Self-avoiding walks and sampling in statistical physics models

Series: 
Probability Working Seminar
Friday, October 15, 2010 - 15:05
1 hour (actually 50 minutes)
Location: 
Skiles 249
,  
School of Math, Georgia Tech
 We will discuss the role that self-avoiding walks play in sampling 'physical' models on graphs, allowing to translate  the complicated calculation of the marginals to a tree recurrence which, under the appropriate conditions (e.g. some form of 'spatial mixing'), reduces to a polynomial recurrence. This talk is mainly based on Dror Weitz' article "Counting independent sets up to the tree threshold".