| Popis: |
Massively parallel (MP) computers provide increased capabilities for a wide range of engineering and physics simulations. High performance computing applications include quantum mechanics for materials modeling, charged particle beams for accelerator design, and computational fluid dynamics for global climate change. Progress in these areas required the development of scalable applications software as well as new mathematical methods and algorithms for parallel computers. This paper reviews recent work at Sandia that enables large-scale engineering simulations on MP computers. Important contributions include the development of iterative and direct solvers for sparse systems of linear equations, an automated meshing algorithm for finite elements, load balancing methods for mapping a computational grid to an array of processors, as well as scalable structural mechanics, computational fluid dynamics, and shock physics models. |
