Collaborative Research; ParCYCLIC: Internet-Enabled Simulation of Earthquake Liquefaction Response on Parallel Computers

这项工作是加州大学圣地亚哥分校(UCSD)和斯坦福大学的研究人员合作完成的。其目的是探索和利用先进的计算和信息技术，进一步开发一种最先进的非线性有限元程序(循环)，用于地震地面响应和液化模拟。地震岩土现象建模和模拟的校准代码将与先进的计算方法相结合，以促进大规模系统的模拟，并扩大实际应用范围。建议的工作涉及三个主要研究内容：1.进一步开发计算程序Cycle：这一开发将引入参数敏感性和优化选项。计算程序的能力将从二维(2D)扩展到3D模拟。此外，还将包括传输边界功能。2.并行和分布式计算能力：通过采用为并行和分布式计算环境设计的最先进的软件范例和求解器技术，循环将得到显著增强。具体地说，将采用和实施多任务单程序多数据(SPMD)程序范例。此外，还将开发并行稀疏矩阵求解器，并将其整合到计算引擎中，以提高性能和扩展模拟能力。基于互联网的交互式网络环境：除了目前的1D远程执行和可视化能力(http://casagrande.ucsd.edu)，将通过并行计算促进对2D和3D的扩展)。新的有用功能将包括一个广泛的在线输入地震记录库，以及一个土壤材料库，其中包含为沙土、粉土和粘土土壤预先定义的校准模型。将在整个开发过程中解决用户友好性问题。将部署先进的联网和网络浏览技术，使远程用户能够以互动和动态的方式查看结果。在开发支持互联网的模拟环境方面获得的经验将与其他研究人员分享，并可能有助于使许多其他有用的计算程序在未来的应用中通过互联网广泛使用。

This effort is a collaboration between researchers at University of California, San Diego (UCSD) and Stanford University. The objective is to explore and utilize advanced computational and information technologies to further develop a state-of-the-art nonlinear finite element program (CYCLIC) for earthquake ground response and liquefaction simulation. Calibrated codes for modeling and simulation of earthquake geotechnical phenomena will be combined with advanced computational methodologies to facilitate the simulation of large-scale systems and broaden the scope of practical applications. The proposed work involves three main research elements: 1. Further development of the computational program CYCLIC: This development will introduce parameter sensitivity and optimization options. The capacity of the computational program will be extended from two-dimensional (2D) to 3D simulation. In addition, a transmitting boundary feature will be included. 2. Parallel and distributed computing capability: CYCLIC will be significantly enhanced by incorporating state-of-the-art software paradigm and solver technologies that are designed for parallel and distributed computing environments. Specifically, a multitask Single-Program-Multiple-Data (SPMD) program paradigm will be employed and implemented. Furthermore, parallel sparse matrix solvers will be developed and incorporated into the computational engine to improve performance and extend the simulation capabilities.3. Internet-based interactive web environment: In addition to the current 1D remote execution and visualization capability (http://casagrande.ucsd.edu), extensions to 2D and 3D will be facilitated through parallel computing. Novel useful features will include an extensive library of on-line input earthquake records, and a soil material library with pre-defined calibrated models for sand, silt, and clay soils. User friendliness will be addressed throughout the development process. Advanced Internet-enabled and web browsing technologies will be deployed to allow remote users to view the results in an interactive and dynamic manner. The experience gained in the development of the Internet-enabled simulation environment will be shared with other researchers and may help make many other useful computational programs widely accessible over the Internet for future applications.