Dr. Elizabeth Gross
University of Hawaii at Manoa
One of the main goals of phylogenomics is to understand the evolutionary history of a set of species. These histories are represented by directed graphs where the leaves represent living species and the interior nodes represent extinct species. While it is common to assume the evolutionary history is a tree, when events such as hybridization are present, networks are more realistic. However, allowing for networks rather than trees complicates the process of inference, and ways to overcome this complication are needed. One recent approach to phylogenetic network inference is rooted in computational algebraic geometry. In this talk, we discuss the role computationally algebraic geometry has played in the statistical problems related to network inference and show how algebraic geometry combined with statistical learning can aid in model selection. This is joint work with Travis Barton, Colby Long, and Joseph Rusinko.