Efficient Parallel Solvers for the FireStar3D Wildfire Numerical Simulation Model
| Autor: | Oleg Bessonov, Sofiane Meradji |
|---|---|
| Přispěvatelé: | Institut de Mathématiques de Toulon - EA 2134 (IMATH), Université de Toulon (UTLN) |
| Rok vydání: | 2019 |
| Předmět: |
Computer science
ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION MathematicsofComputing_NUMERICALANALYSIS 020101 civil engineering 02 engineering and technology 0201 civil engineering law.invention preconditioners Multigrid method Factorization law ill-conditioned linear systems parallelization Conjugate gradient method 0202 electrical engineering electronic engineering information engineering Fluid dynamics Applied mathematics [MATH]Mathematics [math] Conjugate gradient multigrid Computer simulation Preconditioner Linear system smoothers [INFO.INFO-MO]Computer Science [cs]/Modeling and Simulation LU decomposition 020201 artificial intelligence & image processing |
| Zdroj: | Lecture Notes in Computer Science ISBN: 9783030256357 PaCT Malyshkin V. (eds) Parallel Computing Technologies. PaCT 2019. Lecture Notes in Computer Science, vol 11657. Springer, Cham Malyshkin V. (eds) Parallel Computing Technologies. PaCT 2019. Lecture Notes in Computer Science, vol 11657. Springer, Cham, pp.140-150, 2019, 978-3-030-25635-7. ⟨10.1007/978-3-030-25636-4_11⟩ |
| DOI: | 10.1007/978-3-030-25636-4_11 |
| Popis: | International audience; This paper presents efficient parallel methods for solvingill-conditioned linear systems arising in fluid dynamics problems. Thefirst method is based on the Modified LU decomposition, applied as apreconditioner to the Conjugate gradient algorithm. Parallelization ofthis method is based on the use of nested twisted factorization. Anothermethod is based on a highly parallel Algebraic multigrid algorithm with anew smoother developed for anisotropic grids. Performance comparisonsdemonstrate superiority of new methods over commonly used variants ofthe Conjugate gradient method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|