@inproceedings{dfed3373bd6b4591b4fa579765a06c47,
title = "How to make a domain decomposition method more robust",
abstract = "The simulation of processes in highly heterogeneous media comes with many challenges. In particular many domain decomposition methods do not perform well in this case, specially if the decomposition into subdomains does not accommodate the coefficient variations. For three popular domain decomposition methods (two level additive Schwarz, BDD and FETI) we have proposed a remedy to this problem in previous work with coauthors. Here we present the strategy which was used by applying it to the Hybrid Schwarz preconditioner. It is based on identifying a bottleneck estimate in the proof of convergence which cannot be satisfied for the entire solution space. Then the part of the solution which is problematic is isolated via a generalized eigenvalue problem and solved separately.",
keywords = "Domain decomposition, Heterogeneous problems, Hybrid Schwarz, Robustness",
author = "Nicole Spillane",
year = "2014",
month = jan,
day = "1",
doi = "10.1007/978-3-662-43880-0\_40",
language = "English",
isbn = "9783662438794",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "355--362",
booktitle = "Large-Scale Scientific Computing - 9th International Conference, LSSC 2013, Revised Selected Papers",
note = "9th International Conference on Large-Scale Scientific Computations, LSSC 2013 ; Conference date: 03-06-2013 Through 07-06-2013",
}