The Grassmannian of lines is a geometric object that collects together all the lines passing through the origin in some higher-dimensional space. It shows up naturally in many areas of mathematics and physics. Researchers in algebraic statistics are interested in using such geometric spaces as statistical models, where points on the space represent probability distributions. A key question is: given some observed data, how hard is it to find the best-fitting probability distribution on the model? One way to measure this difficulty is through a quantity called the maximum likelihood degree, which counts the number of algebraically possible solutions to that fitting problem. A higher degree generally means the optimization landscape is more complex.
The authors derive an exact, closed-form formula for the maximum likelihood degree of the Grassmannian of lines. This is a concrete algebraic result, meaning they found a precise mathematical expression rather than just a computational estimate. The "positive Grassmannian" in the title refers to a particularly well-behaved region of this space where all coordinates are positive, which connects naturally to probability theory since probabilities must be non-negative.
Beyond the formula, the authors conjecture that within the region of valid probability distributions (called the probability simplex), there is actually only one local maximum of the likelihood function. In practical terms, this would mean that despite the potentially large number of algebraic solutions, finding the best fit is actually straightforward from a statistical standpoint: any reasonable search procedure would converge to the true answer without getting stuck in false peaks. The authors back up this conjecture with substantial computational experiments, testing many cases without finding a counterexample, though a full proof remains open.