Parallel Computing Applied to the Simulation of Electron Transport in Advanced Semiconductor Devices

在先进电子器件的模拟中出现的计算密集型问题中固有的并行性将被识别并被用于开发新的软件工具。目标是在特定机器(如MasPar MP-1216)上使用的算法的优化以及迄今未探索的设备现象的模拟方面的进展。以前的研究涉及到纳米结构半导体器件中量子输运的数值密集型模拟器的开发。从这项工作中提取的一组丰富的矩阵问题，包括复值线性系统解、特征值问题以及正则结构矩阵上的QR和SV分解，已被确定为未来并行处理机工作的候选对象。在电磁波导的处理中也出现了非常类似的矩阵问题，这项工作的范围可以很容易地扩展到包括这一领域的问题。

Parallelism inherent in the computationally intensive problems arising in the simulation of advanced electron devices will be identified and exploited for the development of novel software tools. Advances, both in the optimization of algorithms for use on specific machines such as a MasPar MP-1216, as well as in the simulation of hitherto unexplored device phenomena are targeted. Previous research involved the development of a numerically intensive simulator for quantum transport in nanostructured semiconductor devices. A rich set of matrix problems drawn from this work, including complex valued linear systems solutions, eigenvalue problems as well as QR, and SV decompositions on regularly structured matrices, have been identified as candidates for future work on parallel processing machines. Quite similar matrix problems arise also in the treatment of electromagnetic waveguides, and the scope of this work can be readily expanded to include problems in this field as well.