Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6422600 | Journal of Computational and Applied Mathematics | 2014 | 14 Pages |
Abstract
A variant of the global generalized Hessenberg method is presented which allows varying preconditioning at each restart. Theoretical results that relate the residual norm of this new method with its original version are developed. As two special variants, the flexible global GMRES method and the flexible global CMRH method are investigated both theoretically and experimentally. Numerical examples are conducted to illustrate the performance of these two flexible global methods in comparison with both the original global methods and weighted global methods.
Related Topics
Physical Sciences and Engineering
Mathematics
Applied Mathematics
Authors
Ke Zhang, Chuanqing Gu,