Fall 2013

On Unifying Asymptotic Complexity with Real-world Performance in Matrix-based Network Computations

Thursday, November 21st, 2013 9:35 am10:00 am

Add to Calendar

Many network analysis questions can be posed as solving a system of linear equations with a stochastic random-walk matrix, e.g. PageRank. Monte Carlo methods often have the best asymptotic complexity for these problems. Yet, those same methods are notoriously slow to find high precision answers. In this talk, we pose some questions on unifying asymptotic complexity with real-world performance.