A class of optimized row projection methods for solving large nonsymmetric linear systems
The optimal and the accelerated row projection methods for solving large nonsymmetric linear systems were discussed. These algorithms use a partition strategy into blocks based on sequential estimations of their condition numbers. These algorithms are extremely fast and efficient, but they do not co...
Guardado en:
| Publicado: |
2002
|
|---|---|
| Materias: | |
| Acceso en línea: | https://bibliotecadigital.exactas.uba.ar/collection/paper/document/paper_01689274_v41_n4_p499_Scolnik http://hdl.handle.net/20.500.12110/paper_01689274_v41_n4_p499_Scolnik |
| Aporte de: |
| Sumario: | The optimal and the accelerated row projection methods for solving large nonsymmetric linear systems were discussed. These algorithms use a partition strategy into blocks based on sequential estimations of their condition numbers. These algorithms are extremely fast and efficient, but they do not converge always. A block splitting algorithm which fulfills the conditions based on the sequential estimations of the condition numbers was also discussed. The performance of the projection methods was highly dependent on the way in which the rows of the matrix were split into blocks. |
|---|