Numerical algorithms for large-scale memory-bound problems

在UBC的科学计算实验室(SCL)开发的算法正在突破我们现有硬件的界限，即使应用程序需要更大的问题大小。SCL正在进行的项目包括高维(3+)模拟、大规模逆问题、用于验证和最优控制的水平集方法、大型稀疏线性系统的解以及搜索引擎的数值技术。随着我们的学生人数不断增长，为了快速将软件从基准测试转换到真实世界的示例，我们需要访问集成到我们的计算环境中的大内存机器，并获得本地技术支持。*我们还要求一个单独的、较小的可视化工作站，可实际放置在我们的实验室中，并允许以图形方式探索结果。

Algorithms developed in the Scientific Computing Lab (SCL) at UBC are pushing the boundaries of our existing hardware, even as applications demand larger problem sizes.  Projects underway at SCL include high (3+) dimensional simulations, large scale inverse problems, level set methods applied to verification and optimal control, solution of large sparse linear systems, and numerical techniques for search engines.  Our student population is growing, and in order to rapidly transition software from benchmark to real-world examples, we require access to a large memory machine integrated into our computing environment and with local technical support.  We are also requesting a separate, smaller visualization workstation which can be physically located in our lab and will permit graphical exploration of results.