"Identifiable structural equation models"

Seth Sullivant
North Carolina State University

UK WILDCATS Seminar
Monday, March 1, 2010
4:00 pm, 845 POT

Abstract:

A structural equation model is a statistical model that uses a graph to encode linear relationships between a collection of random variables. These linear relationships determine a positive definite covariance matrix whose entries are polynomials in certain parameters associated to the edges of the graph. The edge parameters are identifiable if they can be recovered from the covariance matrix. I will discuss recent results, where we employ techniques from computational algebra and algebraic combinatorics to make progress on the identifiability problem. This is joint work with Mathias Drton, Rina Foygel, Luis Garcia, and Sarah Spiegelvogel.