Stefan Röllin,
Parallel Iterative Solvers in Computational Electronics.
2005; 168 pages. € 64,00. ISBN 3-89649-983-1

This work investigates parallel iterative solvers for large sparse linear systems of equations that arise in semiconductor device simulation, an important area of computational electronics.
In the first part of this work the most important algorithms for the iterative solution of sparse linear systems are presented. The optimal set of these algorithms is then evaluated for realistic applications through an extensive number of numerical experiments.
In the second part the parallelisation of the algorithms is considered. The target architectures for the parallelisation are shared memory multiprocessors. The numerical results show that a good overall speedup can be reached.

About the Author:

Stefan Röllin was born in 1974 in Baar, Switzerland. He studied mathematics at the Swiss Federal Institute of Technology in Zurich (ETHZ). During his studies he concentrated on numerical mathematics. He has written his master thesis on the subject "Lanczos-type product methods based on two-term recursions" at the Seminar of Applied Mathematics under the supervision of Prof. Martin H. Gutknecht. After his studies, in April 2000, he joined the Integrated Systems Laboratory at ETHZ. His main research interests are iterative linear solvers in general and their parallelisation on shared memory multiprocessor machines.

Keywords: linear systems of equations, sparse matrices, unsymmetric permutations, preconditioned iterative methods, parallel algorithms, shared memory multiprocessors, semiconductor device simulation.

Series in Microelectronics

Direkt bestellen bei / to order directly from: Hartung.Gorre@t-online.de