Role of variability in complex systems

Michael Insana, University of Illinois at Urbana-Champaign

Usage Details

Understanding, predicting and controlling complex systems is a grand challenge for 21st century engineering. Extending our reductionist expertise as scientists to allow the responses of very large systems like populations, economies, and plate tectonics to become predictable requires new techniques for system model using sparsely sampled data. We developed a theory suggesting system variance of a type that contributes degrees of freedom to system properties is essential for generating emergent behavior. To demonstrate these principles, we selected a problem in modeling population dynamics resulting in viral epidemics. We study how degrees of freedom in the population are able to produce the characteristic power-law distributions of infected individuals that demonstrate the system is operating in a self-organized critical state. We developed hybrid models that include mathematically tractable behavior of individuals that interact as agents in an agent-based model. Initially we use published data from Faroe Island sources to develop and validate the model, but we are limited by the high computational load required to monitor large populations. This project aims to apply high-performance computational resources to this system and others.