Szczegóły publikacji
Opis bibliograficzny
Open source JAVA implementation of the parallel multi-thread alternating direction isogeometric L2 projections solver for material science simulations — Otwarte oprogramowanie zawierające implementację w języku Java równoległego solwera wielowątkowego metody zmienno-kierunkowych izogeometrycznych L2 projekcji w zastosowaniach do symulacji inżynierii materiałowej / Grzegorz GURGUL, Maciej WOŹNIAK, Marcin ŁOŚ, Danuta SZELIGA, Maciej PASZYŃSKI // Computer Methods in Materials Science : quarterly / Akademia Górniczo-Hutnicza ; ISSN 1641-8581. — Tytuł poprz.: Informatyka w Technologii Materiałów. — 2017 — vol. 17 no. 1, s. 1–11. — Bibliogr. s. 10–11, Abstr., Streszcz. — KomPlasTech 2017 : XXIV Conference on Computer Methods in Materials Technology : Zakopane, Poland 15–18 January 2017
Autorzy (5)
Słowa kluczowe
Dane bibliometryczne
ID BaDAP | 104202 |
---|---|
Data dodania do BaDAP | 2017-02-21 |
Rok publikacji | 2017 |
Typ publikacji | referat w czasopiśmie |
Otwarty dostęp | |
Czasopismo/seria | Computer Methods in Materials Science |
Abstract
This paper describes multi-thread parallel open source JAVA implementation of an alternating directions isogeometric L2 projections solver. The solver enables for fast numerical simulations of time dependent problems. To apply our solver, the time-dependent problem must be discretized using isogeometric finite element method with B-spline basis functions in spatial domain. The problem is solved using explicit method with respect to time. The application of the explicit method with B-spline based spatial discretization results in a sequence of isogeometric L2 projections that can be solved using our fast solver. The computational cost of solution of either 2D or 3D problem is linear O(N) in every time step. This cost is lower than the cost of traditional multi-frontal solvers, delivering O(N1.5) computational cost for 2D problems and O(N2) computational cost for 3D problems. This cost is also lower from any iterative solver, delivering O(Nk) computational cost, where k is the number of iterations, which depends on the particular iterative solver algorithm. Our algorithm is used for numerical solution of 3D elasticity problem.