Transformative Petascale Particle-in-cell simulations

变革性千万亿级细胞内粒子模拟

• 批准号: 1440071
• 负责人: Warren Mori
• 金额: $ 1.23万
• 依托单位: University of California-Los Angeles
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-09-01 至 2016-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1440071&HistoricalAwards=false
• 关键词: Transformative; Petascale; Particle; cell; simulations

This project accesses computational resources on the Blue Waters computer to answer several compelling science questions. The UCLA Simulation of Plasmas Group has spent several decades developing deep physics understanding of phenomenon in key areas of plasma physics. It has also helped to formulate key compelling questions in each of these areas, that if answered could be transformative. It has built up a comprehensive suite of fully kinetic plasma simulation tools that include a variety of physics, that are highly optimized on a single compute core, and that scale to the full size of leadership-class computers such as Blue Waters. It is anticipated that the obtained resources on Blue Waters will enable transformative simulations in diverse areas of plasma physics including plasmas based acceleration, inertial fusion energy, astrophysics, and space weather. Presently at UCLA and within the OSIRIS Consortium there are six distinct parallel PIC codes, with varying degrees of physics approximations, being used to explore a wide range of phenomena. All of these codes were self-built and are self-maintained; several use the same particle manager for the parallelization. These codes are highly optimized on a single processor and have very high parallel efficiency, e.g., OSIRIS has already been shown to scale to more than 1,500,000 cores on Sequoia and the full Blue Waters machine. These simulations will not only enable answers to compelling science questions, but they will also provide insight into fundamental nonlinear and high energy density science as well as on how to use particle-in-cell codes effectively on leadership class facilities.

该项目访问Blue沃茨计算机上的计算资源，以回答几个引人注目的科学问题。加州大学洛杉矶分校等离子体模拟小组花了几十年的时间来发展对等离子体物理关键领域现象的深入物理理解。 它还帮助在每个领域提出了关键的紧迫问题，如果得到回答，可能会产生变革。它已经建立了一套全面的全动力学等离子体模拟工具，包括各种物理，在单个计算核心上高度优化，并可扩展到Blue沃茨等领导级计算机的全尺寸。 预计在Blue沃茨上获得的资源将能够在等离子体物理学的不同领域进行变革性模拟，包括基于等离子体的加速、惯性聚变能、天体物理学和空间天气。目前，在加州大学洛杉矶分校和OSIRIS联盟内，有六个不同的并行PIC代码，具有不同程度的物理近似，被用于探索各种现象。所有这些代码都是自构建和自维护的;有几个使用相同的粒子管理器进行并行化。这些代码在单个处理器上被高度优化，并且具有非常高的并行效率，例如，OSIRIS已经被证明可以在Sequoia和完整的Blue沃茨机器上扩展到超过150万个核心。这些模拟不仅能够回答令人信服的科学问题，而且还将提供对基本非线性和高能量密度科学的见解，以及如何在领导级设施上有效使用粒子细胞代码。