Acquisition of a High Performance Parallel Workstation Cluster for Research in Mechanics of Materials

9871240鲍尔此行动是对重大研究仪器计划MRI'98 （NSF 98-16）的回应。其目的是为布朗大学的计算机械研究设施增加一个高性能工作站集群。该集群连接形成一个专用的高速并行计算环境，并用于加强布朗大学材料研究科学与工程中心（MRSEC）的活动。MRSEC支持研究的主要目标是开发和应用新的计算方法，以了解原子尺度（纳米尺度）和微观结构尺度（微米尺度）的现象如何决定材料的机械、电子和光学性质。近年来，在实现这一目标方面取得了一些重大进展，包括：拟连续体方法的发展，该方法将原子模拟技术与自适应有限元分析相结合；塑性离散位错模型的发展应力固体中微观组织演化的锐界面和相场模型的建立计算应变对半导体电子特性影响的有限元方法的发展基于物理的断裂模型的发展，利用材料的微观结构特性来预测结构的断裂抗力。该设施将用于开发和应用可扩展的并行算法，以模拟各种长度尺度的材料行为，并结合耦合物理过程。该设施的一个直接影响是允许对材料行为进行真实的三维模拟。进一步的进展需要系统地应用这些技术来充分探索材料变形和破坏的基本机制。该设施为这项工作提供了必要的计算能力，并有望在理解材料行为方面取得重大突破。

9871240BowerThis action is in response to the Major Research Instrumentation Initiative MRI'98 (NSF 98-16). The purpose is to add a cluster of high performance workstations to the Computational Mechanical Research Facility at Brown University. The cluster is connected to form a dedicated, high speed parallel computing environment, and is used to strengthen the activities of the Materials Research Science and Engineering Center (MRSEC) at Brown University. The primary objective of MRSEC supported research is to develop and apply novel computational methods to understand how phenomena at the atomic scale (with nanometer dimensions) and microstructural scale (with micron dimensions) determine the mechanical, electronic and optical properties of materials. Several significant advances have been made towards this goal in recent years, including: development of the quasi-eontinuum method, which couples atomistic simulation techniques with adaptive finite element analysis; development of discrete dislocation models of plasticity; development of sharp interface and phase field models of microstructures evolution in stressed solids; development of finite element methods for computing strain effects on the electronic properties of semiconductors; and the development of physically-based models of fracture, which use microsructural properties of materials to predict the fracture resistance of structures. The facility is to be used to develop and apply scalable, parallel algorithms for simulating material behavior at a range of length scales and incorporating coupled physical processes. An immediate impact of the facility is to allow realistic three-dimensional simulations of material behavior to be run routinely. Further progress requires a systematic application of these techniques to explore fully the fundamental mechanisms of deformation and failure in materials. This facility provides the necessary computing power for this effort, and holds the promise of major breakthroughs in understanding the behavior of materials.