Laboaratory J A Dieudonne UMR CNRS and Universite de Nice Sophia Antipolis Parc Valrose F Nice Cedex France

Sophia Antipolis

Description

Niveau: Supérieur, Doctorat, Bac+8
Laboaratory J.-A. Dieudonne, UMR CNRS 6621 and Universite de Nice-Sophia Antipolis, Parc Valrose, F-06108 Nice Cedex 02, France PACS: 65N30; 65M60; 65M70 famous one is certainly the quadrangle based spectral element method which was developed in the 80s [17,21] and then largely adopted (see, e.g. [10,16] and references herein). However, in order to handle highly complex geometries the use of triangular (tetrahedral in 3D) elements is generally preferred. Therefore, hp-finite element methods have been deeply investigated during this last decade [2,6,22]. In the field of spectral methods, it was suggested to restrict the Gauss–Lobatto mesh of the * Keywords: Spectral elements; Triangular and quadrangular mesh elements; Fekete points; Gauss–Lobatto points 1. Introduction The main advantage of standard spectral methods relies on the exponential convergence property as soon as smooth solutions are involved (see, e.g. [5,11,20]). The main drawback is their inability to handle complex geometries. Di?erent strategies are, however, possible to overcome this di?culty. One of these consists in the combination of the standard spectral approximation with a penalty method [9], but the most Received 13 October 2003; received in revised form 15 January 2004; accepted 15 January 2004 Available online 21 February 2004 Abstract In this paper, we compare a triangle based spectral element method (SEM) with the classical quadrangle based

• following linear


• fekete point


• i?1 ui


• based spectral


• kd polynomials


• pn ?t


• s? ?


• spectral methods


• ?t ?


• convergence properties

Sujets

• Antipolis
• Fekete problem
• Spectral method

Informations

• Publié par: profil-zyak-2012
• Nombre de lectures: 14
• Langue: English

Extrait

JournalofComputationalPhysics198(2004)349–362 www.elsevier.com/locate/jcp Spectralelementmethodsontrianglesand quadrilaterals:comparisonsandapplications RichardPasquetti,FrancescaRapetti * LaboaratoryJ.-A.Dieudonne
,UMRCNRS6621andUniversite
deNice-SophiaAntipolis,ParcValrose,F-06108NiceCedex02,France Received13October2003;receivedinrevisedform15January2004;accepted15January2004 Availableonline21February2004 Abstract Inthispaper,wecompareatrianglebasedspectralelementmethod(SEM)withtheclassicalquadranglebasedSEM andwithastandardspectralmethod.Forthesakeofcompleteness,thetriangle-SEM,makinguseoftheFeketepoints ofthetriangle,isfirstrevisited.Therequirementofahighlyaccuratequadratureruleisparticularlyemphasized.Thenit isshownthattheconvergencepropertiesofthetriangle-SEMcomparewellwiththoseoftheclassicalSEM,bysolving anellipticequationwithsmooth(butsteep)analyticalsolution.Itisalsoprovednumericallythattheconditionnumber growssignificantlyfasterfortrianglesthanforquadrilaterals.Finally,theattentionisfocusedonadiffractionproblem toshowthehighflexibilityofthetriangle-SEM.
2004ElsevierInc.Allrightsreserved. PACS:65N30;65M60;65M70 Keywords:Spectralelements;Triangularandquadrangularmeshelements;Feketepoints;Gauss–Lobattopoints 1.Introduction Themainadvantageofstandardspectralmethodsreliesontheexponentialconvergencepropertyas soonassmoothsolutionsareinvolved(see,e.g.[5,11,20]).Themaindrawbackistheirinabilitytohandle complexgeometries.Differentstrategiesare,however,possibletoovercomethisdifficulty.Oneofthese consistsinthecombinationofthestandardspectralapproximationwithapenaltymethod[9],butthemost famousoneiscertainlythequadranglebasedspectralelementmethodwhichwasdevelopedinthe80s [17,21]andthenlargelyadopted(see,e.g.[10,16]andreferencesherein). However,inordertohandlehighlycomplexgeometriestheuseoftriangular(tetrahedralin3D)elements isgenerallypreferred.Therefore,hp-finiteelementmethodshavebeendeeplyinvestigatedduringthislast decade[2,6,22].Inthefieldofspectralmethods,itwassuggestedtorestricttheGauss–Lobattomeshofthe * Correspondingauthor. E-mailaddress:[e-mail](F.Rapetti). 0021-9991/$-seefrontmatter
2004ElsevierInc.Allrightsreserved. doi:10.1016/j.jcp.2004.01.010