US-France Cooperative Research (INRIA): Algorithms for Large-Scale Constrained Optimization

这一为期三年的奖项支持西北大学的Jorge Nocedal和INRIA(法国国家计算机科学和应用数学研究所)的Jean Charles Gilbert正在进行的美法计算机科学合作。研究人员此前致力于无约束最优化算法的研究，将其应用于气象学中的同化问题。他们建议将研究扩展到解决涉及大量变量的大规模约束优化问题。目标是开发比目前可用的更快、更灵活的算法和软件。研究人员选择了一个具有任意多个变量的重要非线性问题--航天飞机重新进入大气层时的轨迹优化。他们将应用序列二次规划方法来解决这个问题。该提案利用了互补的专业知识。这位美国研究人员为这次合作带来了大规模优化软件开发方面的丰富经验。法国研究人员在解决最优控制问题和最近对航天飞机控制的研究方面的专门知识补充了这一点。该项目可能会导致实际应用，但将促进算法和软件的使用，这些算法和软件可能对具有大量变量的工程和科学问题有用。

This three-year award supports ongoing U.S-France collaboration in computer science between Jorge Nocedal of Northwestern University and Jean Charles Gilbert of INRIA (The French National Institute for Research in Computer Science and Applied Mathematics). The investigators worked previously on algorithms for unconstrained optimization as applied to assimilation problems in meteorology. They propose to extend their research to solving large-scale constrained optimization problems involving a large number of variables. The objective is to develop faster and more flexible algorithms and software than is currently available. The investigators selected an important nonlinear problem with an arbitrarily large number of variables - the optimization of the trajectory of a space shuttle as it re-enters the atmosphere. They will apply sequential quadratic programming methods to this problem. The proposal takes advantage of complementary expertise. The U.S. investigator brings to this collaboration extensive experience in the development of software for large scale optimization. This is complemented by the French investigator's expertise in solutions of optimal control problems and recent studies of control of the space shuttle. The project may lead to practical applications, but will advance the use of algorithms and software which may useful in engineering and scientific problems with large numbers of variables.