Validating Large Scale Simulations of Socio-Political Phenomena

This project involves research that supports the development of a prototype system to identify active factors in large complex HSCB simulation models. The approach is based on a Bayesian inference technique for finding active factors in fractionated screening experiments. The project will a) continue research on algorithm enhancements based on the results of Phase I, 2) continue research into scalable methodology for applying the algorithm to large models, c) develop a prototype, and d) test the prototype against exit criteria using a known mathematical model, a large simulation leveraged during development and a complex HSCB model identified by DARPA at the end of the project. The key benefit is that the resulting product will fill the gap in HSCB modeling that is seen when models are too large and too complex to accommodate Verification and Validation. This model will provide new insight into which factors are truly “driving factors” when applying a simulation to determine the effects of certain input changes (e.g. based on policy or operational decisions) on specific outputs at specific times. If commercialized, the product will include XML interfaces that can be loosely coupled to any complex model, whether it is a physical science or HSCB model.