Montanari Andrea (Stanford University)

Convergence times in large coordination games and Ising models

(based on joint work with A. Saberi and Y. Kanoria)

Social network data are increasingly available and comprehensive. This raises the possibility of to study a number of social and economic phenomena and their dependence on the underlying network structure. Here we consider a family of classical game theoretic models: coordination games under noisy best response dynamics, and their statistical mechanics analogues, Ising model.

Since the pioneering work of Ellison (1993), specific network structures have been shown to have dramatic influence on the convergence behavior of such networks. The objective of this talk is to translate this intuition into effective algorithms.