Collaborative Research: CDI- Type II: Towards Analyzing Complex Petascale Datasets: The Milky Way Laboratory

银河系实验室将创建交互式访问有史以来最大的宇宙学结构数值模拟之一。它的数据量接近PB，它说明了新兴的数据挑战-新的突破如何要求我们做科学的方式非增量变化。随着数据集的规模每年翻一番，传统的分析技术已不再适用。科学家们需要一种新的仪器，比如“数据显微镜”，一种数据观测仪，既能观察到堆积如山的数字数据中的细节，也能观察到大规模的特征。增加原始计算机能力是必要的，但还不够-我们还需要使用新的计算思维来开发所需的新方法。NSF支持的研究旨在实现数据密集型计算分析的突破。 特别的科学集中在银河系的起源和演化的引人注目的问题，太阳和我们的太阳系的家园。了解我们在宇宙中的位置将反过来导致对神秘的“暗物质”的性质的洞察，这种物质的引力将银河系结合在一起，但对此知之甚少。该项目将创建一个可公开访问的“银河系实验室”，将银河系形成的最大最先进的计算机模拟的输出与我们银河系中恒星（如太阳）属性的最大观测数据集相结合，并提供回答超出任何数据集范围的问题的方法。公开访问的数据和开发的工具将使科学家能够以前所未有的规模进行自己的创新实验，这表明在当今的大学环境中解决PB级问题是可能的。该项目将为解决其他科学领域类似的数据密集型问题提供蓝图。国际团队由来自学术界（约翰霍普金斯和加州大学圣克鲁斯）和工业界的天文学家、粒子物理学家和计算机科学家组成，以实现其雄心勃勃的目标。该项目将使用美国国家科学基金会资助的5.5 PB数据范围系统，该系统正在约翰霍普金斯大学建设，这是一个独特的数据密集型环境，旨在进行像银河实验室这样的实验。数据范围将位于JHU的一个新的计算设施，由NSF ARRA拨款资助。

The Milky Way Laboratory will create interactive access to one of the largest numerical simulations of cosmological structure ever created. With its data volume close to a petabyte, it illustrates the emerging data challenge -- how new breakthroughs require non-incremental changes in the way we do science. As datasets are doubling in size every year, traditional analysis techniques are no longer adequate. Scientists need a new instrument, like a "microscope of data", a data-scope, that can look at both the details and the large-scale features in the mountains of digital data. Increased raw computer power is necessary, but not enough -- we also need to use new computational thinking to develop the new approaches needed. The NSF-supported research aims to achieve such a breakthrough in data-intensive computational analysis. The particular science focuses on the compelling question of the origin and evolution of the Milky Way Galaxy, the home of the Sun and our solar system. Understanding our place in the Universe will in turn lead to insight into the nature of the enigmatic 'dark matter', the substance whose gravity holds the Milky Way together, but about which little is known. The project will create a publicly accessible 'Milky Way Laboratory' that will combine the output of the largest state-of-the-art computer simulations of the formation of the Milky Way with the largest observational datasets of the properties of stars (like the Sun) in our Galaxy, and provide the means to answer questions that go beyond the scope of either dataset. The publicly accessible data and the tools developed will enable scientists to run their own innovative experiments at unprecedented scales, demonstrating that it is possible to tackle problems on the petabyte scale in a university setting today. The project will provide a blueprint to address similar data-intensive problems in other areas of science.The international team consists of astronomers, particle physicists and computer scientists from academia (Johns Hopkins and UC Santa Cruz) and industry to accomplish its ambitious goals. The project will be using the NSF-funded 5.5-petabyte Data-Scope system, under construction at the Johns Hopkins University, a unique data-intensive environment designed to carry out experiments like the Milky Way Laboratory. The Data-Scope will be located at a new computational facility at JHU, funded by an NSF ARRA grant.