Title: Distributed problem formulations for the design of complex
systems

Multidisciplinary optimization (MDO) is a collection of systematic 
approaches to the design optimization of complex, coupled engineering 
systems. Multidisciplinary systems, such as aerospace vehicles, involve
optimization with respect to many aspects, or disciplines, of the design
problem.  The expense and difficulty of treating MDO problems as 
conventional NLP has motivated researchers to propose alternative optimal 
design problem formulations and attendant optimization algorithms. The 
proposed approaches usually involve some degree of problem distribution.
We discuss the effect of the degree of distribution on the tractability of 
the resulting optimization problem. We propose a dynamic approach to problem 
formulation for MDO that takes advantage of the problem structure, maximizes 
disciplinary autonomy, and allows for interchangeable problem statements
to be used depending on the available resources and the characteristics of 
the application problem.