Design and Analysis of Experiments for Large Scale Supercomputer

这是一项为大规模超级计算机模拟开发新的实验技术的工作，特别关注康奈尔国家超级计算机设施的模拟应用。本课题包括一个模拟实验设计的研究计划，以及对超级计算模拟实验设计中常见问题和策略的综述。这个项目将产生两个实际后果。第一个将对模拟建模者有直接的好处：来自统计学和运筹学领域的现有模拟方法将在超级计算模拟环境中进行审查和评估。第二个结果将有利于模拟用户和方法论研究者。在模拟方法和实验设计方面的进一步研究的重要领域将被确定和追求。

This is work to develop new experiment techniques for large scale supercomputing simulations, with special focus on simulation applications at the Cornell National Supercomputer Facility. The proposed project consists of a research program on simulation experiment design, and a survey of common problems and strategies in supercomputing simulation experiment design. This project will have two practical consequences. The first will be of immediate benefit to simulation modelers: existing simulation methodologies from the fields of statistics and operations research will be reviewed and assessed in the supercomputing simulation context. The second consequence will benefit both simulation users and methodology researchers. Important areas for further research in simulation methodology and experiment design will be identified and pursued.