A0653
Title: Statistical inference in Ising and Potts models
Authors: Somabha Mukherjee - National University of Singapore (Singapore) [presenting]
Abstract: Some of the existing literature is summarized in the field of statistical inference in the classical Ising and Potts model, followed by presenting some of the recent results in the same area for tensor and more general regression versions of these classical models. The main focus is on consistent estimation of the inverse temperature and external field parameters of these models and proving asymptotics of these estimators. As seen, these asymptotics are accompanied by surprising phase transitions in terms of the true parameter position and the rate of convergence of the sufficient statistics. A comparative discussion is also given on the efficiencies of two of the most important consistent estimators in these models: the maximum likelihood and the maximum pseudolikelihood estimators.
