Probabilistic Programming for Analyzing Complex Systems (PPACS)

Development planning is a critical task in the US Air Force (USAF) that is required to efficiently acquire new and improved vehicles, analyze strategic and tactical gaps in the USAFs missions, guide investment strategies of new technologies, and plan for future workforce skills. One of the key development planning tasks is analyzing the impacts of alternative designs on the potential system of systems (SoS) concept of operations (CONOPs), its ability to meet the desired measures of performance and effectiveness (MOPs/MOEs), and the sensitivity of the system to various internal and external factors. Yet performing this analysis is difficult and costly due to the complexity and variety of systems, as well as the uncertainty within and impacting the systems. Bayesian networks (BNs) have been used to satisfy these development planning. However, BNs are insufficient as a technological base because of their rigid structure and limited support for advanced analysis. To address these development planning gaps in multi-domain USAF systems and operations, Charles River Analytics proposes a method for Probabilistic Programming for Analyzing Complex Systems (PPACS). This comprehensive, generalized tool uses advanced probabilistic programming languages and probabilistic relational models to support a wide variety of development planning needs. This will enable systems