Radu, Bogdan (2022)
Finite element mass lumping for H(div) and H(curl).
Technische Universität Darmstadt
doi: 10.26083/tuprints-00021948
Dissertation, Erstveröffentlichung, Verlagsversion
Kurzbeschreibung (Abstract)
In this work, we consider the efficient implementation of finite element approximations for porous media, poroelasticity, and wave propagation problems. This is conducted by using mass-lumping, an approximation technique that enables a faster inversion of the mass matrix. This method has been primarily used for discretizing wave propagation problems in H1, as it allows for the efficient application of explicit time-stepping schemes. Another example of the usage of mass-lumping is the mixed finite element discretization of porous medium flow. Here, mass-lumping is utilized to reduce the algebraic saddle-point structure to a symmetric positive definite system, which is a lot easier to solve. In this work, we specifically examine mass-lumping strategies for the functional spaces H(div) and H(curl). In the first part of this thesis, we investigate the discretization of porous medium flow and poroelasticity. We will analyze already existing first-order convergent methods with mass-lumping and propose extensions that have better convergence properties and are optimal in the number of degrees of freedom. We will complement this with a rigorous error analysis, which confirms the accuracy of the method. In the second part, we consider discretization techniques with mass-lumping for the acoustic wave equation in H(div) and for Maxwell's equations in H(curl). Specifically, we will propose first and second-order accurate methods. Moreover, we will propose a new technique for the first-order convergent method that will cut the number of degrees of freedom in half (or even more). For all the methods we propose, we also provide full error convergence analyses. Furthermore, all our results will be validated by numerical experiments.
Typ des Eintrags: | Dissertation | ||||
---|---|---|---|---|---|
Erschienen: | 2022 | ||||
Autor(en): | Radu, Bogdan | ||||
Art des Eintrags: | Erstveröffentlichung | ||||
Titel: | Finite element mass lumping for H(div) and H(curl) | ||||
Sprache: | Englisch | ||||
Referenten: | Egger, Prof. Dr. Herbert ; Kanschat, Prof. Dr. Guido | ||||
Publikationsjahr: | 2022 | ||||
Ort: | Darmstadt | ||||
Kollation: | 144 Seiten | ||||
Datum der mündlichen Prüfung: | 31 März 2022 | ||||
DOI: | 10.26083/tuprints-00021948 | ||||
URL / URN: | https://tuprints.ulb.tu-darmstadt.de/21948 | ||||
Kurzbeschreibung (Abstract): | In this work, we consider the efficient implementation of finite element approximations for porous media, poroelasticity, and wave propagation problems. This is conducted by using mass-lumping, an approximation technique that enables a faster inversion of the mass matrix. This method has been primarily used for discretizing wave propagation problems in H1, as it allows for the efficient application of explicit time-stepping schemes. Another example of the usage of mass-lumping is the mixed finite element discretization of porous medium flow. Here, mass-lumping is utilized to reduce the algebraic saddle-point structure to a symmetric positive definite system, which is a lot easier to solve. In this work, we specifically examine mass-lumping strategies for the functional spaces H(div) and H(curl). In the first part of this thesis, we investigate the discretization of porous medium flow and poroelasticity. We will analyze already existing first-order convergent methods with mass-lumping and propose extensions that have better convergence properties and are optimal in the number of degrees of freedom. We will complement this with a rigorous error analysis, which confirms the accuracy of the method. In the second part, we consider discretization techniques with mass-lumping for the acoustic wave equation in H(div) and for Maxwell's equations in H(curl). Specifically, we will propose first and second-order accurate methods. Moreover, we will propose a new technique for the first-order convergent method that will cut the number of degrees of freedom in half (or even more). For all the methods we propose, we also provide full error convergence analyses. Furthermore, all our results will be validated by numerical experiments. |
||||
Alternatives oder übersetztes Abstract: |
|
||||
Status: | Verlagsversion | ||||
URN: | urn:nbn:de:tuda-tuprints-219485 | ||||
Sachgruppe der Dewey Dezimalklassifikatin (DDC): | 500 Naturwissenschaften und Mathematik > 510 Mathematik | ||||
Fachbereich(e)/-gebiet(e): | 04 Fachbereich Mathematik 04 Fachbereich Mathematik > Numerik und wissenschaftliches Rechnen |
||||
Hinterlegungsdatum: | 16 Aug 2022 12:03 | ||||
Letzte Änderung: | 17 Aug 2022 12:21 | ||||
PPN: | |||||
Referenten: | Egger, Prof. Dr. Herbert ; Kanschat, Prof. Dr. Guido | ||||
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: | 31 März 2022 | ||||
Export: | |||||
Suche nach Titel in: | TUfind oder in Google |
Frage zum Eintrag |
Optionen (nur für Redakteure)
Redaktionelle Details anzeigen |