Scientific Computing Research Environments for the Mathematical Sciences (SCREMS)

该项目将为计算密集型应用统计学方法的研究提供计算支持：二维和三维数据中的小波方法，基因表达数据的微阵列数据分析，寻找黑盒函数可解释近似的蒙特卡罗方法，功能磁共振图像的空间平滑，以及流式细胞仪数据的自动聚类。统计学的研究前沿现在与t检验和卡方检验相去甚远。现代科学设备在复杂的环境中收集数以百万计的数据测量。例如，生物学家在单芯片上测量数千个基因的表达水平，大脑科学家在受试者执行任务时观察大脑活动的地图。处理这些问题和相关问题的数据需要新的统计方法来从随机性中挑选出模式。这些方法反过来又需要在速度和内存方面的辅助计算能力。该项目将支付计算资源，以支持五个这样的研究领域。

This project will provide computing support forresearch on computationally intensive appliedstatistical methods: wavelet methods for findingfiliments in two and three dimensional data,microarray data analysis of gene expression data,Monte Carlo method of finding interpretableapproximations to black box functions,spatial smoothing of functional MRI images,and automatic clustering of flow cytometry data.The research frontiers of statistics are now afar cry from the t-tests and chi-squared tests ofintroductory courses. Modern scientific equipmentgathers millions of data measurements in complexsettings. For example, biologists measure theexpression levels of thousands of genes on a singlechip, and brain scientists watch maps of brainactivity as a subject performs a task. Handlingdata from these and related problems requires newstatistical methods to sort out patterns fromrandomness. These methods in turn requiresubstantial computating power, in both speed andmemory. This project will pay for computatingresources to support five such research areas.