TU Darmstadt / ULB / TUbiblio

GPU-accelerated Optimization with Structural Analysis

Weber, Daniel ; Mueller-Roemer, Johannes Sebastian ; Ferreira, Stephanie ; Besler, Sebastian ; Stork, André (2023)
GPU-accelerated Optimization with Structural Analysis.
International Association for the Engineering Analysis Community (NAFEMS World Congress) 2023. Tampa, Florida (15.05.2023 - 18.05.2023)
Konferenzveröffentlichung, Bibliographie

Kurzbeschreibung (Abstract)

In this work we demonstrate how optimization scenarios can benefit from GPU-accelerated structural analysis simulations with examples in the domains design of experiment, topology optimization and multi-phase topology optimization. We show that it is important to also come up with a GPU-accelerated optimization algorithm to minimize overall run times. With an efficient novel linear solver, the resulting speedups enable in-situ optimizations, i.e., the possibility to inspect the optimization algorithm during run time enabling to gather a deeper understanding of the algorithms or to abort optimization runs earlier if configurations have been specified incorrectly.

Typ des Eintrags: Konferenzveröffentlichung
Erschienen: 2023
Autor(en): Weber, Daniel ; Mueller-Roemer, Johannes Sebastian ; Ferreira, Stephanie ; Besler, Sebastian ; Stork, André
Art des Eintrags: Bibliographie
Titel: GPU-accelerated Optimization with Structural Analysis
Sprache: Englisch
Publikationsjahr: 17 Mai 2023
Veranstaltungstitel: International Association for the Engineering Analysis Community (NAFEMS World Congress) 2023
Veranstaltungsort: Tampa, Florida
Veranstaltungsdatum: 15.05.2023 - 18.05.2023
Kurzbeschreibung (Abstract):

In this work we demonstrate how optimization scenarios can benefit from GPU-accelerated structural analysis simulations with examples in the domains design of experiment, topology optimization and multi-phase topology optimization. We show that it is important to also come up with a GPU-accelerated optimization algorithm to minimize overall run times. With an efficient novel linear solver, the resulting speedups enable in-situ optimizations, i.e., the possibility to inspect the optimization algorithm during run time enabling to gather a deeper understanding of the algorithms or to abort optimization runs earlier if configurations have been specified incorrectly.

Freie Schlagworte: Computer simulation, Finite element analysis, Optimization
Fachbereich(e)/-gebiet(e): 20 Fachbereich Informatik
20 Fachbereich Informatik > Graphisch-Interaktive Systeme
Hinterlegungsdatum: 09 Jul 2024 09:07
Letzte Änderung: 09 Jul 2024 09:07
PPN:
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen