Why Are Intervals and Imprecision Important In Engineering Design?

This paper was presented at the Proceedings of the Reliable Engineering Computing Workshop (REC'06), Savannah, GA, February 22-24.It is valuable in engineering design to distinguish between two different types of uncertainty: inherent variability and imprecision. While variability is naturally random behavior in a physical process or property, imprecision is uncertainty that is due to a lack of knowledge or information. There are many sources of imprecision in design. Sequential decision making introduces imprecision because the results of future decisions are unknown. Statistical data from finite samples of environmental factors are inherently imprecise. Bounded rationality leads to imprecise subjective probabilities. Expert opinions and judgments often are imprecise due to a lack of information or conflict. Behavioral simulations and analysis models are imprecise abstractions of reality. Knowledge of a decision maker's preferences may be imprecise due to bounded rationality or other constraints. Consequently, the engineering design community needs efficient computational methods for interval data and imprecise probabilities in order to support decision making in the design process. This paper introduces these sources and needs, with the aim of forming a foundation for future collaboration with the reliable engineering computing community.NSF (Grant #DMI-0522116