Postdoc: Multidisciplinary Research in Numerical Relativity within the Context of High Performance, Massively Parallel Computing
Postdoc: Multidisciplinary Research in Numerical Relativity within the Context of High Performance, Massively Parallel Computing
批准号:
9503978
负责人:
Edward Seidel
金额:
$4.62万
依托单位国家:
美国
项目类别:
Standard Grant
财政年份:
1995
资助国家:
美国
项目状态:
已结题
起止时间:
1995-11-15 至 1998-10-31
中文摘要
[9503978]塞德尔:不同学科之间的交叉受精是计算科学许多成功领域的标志,但在数值相对论中却迟迟没有出现。该提案旨在使解决爱因斯坦方程的一种有前途的新方法成熟,该方法基于一种新的公式,该公式将它们明确地置于一阶,通量保守的双曲形式中。这种方法为从其他领域借鉴强大的现代算法打开了大门。在此基础上,PI计划将适当的数学形式、可扩展的计算机、复杂的并行和数值算法(包括双曲系统的先进方法和优化的线性求解器)和实验性的科学可视化技术(如虚拟现实)整合到一个代码中。已经使用这种全局方法开发了一个初步版本,在完全不同的并行架构上运行,性能出色。PI建议构建一个通用控制器,将计算、可视化和数据分析分布到元中心。物理应用将包括:一个普通的三维引力波实验室,三维黑洞模拟,玻色子恒星作为碰撞物体的测试平台,以及完全广义相对论流体动力学。可以开发一个单一的,非常通用的和强大的研究工具,并且由此产生的经验可以应用到计算科学的其他领域,使本建议成为真正的多学科努力。
英文摘要
9503978 Seidel The cross fertilization across different disciplines that has been the hallmark of many successful fields of computational science has been slow to occur in Numerical relativity. The proposal is designed to bring maturity to a promising new approach for solving Einstein's equations, based on a new formulation that casts them in an explicitly first order, flux-conservative, hyperbolic form. This approach opens the door to borrowing powerful modern algorithms from other fields. Building on this the PI's plan to incorporate appropriate mathematical formalisms, scaleable computers, sophisticated parallel and numerical algorithms (including advanced methods for hyperbolic systems and optimized linear solvers) and experimental scientific visualization techniques (such as virtual reality) into a single code. A preliminary version has already been developed with this global approach, running with outstanding performance on fundamentally different parallel architectures. The PI's propose the construction of a general controller that distributes computation, visualization, and data analysis across the metacenter. The physical applications will include: a general 3D gravitational wave laboratory, 3D black hole simulations, bosonic stars as a testbed for colliding objects, and fully general relativistic hydrodynamics. A single, very general and powerful research tool can be developed, and the resulting experience can be applied into other fields of computational science, making this proposal a truly multidisciplinary effort.
期刊论文(0)
专著(0)
科研奖励(0)
会议论文
U.S. -Spain Cooperative Research: Advanced Methods for 3D Numerical Relativity
-
批准号:9414185
-
项目类别:Standard Grant
-
资助金额:$1.97万
-
财政年份:1995
-
负责人:Edward Seidel
-
依托单位:
Computational Relativity
-
批准号:9407882
-
项目类别:Continuing Grant
-
资助金额:$7.4万
-
财政年份:1994
-
负责人:Edward Seidel
-
依托单位:
海外基金