ITR Development of Parallel Coupled Cluster Methods

艾奥瓦州立大学的马克·戈登、里奇·肯德尔和迈克·施密特以及密歇根州立大学的皮奥特·皮库赫通过信息技术研究计划，支持化学系并行化从头算电子结构方法，特别是耦合团簇理论及其改进。耦合团簇理论是一种计算昂贵的方法，但它被认为是求解多电子薛定谔方程的最准确的方法。因此，开发在大规模并行体系结构上实现的高度可伸缩的代码将产生广泛的影响。从这一理论水平对能量学和化学性质的准确预测被用于化学的所有领域。这里开发的理论方法将在包括有机、生物和材料化学以及许多物理和工程领域在内的广泛的现象的模拟中找到普遍的适用性。此外，还将开发新的范例，作为并行化物理科学中其他复杂代码的模板。由此产生的计算工具将免费分发给世界各地的学术界、政府实验室和工业界的研究人员。

Mark Gordon, Ricky Kendall, and Mike Schmidt of Iowa State University and Piotr Piecuch of Michigan State University are supported by the Chemistry Division through the Information Technology Research Program to parallelize ab initio electronic structure methods, specifically coupled cluster theory and improvements upon it. Coupled cluster theory is a computationally expensive methodology but it is regarded as the most accurate means for solving the many-electron Schroedinger equation. Therefore, developing a highly scalable code for implementation on massively parallel architectures will have broad impact. The accurate predictions of energetics and chemical properties from this level of theory are used in all areas of chemistry.The theoretical methods developed here will find general applicability in modeling a wide range of phenomena, including organic, biological, and materials chemistry, as well as many areas of physics and engineering. In addition, new paradigms will be developed as templates for parallelizing other complex codes in the physical sciences. The resulting computational tools will be freely distributed to researchers worldwide in academia, government labs, and industry.