MRI: Acquisition of an Advanced Computing Instrument to Integrate Data-Driven Research and Data intensive computing at Johns Hopkins University

该项目为购买高性能计算和存储系统提供资金，以创建将设在约翰·霍普金斯大学(JHU)的数据密集型科学计算(DISCO)环境。迪斯科将支持一系列科学和工程领域的研究，包括风能、海洋循环、设计材料、添加剂制造、大数据、基因组学和从膜到器官尺度的生物功能。它提供了一个框架，用于从模拟中创建新数据、分析数据，并使数据可供远程用户分析。DISCO还将使各机构能够开发和共享新的代码和数据分析方法。这个项目涉及NSF的两个大想法：不断增长的融合研究和利用数据革命，以及材料基因组倡议。迪斯科将立即支持40多个教职员工研究项目，涉及200多名科学家。迪斯科将与摩根州立大学合作管理，摩根州立大学是一所历史上的黑人大学，该大学将获得至少5%的资源来支持其教育和计算研究项目。通过与XSEDE的合作，将向全国其他用户提供20%的资源进行管理。各种科学计算主题的培训机会将直接纳入JHU和摩根州立大学的常规课程。计划开设计算化学、基因组学、分子动力学、机器学习和蛋白质化学等课程。课程材料将通过网络提供，JHU现有的大型在线开放课程(MOOC)将得到扩展。这些材料将确保正确使用资源的培训，在不同学科之间分享最佳实践，并促进跨学科合作。这个主要研究仪器(MRI)项目的目标是创建一个数据密集型科学计算(DISCO)环境，将高性能计算与用于生成、分析和传播不断增长的数据集的工具相结合。该群集将包含超过5 PB的存储和针对不同研究项目和复杂、优化的工作流进行优化的异类计算节点：标准双处理器节点、大内存节点和图形处理器(GPU)增强型节点。本网站的一个软件实例将加强分析，并提供一个通过门户网站传播数据的平台。这一强大的能力将成为开发和共享代码和基础设施工具的强大资源，供约翰·霍普金斯大学、摩根州立大学等跨学科数据科学家使用。DISCO将使风电场、海洋流动和生理流体动力学的计算流体动力学研究成为可能。材料的多尺度建模研究将使用机器学习来定制材料属性，并在不同尺度上集成物理模型，以提高添加剂制造的性能，并将缺陷和微观结构与晶体、聚合物和其他软材料的宏观性能联系起来。DISCO还将使新的基因组测序、物种内变异研究和微生物组研究成为可能。在更大的尺度上，对生物系统动力学的研究将检查蛋白质、膜、细胞和器官的功能，包括纳米颗粒对功能的影响。这种新的集成环境将重塑计算研究实践，并将有助于激进的变革性研究。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The project funds the purchase of a high-performance computing and storage system to create a Data Intensive Scientific Computing (DISCO) environment that will be located at Johns Hopkins University (JHU). DISCO will support research in a range of science and engineering areas including wind energy, ocean circulation, materials by design, additive manufacturing, big data, genomics, and biological function on membrane to organ scales. It provides a framework for creating new data from simulations, analyzing data, and making data available for analysis by remote users. DISCO will also enable the development and sharing of new code and data analysis methods across institutions. This project addresses two of NSF's big ideas: Growing Convergence Research and Harnessing the Data Revolution, as well as the Materials Genome initiative. DISCO will immediately support over 40 faculty research projects involving over 200 scientists. DISCO will be managed in collaboration with Morgan State University, a historically black university, which will be allocated at least 5% of resources to support its educational and computational research programs. Other users nationally will be offered 20% of the resources to be managed through a partnership with XSEDE. Opportunities for training on a diverse set of scientific computing topics will be directly integrated into regular courses across JHU and Morgan State. Courses on computational chemistry, genomics, molecular dynamics, machine learning and protein chemistry are planned. Course materials will be made available through the web and the existing set of Massive Open Online Courses (MOOCs) at JHU will be expanded. These materials will ensure training in the proper use of resources, share best practices between different disciplines and promote interdisciplinary collaboration.The objective of this Major Research Instrumentation (MRI) project is to create a Data Intensive Scientific Computing (DISCO) environment that integrates high performance computing with tools for generating, analyzing and disseminating data sets of ever increasing size. The cluster will contain over 5 petabytes of storage and heterogeneous compute nodes optimized for different research projects and complex, optimized work flows: standard dual processor nodes, large memory nodes, and Graphics Processing Unit (GPU) enhanced nodes. An instance of SciServer software will enhance analysis and provide a platform for disseminating data through a web portal. This robust capability will become a powerful resource for developing and sharing code and infrastructure tools used by interdisciplinary data scientists at Johns Hopkins University, Morgan State University and beyond. DISCO will enable computational fluid dynamics studies of wind farms, ocean flow and physiological fluid dynamics. Research in multiscale modeling of materials will use machine learning to tailor material properties and integrate physical models at different scales to improve performance of additive manufacturing and connect defects and microstructures to macroscopic performance of crystals, polymers and other soft materials. DISCO will also enable sequencing of new genomes, study of variance within species and research on the microbiome. On larger scales, research on the dynamics of biological systems will examine function of proteins, membranes, cells and organs, including the impact of nanoparticles on function. This new integrated environment will reshape computational research practices and will be conducive to radical transformative research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.