Dear Yassine, Yes, you are right. There is a superconvergence phenomenon in both cases. I do not see what is the reason, the integration method is correct. May be, the fact that the mesh is nearly structured. It would be interesting to make the experiment in a fully non-structured case.
Best regards, Yves. Le 16/02/2016 11:05, Yassine ZAIM a écrit : > Dear Dr. Yves, > Thank you so much for your answer and your time. > > I try with "MESH_NOISED = 1;" > For exemple : > for "NX = 127; /Number of dof : 16384/" : I found the error_H1 > = /1.239379991943454e-05/ > for "NX = 255; /Number of dof : 65536/" : I found the error_H1 > =/3.083252284962158e-06/ > /Order = Log2(//1.239379991943454e-05///3.083252284962158e-06)=//2.0071./ > Maybe we will have the superconvergence phenomena in the both cases. > > Dear Thomas, > Thank you so much for your answer and your time. > > For me I work with C++, I don't use any interfaces, because I need to > do some modifications later. The functions used for computation of > error are getfem::asm_L2_norm and getfem::asm_H1_norm. > > Best regards. > > > 2016-02-16 9:36 GMT+00:00 Thomas Linse <[email protected] > <mailto:[email protected]>>: > > How did you calculate the L2 and H1 norms? I found "correct" > convergence for a similar 1D problem when calculating the norms > "by hand" (using Gauss integration), whereas the norms calculated > using "gf_compute" from the scilab interface showed a larger > convergence rate. > > Maybe "gf_compute" (which callsgetfem::asm_L2_normor dasm_H1_norm) > does not use the necessary order of integration? > > Kind regards, > > Thomas > > > On 16.02.2016 09:59, Yves Renard wrote: >> >> Dear Yassine, >> >> Superconvergence phenomena are quite common when a structured >> mesh is used. If you use the option "MESH_NOISED", this should >> (unfortunately !) disapear. >> >> Best Regards, >> >> Yves. >> >> >> Le 15/02/2016 15:01, Yassine ZAIM a écrit : >>> Hi, >>> I want to ask you if someone has already solved the laplacian >>> problem that exists in the tests directory with the Q_1 element >>> in 2 dimension. Because I find the order of convergence equal to >>> "2" for the L2 and H1 norme. what is not normal. >>> Best regards. >>> >>> Bonjour, >>> S.v.p j'aimerais savoir si quelqu'un de vous a déjà testé le >>> problème de laplacian qui exist dans la répertoire des tests >>> avec l'élément Q_1 en dimension 2. J'ai testé ça et je trouve le >>> même ordre pour les deux norme L2 et H1 (Ordre(L2) = Ordre(H1) = >>> 2). Ce qui contredit la théorie. >>> Merci d'avance pour vos réponse. >>> Cordialement. >>> >>> -- >>> *ZAIM Yassine * >>> *PhD Student in Applied Mathematics* >>> * >>> * >>> >>> >>> _______________________________________________ >>> Getfem-users mailing list >>> [email protected] <mailto:[email protected]> >>> https://mail.gna.org/listinfo/getfem-users >> >> >> -- >> >> Yves Renard ([email protected] >> <mailto:[email protected]>) tel : (33) 04.72.43.87.08 >> Pole de Mathematiques, INSA-Lyon fax : (33) 04.72.43.85.29 >> 20, rue Albert Einstein >> 69621 Villeurbanne Cedex, FRANCE >> http://math.univ-lyon1.fr/~renard >> <http://math.univ-lyon1.fr/%7Erenard> >> >> --------- >> >> >> _______________________________________________ >> Getfem-users mailing list >> [email protected] <mailto:[email protected]> >> https://mail.gna.org/listinfo/getfem-users > > > _______________________________________________ > Getfem-users mailing list > [email protected] <mailto:[email protected]> > https://mail.gna.org/listinfo/getfem-users > > > > > -- > *ZAIM Yassine * > *PhD Student in Applied Mathematics* > * > * -- Yves Renard ([email protected]) tel : (33) 04.72.43.87.08 Pole de Mathematiques, INSA-Lyon fax : (33) 04.72.43.85.29 20, rue Albert Einstein 69621 Villeurbanne Cedex, FRANCE http://math.univ-lyon1.fr/~renard ---------
_______________________________________________ Getfem-users mailing list [email protected] https://mail.gna.org/listinfo/getfem-users
