Large-Scale and Parallel Nonlinear Optimization

点击翻译按钮获取中文摘要

9414355 Nash This research is focused on the development of algorithms for nonlinear programming, particularly, problems that involve many variables. The emphasis is on algorithms suitable for parallel computer implementation. The aim is at methods suitable for coarse-grained parallel computers where each processor is itself moderately powerful. The domain of target processors of interest include computers such as the Intel Paragon and Cray supercomputers. The algorithms developed are aimed to be general- purpose, easy to use, and ultimately implemented as a computer software. Nonlinear programming problems exist in several sectors of the economy, including both the industrial and military systems. However, nonlinear programming models can be difficult to solve. The general-purpose nature of the algorithms to be developed will make available the power of computers to users of nonlinear models, and hence extend the range of nonlinear programming as an everyday modeling tools.