MRI Acquisition: Multi-Environment Research Computer for Exploration and Discovery (MERCED) Cluster

科学计算在知识发现和科学进步中发挥着核心作用。加州大学默塞德（UCM）超过13%的终身教职员工专注于计算科学领域。UCM的这一专业知识贯穿于大学的三个学院，包括自然科学、工程学和社会科学的教师领导的实验室。该提案寻求支持，以收购一个名为探索和发现多环境研究计算机（默塞德）的共享高性能计算（HPC）集群。集群包括异构节点，包括具有大内存的节点、具有大存储空间的节点和GPU节点。三个具体的研究领域受益于这个集群，包括复杂系统的建模和仿真，数据支持的科学和数值优化。在复杂系统的模拟中，该集群将推动高性能分子动力学、摩擦建模原子相互作用计算、化学反应和化学生物学的第一原理模拟以及流体动力学的研究。在数据支持的研究中，集群将影响计算系统生物学的研究，例如处理高通量基因组学数据，认知科学，例如人类语言的进化，历史和世界文化，例如整合代表多个空间，时间和主题尺度的许多数据集，并允许分析和空间可视化人类事务中的历史数据，以及使用GPU计算来加速数据库系统的性能。在数值优化方面，该集群将加强对电网优化、3D数据处理的GPU计算和信号重建算法的研究。该集群加强了UCM以及附近大学的教育和外联活动。

Scientific computing plays a central role in knowledge discovery and scientific advancement. Over 13% of tenured/tenure-track faculty at the University of California Merced (UCM) specialize in areas focused on computational science. This expertise at UCM cuts through the university's three schools and includes faculty-led labs in the natural sciences, engineering, and social sciences.This proposal seeks support for the acquisition of a shared high-performance computing (HPC) cluster named the Multi-Environment Research Computer for Exploration and Discovery (MERCED). The cluster includes heterogeneous nodes, including nodes with large memory, nodes with large storage space, and GPU nodes. The three specific research areas benefit from this cluster, including modeling and simulation of complex systems, data enabled science, and numerical optimization. In the simulation of complex systems the cluster would advance research in high-performance molecular dynamics, computations of atomistic interactions for friction modeling, first principles simulations of chemical reactions and chemical biology, and fluid dynamics. In data enabled study, the cluster would impact the research in computational system biology, such as handling high-throughput genomics data, cognitive sciences, such as the evolution of language in human populations, history and world culture, such as integration of many datasets representing multiple spatial, temporal and thematic scales and permitting analysis and spatial visualization of this historical data in human affairs, and using GPU computing for speeding up the performance of database systems. In numerical optimization, the cluster would enhance the research in optimizing electric grid, GPU computing for 3D data processing, and signal reconstruction algorithms. The cluster enhances the educational and outreach activities at UCM as well as in the nearby universities.