Calculation of Large-Scale Engineering Flows by Parallel Spectral Element Methods

并行谱元法计算大型工程流量

• 批准号: 8806925
• 负责人: Anthony Patera
• 金额: $ 25.89万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1988
• 资助国家: 美国
• 起止时间: 1988-08-01 至 1991-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8806925&HistoricalAwards=false
• 关键词: Calculation; Large; Scale; Engineering; Flows

The economical solution of complex three dimensional fluid flow problems by direct numerical simulation techniques requires simultaneous, compatible advances on the algorithmic and architectural fronts. The architectural foundations for the next ten-fold improvement in simulation performance are now established in the form of resource efficient massively parallel processors. Work to date on parallel algorithms has focussed on the fundamental aspects of parallel soultion of partial differential equations (e.g., load balancing, communication latency) in the context of simple discretizations of idealized in non-general geometires. The extensions of these parallel concepts and constructs to the numerical solution of complex engineering flow problems is the subject of the current proposal. The ultimate goal is the parallel solution of a broad class of flow problems at supercomputer speed, but at a fraction of current supercomputer cost.

复杂三维流体流动的经济解法 直接数值模拟技术的问题需要 同时，在算法和 建筑前沿 未来的建筑基础 模拟性能提高了十倍， 以资源高效的大规模并行的形式建立 处理器. 迄今为止，并行算法的工作主要集中在 部分并行解决的基本方面 微分方程（例如，负载平衡，通信 延迟）的理想化的简单离散化的上下文中， 非一般几何学 这些并行的扩展 概念和结构的数值解复杂 工程流程问题是当前提案的主题。 最终目标是并行解决一个广泛的类， 在超级计算机的速度流动问题，但在一小部分， 超级计算机的成本。