Scientific Computing Research Environments for the Mathematical Sciences (SCREMS)

NSF提案DMS-0209695 PI：Ipsen，坎贝尔，凯利和迈耶摘要所提出的研究涉及稀疏最优控制问题的数值解。本文的研究课题来源于波音公司的软件项目SOCS（Sparse Optimal Control Software）。 控制问题中的约束是由微分代数方程组（DAE）表示的，微分方程组的解具有严重的挑战性。PI建议推进高索引DAE技术的最新发展，以解决现实的大规模应用;通过使用非线性求解器的内点方法和线性求解器的Krylov空间方法来提高软件的可扩展性;开发原型软件，测试新算法的精度和性能。研究如何改进软件，解决航空航天和机械工程应用中的困难和复杂问题，例如飞行控制系统、机械臂和信息检索系统设计。利用计算来探索新的科学和工程概念是该项目的核心。这个多学科项目是由与波音公司在其工业软件项目SOCS上的合作推动的，SOCS解决了许多发生在化学反应中的大规模最优控制问题，以及各种各样的航空航天问题。

NSF Proposal DMS-0209695PIs: Ipsen, Campbell, Kelley, and MeyerABSTRACTThe proposed research concerns the numerical solutionof sparse optimal control problems. The research issuesoriginate from the Boeing Company's software projectSOCS (Sparse Optimal Control Software). Theconstraints in the control problems are represented bysystems of differential algebraic equations (DAEs), aclass of differential equations whose solution posesserious challenges. The PIs propose to advance thestate-of-the-art in high-index DAE technology to solverealistic, large-scale applications; to increasesoftware scalability through the use of interior pointmethods for non-linear solvers and Krylov-space methodsfor linear solvers; and to develop prototyping softwareto test accuracy and performance of new algorithms.Research will be conducted on improving software fordifficult, complex problems in aerospace and mechanicalengineering applications, such as design of flightcontrol systems, robot arms, and information retrievalsystems. The use of computation to explore new scienceand engineering concepts is central to the project.This multi-disciplinary project is driven bycollaboration with the Boeing company on its industrialsoftware project SOCS which solves many large-scaleoptimal control problems that occur in chemicalreactions, and a large variety of aerospace problems.