Scalable, efficient epidemiological simulation

We describe the design and implementation of a system for simulating the spread of disease among individuals in a large urban population over the course of several weeks. In contrast to traditional approaches, we do not assume uniform mixing among large sub-populations or split the population into spatial or demographic subpopulations determined a priori. Instead, we rely on empirical estimates of the social network, or contact patterns, that are produced by TRANSIMS, a large-scale simulation of transportation systems.