topological invariant called cohomology for these spaces depends only on the tropical root valuation data rather than on the full complexity of the braid, which provides a powerful simplification for future computations.