Scientific Computing Research Environments for the Mathematical Sciences (SCREMS)

North Carolina State University, Raleigh NC

Investigators

Abstract

NSF Proposal DMS-0209695 PIs: Ipsen, Campbell, Kelley, and Meyer ABSTRACT The proposed research concerns the numerical solution of sparse optimal control problems. The research issues originate from the Boeing Company's software project SOCS (Sparse Optimal Control Software). The constraints in the control problems are represented by systems of differential algebraic equations (DAEs), a class of differential equations whose solution poses serious challenges. The PIs propose to advance the state-of-the-art in high-index DAE technology to solve realistic, large-scale applications; to increase software scalability through the use of interior point methods for non-linear solvers and Krylov-space methods for linear solvers; and to develop prototyping software to test accuracy and performance of new algorithms. Research will be conducted on improving software for difficult, complex problems in aerospace and mechanical engineering applications, such as design of flight control systems, robot arms, and information retrieval systems. The use of computation to explore new science and engineering concepts is central to the project. This multi-disciplinary project is driven by collaboration with the Boeing company on its industrial software project SOCS which solves many large-scale optimal control problems that occur in chemical reactions, and a large variety of aerospace problems.