Optimal Design with Design Networks

As part of decision-theoretic collaborative design, this work addresses representation and computation issues for a set-based design agent. We refine the definition of design networks to be more expressive for design knowledge and more effective for guiding model construction and verification. We propose a compiled structure of design networks and a suite of algorithms that computes optimal designs efficiently. The result provides a mechanism that combines probabilistic, constraint-based, and decision-theoretic reasoning for single-agent optimal design, and fills in a gap in optimal collaborative design.