Collaborative Research: Towards Petascale Cosmological Simulations

该奖项是根据2009年美国复苏和再投资法案（公法111-5）资助的。该奖项支持一项研究计划，以开发宇宙结构形成的千万亿次数值模拟。这项工作将利用自适应细化树（ART）宇宙学模拟代码。具体目标是：(1)通过改进千万亿级平台上多核/多处理器节点的负载平衡、通信性能和混合并行化模式，实现ART宇宙码的千万亿级性能；(2)开发用于千万亿次ART模拟的鲁棒容错机制；(3)提供用户级界面和文档，以便在开源许可下公开发布ART；(4)产生最先进的宇宙模拟，这将为天体物理学界带来丰富的科学成果和有价值的模拟结果库。该项目结合了在负载平衡、可伸缩性和容错研究方面具有丰富经验的计算机科学家的专业知识，以及一组在代码开发和运行最先进的宇宙模拟方面具有丰富经验的数值宇宙学家。这项工作产生的千万亿次宇宙学模拟将导致我们对星系形成的理解取得重大突破，并将为即将到来的大型观测调查提供关键的理论支持，这些调查旨在探测我们宇宙的物质和能量含量，并限制神秘暗物质和暗能量的性质。这个项目将通过在一个项目上建立一个紧密合作的团队，弥合计算机科学和天体物理学之间的跨学科界限。通过不同层次的研究生和博士后学者的参与，该项目将为年轻的计算科学家和数值天体物理学家提供前沿模拟技术的专业培训。该项目的最终结果——一个公开可用的、开源的千万亿次宇宙学模拟代码——将对一般天体物理学社区具有巨大价值，并将被广泛使用，正如其他目前存在的、公开可用的宇宙学代码的广泛使用所显示的那样。该代码还可以用作当前和未决暗能量实验的校准工具。在更广泛的范围内，该项目将对本科教育和公众产生直接影响。这里提出的这种宇宙学模拟的可视化激发了普通公众的好奇心和一种只有通过探索宇宙才能获得的发现感。为此，与芝加哥的阿德勒天文馆合作，将为全穹顶天文馆展览和公共讲座开发模拟的三维可视化，并在网上提供。

OCI-0904482/OCI-0904484/OCI-0904670Kravtsov/Rudd/LanThis award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).This award supports a research program to develop petascale numerical simulations of structure formation in the universe. This work will utilize the Adaptive Refinement Tree (ART) cosmological simulation code. Specific goals are: (1) to achieve petascale level performance for the ART cosmological code by improving load balancing, communication performance, and the hybrid parallelization mode on multi-core/multi-processor nodes on petascale platforms; (2) to develop a robust fault-tolerance mechanism for petascale ART simulations; (3) to provide a user-level interface and documentation for public release of ART under an open source license; and (4) to produce state-of-the-art cosmological simulations that will result in a rich scientific output and a valuable repository of simulation results for the astrophysical community. The project combines the expertise of a computer scientist with extensive experience in research on load balancing, scalability, and fault tolerance, with a group of numerical cosmologists with significant experience in code development and running state-of-the-art cosmological simulations. The petascale cosmological simulations resulting from this work will lead to major breakthroughs in our understanding of galaxy formation and will provide critical theoretical support for forthcoming large observational surveys designed to probe the matter and energy content of our universe and constrain properties of the mysterious dark matter and dark energy.This project will bridge the cross-disciplinary boundaries between Computer Science and Astrophysics by developing a tightly-knit team collaborating on a single project. By involving graduate students of different levels and a postdoctoral scholar, this project will serve as an excellent vehicle for professional training of young computational scientists and numerical astrophysicists in cutting edge simulation techniques. The ultimate result of this project - a publicly available, open source petascale cosmological simulation code - will be of great value for the general astrophysics community, and will be widely used, as shown by the widespread usage of other, currently existing, publicly available cosmological codes. The code can also be used as a calibration tool for current and pending dark energy experiments. On a broader scale, this project will have a direct impact on undergraduate education and the general public. Visualizations of cosmological simulations of the kind proposed here trigger a curiosity and a sense of discovery in the general public afforded only by the exploration of the universe. To this end, in collaboration with the Adler Planetarium in Chicago, three-dimensional visualizations of the simulations will be developed for full-dome planetaria shows and public lectures and be made available on the web.