CAREER: Rheology of Materials of Earth's Mantle: High-end Computational/Visualization Research and Education

该计划旨在系统地应用计算机和材料科学的思想和技术来解决地球材料基本问题的理论研究中的挑战性问题。其中一个问题是要了解地球地幔的组成材料的流变学性质在流变学相关的压力和温度条件下，仍然是实验无法达到的。流变学是一个关键因素，它对地震观测和其他来源所揭示的复杂地幔动力学有很大的影响。大规模原子模拟将在大规模并行机器上使用第一性原理量子力学（QM），经典分子动力学（MD）和混合QM/MD方法的组合进行。 这些模拟将预测主要硅酸盐和氧化物地幔矿物中几个关键的流变学相关性质和过程，包括点缺陷、位错和相关的长程机械现象，以及各向异性应力下多晶矿物的变形。由此产生的大量多变量数据集将在沉浸式和交互式环境中可视化，以深入了解矿物的复杂力学行为。 上述跨学科研究活动将纳入教学/学习/培训活动。这些项目包括培养新一代具有计算和地球材料科学专业知识的学生（通过计算机和地球科学双学位研究生课程），开发基于Access Grid的远程教学/学习环境（包括路易斯安那州的少数民族机构），为代表性不足的群体组织计算科学研讨会和狂欢节会议，这些会议以前在路易斯安那州州立大学（LSU）举办。 这些活动的成功执行将有学生和教师的广泛实质性参与，包括那些来自代表性不足的群体在物理学和信息技术的接口的研究和教育。 该项目将利用LSU最近收购的世界级1024处理器Linux集群、Immersadesk虚拟现实和Access Grid设施。 --

This proposal aims to systematically apply ideas and techniques of computer and materials sciences to challenging problems in the theoretical investigation of fundamental issues of Earth materials. One such issue is to understand the rheological properties of the component materials of Earth's mantle at geophysically relevant pressure and temperature conditions that are still experimentally inaccessible. Rheology is a key factor, which has strong influence on the complicated mantle dynamics implied by seismological observations and other sources. Large-scale atomistic simulations will be performed on massively parallel machines using a combination of first-principles quantum mechanical (QM), classical molecular dynamics (MD) and hybrid QM/MD approaches. The simulations will predict several crucial rheology-related properties and processes in major silicate and oxide mantle minerals, including point defects, dislocations and associated long-range mechanical phenomena, and deformation of polycrystalline minerals under anisotropic stresses. The resulting massive multivariate datasets will be visualized in an immersive and interactive environment to gain insight into complex mechanical behavior of minerals. The aforementioned interdisciplinary research activities will be integrated into teaching/learning/training activities. They include training a new generation of students with expertise in computational and Earth materials sciences (through a dual-degree graduate program in computer and geosciences), developing an Access Grid-based remote teaching/learning environment (encompassing minority institutions in Louisiana) and organizing computational science workshop for underrepresented groups and Mardi Gras conference that have previously been organized at Louisiana State University (LSU). The successful execution of these activities will have broad substantive involvement of students and faculty including those from underrepresented groups in research and education at the interface of geopysics and information technology. The project will exploit the world-class 1024-processor Linux cluster, Immersadesk virtual reality, and Access Grid facilities recently acquired by LSU. --