Efficient Iterative Solution of Large Linear Systems on Heterogeneous Computing Systems

Acknowledgments This dissertation concludes four years of intense scientific research in the numerical analysis group at Delft University of Technology. It was an unforgettable experience and I learned many valuable lessons, both personally and professionally. I would not have been able to finish such a large body of work without the help of many people, who I want to thank here. First of all, I want to mention that this PhD research was financially supported by the Delft Centre for Computational Science and Engineering (DCSE). Also, this work is performed as part of the research project " Development of an Immersed Boundary Method, Implemented on Cluster and Grid Computers, with Application to the Swimming of Fish " , which is joint work with Barry Koren and Yunus Hassen from CWI. The Netherlands Organisation for Scientific Research (NWO) is gratefully acknowledged for the use of the DAS–3. My most sincere gratitude goes to my daily supervisor and copromotor Martin van Gijzen. His endless enthusiasm and constructive criticism made working with him a real pleasure. Martin always managed to provide the perfect balance between freedom and supervision. If I am a better scientist than before, it is largely due to Martin's influence. I could truly not have asked for a better supervisor. The other person that I am greatly indebted to is my promotor prof.dr. Kees Vuik. He provided invaluable feedback during the many meetings that we had. It was a true privilege being able to work in his numerical analysis group. I also thoroughly enjoyed working with dr. Gerard Sleijpen. He truly knows everything about everything and I loved picking his brain for inspiration and knowledge. I also immensely enjoyed our time together in Gifu and Tokyo with Yoshi and Emiko. I was fortunate enough to have numerous inspiring conversations with Scott MacLachlan, Jok Tang, and Peter Sonneveld. Their immense body of knowledge has helped me improve my understanding of difficult mathematical material in many ways. I want to thank my colleagues Barry Koren and Yunus Hassen from CWI for many constructive meetings during the beginning of my PhD research. I only regret that our research paths deviated somewhat towards the end. That is typical of being on the frontier of science, where it is always hard to predict how things will go. I thank from the bottom of my heart prof. Kuniyoshi Abe from Gifu Shotoku University and prof. Emiko …