The Euclidean Distance (ED) degree of a geometric object is a number that tells you how many critical points appear when you try to find the point on that object closest to a randomly chosen target point. This concept matters in applications like computer vision, robotics, and data science, where you frequently need to solve "nearest point" problems on curved surfaces or other constrained geometric shapes. The higher the ED degree, the more complex the optimization landscape, and knowing this number helps researchers understand how hard a given nearest-point problem is to solve.
The paper introduces a software package called EuclideanDistanceDegree, written for Macaulay2, which is a widely used computer algebra system for algebraic geometry and commutative algebra. The package gives users two broad approaches to computing ED degrees: symbolic methods, which work with algebraic equations directly using tools like polynomial minors and conormal varieties, and numerical methods, which use techniques from numerical algebraic geometry to approximate solutions when exact symbolic computation becomes too expensive. Having both options is useful because symbolic methods give exact answers but can struggle with complex examples, while numerical methods scale better but produce approximate results.
The authors demonstrate the package through a variety of worked examples, and they also provide supporting code in a public GitHub repository. The practical value is that researchers who encounter nearest-point problems in their own work can now compute ED degrees without having to implement the underlying mathematics themselves. This lowers the barrier to using ED degrees as a diagnostic tool in optimization and applied geometry, and it brings together symbolic and numerical strategies in one accessible package.