Title: Algebraic statistics for random graph models
Speaker: Sonja Petrovic
Speaker Info: University of Illinois at Chicago
Algebraic statistics is a maturing field focusing on applications of algebraic geometry to the study of statistical models. Many interesting families of models correspond to classical families of algebraic varieties. In particular, toric ideals play a special role in the field. One family of "toric models" is given by the p1 random graph model, which models interactions in a social network. The Markov basis for this model can be understood via the special fiber ring of bipartite graphs.Date: Tuesday, February 09, 2010
I will introduce the basic definitions necessary for understanding the algebraic geometry- statistics dictionary for this family of examples.
(Joint work with Stephen Fienberg and Alessandro Rinaldo, of Carnegie Mellon University.)