Scalable Agent-based Modeling Framework

This project works to parallelize spatially-explicit agent-based models and scale the parallel implementation to thousands of processors in order to model more complex phenomena than previously possible. Coordinating thousands of processors to simulate a single model is a challenging task, but our parallel approach uses several parallel techniques to efficiently utilize the computational resources and reduce communication overhead. Our modeling approach has been used to simulate hundreds of millions of agents interacting on hundreds of millions of environment cells and is currently being used to simulate mosquitoes in the Chicago area to study West Nile disease transmission.

Project Team: Eric Shook, Wenwu Tang, Shaowen Wang