High performance computing for statistical research and training

达尔豪斯大学数学与统计学系的统计研究涉及对大型数据集和计算机密集建模方法的分析，以追求教师和研究生的跨学科研究目标。有效地参与这种联合研究需要大量的计算资源。我们的部门计算设施由NSERC资助，是部门内的共享资源，可供所有数学和统计教师，研究生，博士后研究员和访问者在部门。在这里，我们建议升级当前的计算集群，将现有的15个节点替换为最先进的节点。此升级为我们提供了一种将计算能力提高许多倍的方法，同时不会增加集群的功耗，从而避免昂贵的基础设施改进。升级现有的并行计算集群将有助于开展海洋科学、遗传学和医学统计应用方面的研究项目。

Research in the Statistics Division of the Department and Mathematics & Statistics at Dalhousie University involves analysis of large data sets and computer intensive modelling approaches to pursue interdisciplinary research goals for both faculty and graduate students. To effectively participate in such joint research requires significant computational resources. Our departmental computing facilities funded by NSERC is a shared resource within the department that is available to all Mathematics and Statistics faculty, graduate students, post doctoral fellows and visitors in the department. Here, we propose to upgrade the current computing cluster by replacing the existing 15 nodes with state-of-art ones. This upgrade provides us a means to increase our computing power manyfold, while not increasing the power consumption of the cluster and so avoiding expensive infrastructure improvements. Upgrading the existing parallel computing cluster will assist in carrying out research programs in statistical applications in the marine science, genetics, and medicine.