Mathematical Sciences Computing Research Environments

科罗拉多大学丹佛分校计算数学中心将购买一台8处理器的Sun微系统SPARC服务器1000E，该服务器将专门用于支持数学科学的研究。该设备将用于几个研究项目，包括：飞行器周围时间相关可压缩湍流直接数值模拟的并行隐式多重网格方法。亚音速和高超音速下的过渡和湍流将直接近似，而不需要湍流模型。地下流动的欧拉-拉格朗日和混合方法。多孔介质中的流动，重点是在石油开采和地下水污染物输送方面的应用，将采用新的数值方法进行模拟，这些方法旨在精确模拟通过不规则地质构造的对流主导流动。稳定有限元方法及其大规模并行实现。将开发和分析具有高精度和鲁棒性的广泛应用的离散化，并设计和实现域分解方程求解器。并行迭代法。对于高变形有限元网格上实体力学模型的离散化所产生的方程，将开发出鲁棒、快速的方程求解器。大型对称特征问题的预条件并行方法。本文将为对称广义特征问题的几个极值特征对的计算提供快速、准确、鲁棒的方法。项目涵盖科学计算的主要领域：模型开发和应用、离散化方法、线性方程和特征值问题。它们都以大规模并行计算为中心主题。

9508328 Russell The Center for Computational Mathematics at the University of Colorado at Denver will purchase an eight-processor Sun Microsystems SPARC server 1000E, which will be dedicated to the support of research in the mathematical sciences. The equipment will be used for several research projects, including in particular: Parallel Implicit Multigrid Methods for Direct Numerical Simulation of Time Dependent Compressible Turbulent Flow around Flight Vehicles. Transitional and turbulent flow at subsonic and hypersonic speeds will be approximated directly, without turbulence models. Eulerian-Lagrangian and Mixed Methods for Subsurface Flows. Flows in porous media, with emphasis on applications to oil recovery and groundwater contaminant transport, will be simulated with new numerical methods designed to be accurate for convection dominated flows through irregular geological formations. Stabilized Finite Element Methods and their Massively Parallel Implementation. Discretizations with strong accuracy and robustness for a wide range of applications will be developed and analyzed, and domain-decomposition equation solvers will be designed and implemented. Parallel Iterative Methods. Robust, fast equation solvers will be developed for equations arising from discretizations of models of solid mechanics on highly distorted finite element meshes. Preconditioned Parallel Methods for Large Symmetric Eigenproblems. Fast, accurate, robust methods will be developed for computation of several extreme eigenpairs of symmetric generalized eigenproblems. The projects span major areas of scientific computation: model development and application, discretization methods, linear equations, and eigenvalue problems. All share large-scale parallel computation as a central theme.