SCREMS: Scientific Computing Research Environment for the Mathematical Sciences at Buffalo

布法罗大学数学系的一组应用数学家建议购买一个分布式共享内存多处理器集群、工作站和补充软件。新设备将用于支持正在进行的建模和模拟数学研究项目。这些研究项目包括光纤系统中的非线性和随机效应的研究，肾脏血液动力学中的耦合和不确定性的分析，以及晶体表面模型中的动力学和不适定性的研究，特别是在强各向异性的条件下。这些项目有几个共同的特点。由于每个项目的跨学科合作，共同研究者需要一个计算环境，使他们能够快速开发初始模型系统，通常是偏微分方程或常微分方程系统，并探索这些方程的解决方案，而无需大量的编码。随着模型的成熟，控制方程可以通过中等规模的数值模拟来求解。这些动态模型的有效数值模拟需要具有大存储容量和高带宽的计算基础设施。再次，由于所研究的研究项目的跨学科性质，参数探索和模型验证需要大量的计算工作。此外，随机效应可以在模型动力学中发挥重要作用，增加计算负担。通过先进的可视化软件，可以方便地将模型结果与实验结果进行比较，并将这些结果进行交流。拟购买的集群将满足这些要求，并将大大加强共同调查员？布法罗大学数学系的一组应用数学家建议购买一个分布式共享内存多处理器集群、工作站和辅助软件。新设备将用于支持计算密集型工程和生物医学应用领域正在进行的建模和仿真研究。由于合作研究人员在高度跨学科的研究领域工作，在这些项目中取得的数学研究进展迅速影响应用科学领域。共同研究者的跨学科性质为研究生的培训提供了一个刺激的环境。合作研究者还为本科生提供了体验数学研究的机会。拟议的设备将有利于这些学生-研究生在他们的研究生涯，和本科生，因为他们离开大学研究生学习或进入劳动力市场。

A group of applied mathematicians in the Department of Mathematics atthe University at Buffalo proposes to purchase a distributed-shared memory multiprocessor cluster,workstations, and supplementary software. The new equipment will be used to support ongoing mathematical research projects in modeling and simulations. These research projects include the study of nonlinear and stochastic effects in fiber optics systems, analysis of coupling and uncertainty in renal hemodynamics, and an investigation of the dynamics and ill-posedness in models of crystalline surfaces especially under conditions of strong anisotropy. These projects share several features. Becauses of the interdisciplinary collaboration of each project, the co-investigators require a computing environment that allows them to rapidly develop initial model systems, usually systems of partial or ordinary differential equations, and explore solutions to these equations without extensive coding. As a model matures, the governing equations may then be solved by moderate-sized numerical simulations. Efficient numerical simulation of these dynamic models requires a computing infrastructure with large memory capacity and high bandwidth. Again because of the interdisciplinary nature of the research projects under study, parameter exploration and model validation require significant computational effort. Moreover, stochastic effects can play an important role in the model dynamics, adding to the computing burden. Comparison of model results with experiments, and communication of these results, is facilitated by advanced visualization software. The cluster to be purchased will serve these requirements, and will significantly enhance the co-investigators? research capabilities.A group of applied mathematicians in the Department of Mathematics at the University at Buffalo proposes to purchase a distributed-shared memory multiprocessor cluster, workstations, and supplementary software. The new equipment will be used to support ongoing modeling and simulation research in fields of engineering and bio-medical application that are computationally intensive. Because the co-investigators work in highly interdisciplinary fields of study, the mathematical research advances made in these projects quickly impact fields of applied science. The cross-disciplinary nature of the co-investigators provides a stimulating environment for the training of graduate students. The co-investigators also provide opportunities for undergraduates to experience research in mathematics. The proposed equipment will benefit these students - the graduate students in their research careers, and the undergraduate students as they leave the university for graduate study or enter the workforce.