Events

Video screening: Arun G. Chandrasekhar - Tractable and Consistent Random Graph models

In the framework of the lecture series of RECENS we have a special video screening event on 10th May of 2016. We'll watch the 2014 lecture of Arun G. Chandrasekhar, Tractable and Consistent Random Graph Models (presented in Becker Friedman Institute For Research Economics at the University of Chicago). It will start at 3 PM, after the screening we will discuss about the lecture. Please join us!

Venue: RECENS Conference Room, downstairs in the building of Institute for Sociology

 

Arun G. Chandrasekhar defines a general class of network formation models, Statistical Exponential Random Graph Models (SERGMs), that nest standard exponential random graph models (ERGMs) as a special case. Their definition provides the first general results on when these models’ parameters become accurate as the number of nodes grows. While standard techniques of estimating ERGMs have exponentially slow mixing times for many specifications, reformulating network formation as a distribution over the space of sufficient statistics, instead of the space of networks, makes estimation practical and easy.

A related, but distinct, class of models is defined as subgraph generation models (SUGMs), which are useful for modeling sparse networks. Choice-based (strategic) network formation models can be written as SERGMs and SUGMs, as demonstrated with network data from rural Indian villages.