Collaborative Research: SI2-SSI: A Linear Algebra Software Infrastructure for Sustained Innovation in Computational Chemistry and other Sciences

合作研究：SI2-SSI：用于计算化学和其他科学持续创新的线性代数软件基础设施

• 批准号: 1148125
• 负责人: Robert van de Geijn
• 金额: $ 144.53万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-06-01 至 2016-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1148125&HistoricalAwards=false
• 关键词: Collaborative; Research; SI2; SSI; Linear

Linear algebra is a branch of mathematics that provides the foundation for a significant fraction of computations in science and engineering. Historically, the importance of linear algebra is such that highly specialized codes written by computer scientists have been used by the community of scientific programmers as a vital part of their application programs. With the rapid changes in computer architecture during the last several years, it would seem that corresponding modifications in linear algebra routines would be warranted. However, such progress is not in evidence; the development of such routines has been just incremental, involving successive rewrites of routines that had their genesis in the last quarter of the last century. Correspondingly, there is somethingof a disconnect between the current `state-of-the-art' linear algebra libraries, modern computer architectures, and applications that utilize the libraries. The new project will create a new, vertically integrated framework and implementation that revisits every layer of software, from low-level kernels to higher level functionality. The vertical integration is completed with a new generation of software for computationalchemistry applications, guaranteeing that the developed software, to be freely available to the public, supports sustained innovation in that domain and other sciences. The development builds on the FLAME project, which has been funded by NSF and industry for more than a decade.

线性代数是数学的一个分支，它为科学和工程中很大一部分计算提供了基础。从历史上看，线性代数的重要性如此之大，以至于由计算机科学家编写的高度专业化的代码一直被科学程序员社区用作其应用程序的重要部分。随着计算机体系结构在过去几年中的快速变化，似乎有必要对线性代数例程进行相应的修改。然而，这样的进步并不明显；这种例程的发展只是渐进的，包括对上个世纪最后四分之一年起源的例程的连续重写。相应地，当前最先进的线性代数库、现代计算机体系结构和利用这些库的应用程序之间存在某种程度的脱节。新项目将创建一个新的垂直集成框架和实现，重新访问从低级内核到高级功能的每一层软件。垂直一体化是通过用于计算和化学应用的新一代软件来完成的，从而保证所开发的软件可免费向公众提供，支持该领域和其他科学领域的持续创新。这一开发建立在FLAME项目的基础上，该项目由美国国家科学基金会和工业界资助了十多年。