[dune-pdelab] inerpolate function in PDELab

Steffen Müthing steffen.muething at iwr.uni-heidelberg.de
Fri Jul 25 17:22:25 CEST 2014


Hi Xavier,

Am 25.07.2014 um 17:06 schrieb KUHN Xavier EUROGICIEL INGENIERIE <Xavier.KUHN at cea.fr>:

> Hi Steffen,
> 
>> How exactly did you set the different initial value?
> 
> For velocity in LU_Velocity class:
> inline void evaluateGlobal(const DomainType & x, RangeType & y) const
>  {
>    y[0]=1;   
>    y[1] = 2;
>  }
> For pressure in ZeroVectorFunction class called by ZeroScalarFunction:
> 
> inline void evaluateGlobal(const DomainType & x, RangeType & y) const
>  {
>    y[0]=3;
>  }
> 

that all looks fine...

> 
>> Oh, and another question: What compiler are you using? We discovered a bug in PDELab that might bite you here if you are using an older compiler.
> 
> This morning my OS was updated from debian 6.0 (squeeze) to debian 7 (wheezy) and now it is ok... So as you said the problem is the compiler.
> With wheezy the compiler is gcc 4.7.2 and with squeeze it was gcc 4.4.5.

Yes, we’ve had some problems with gcc 4.4.5 and subspaces - those are fixed in the branch releases/2.0, but the problem was discovered after
the 2.0.0 release. But with gcc 4.7, you should be fine (and have faster compile times, as well… :-) ).

Steffen

> 
> Thank you for your help.
> 
> Best 
> Xavier 
> 
> -----Message d'origine-----
> De : Steffen Müthing [mailto:steffen.muething at iwr.uni-heidelberg.de] 
> Envoyé : vendredi 25 juillet 2014 10:58
> À : KUHN Xavier EUROGICIEL INGENIERIE
> Cc : dune-pdelab mailing list
> Objet : Re: [dune-pdelab] inerpolate function in PDELab
> 
> Hi Xavier,
> 
> Am 23.07.2014 um 15:39 schrieb KUHN Xavier EUROGICIEL INGENIERIE <Xavier.KUHN at cea.fr>:
> 
>> Hello,
>> 
>> I am currently learning how to use PDELab to implement Navier-Stokes equations.
> 
> great!
> 
>> 
>> In navier-stokes folder of dune-pdelab-howto, interesting examples are already implemented but I do not understand how interpolate function works. When I run cgstokes_instat with LU2, pressure is initialized to 0 with ZeroScalarFunction and velocity to 0 with LU_Velocity. By directly changing these initial values in cgstokes_inital.hh file, it appears pressure gets the same value of the first component of the velocity. Could you help me to understand ?
>> I checked in the previous version of pdelab and initial functions were given to interpolate with  NavierStokesDirichletFunctionAdapterFactory (deleted in the new one) and GridFunctionSpace did not have the same structure. I did not encounter this problem. Do you think I should reuse this deleted class ?
> 
> This is strange - I can change velocity and pressure initial values individually without a problem. How exactly did you set the different initial value?
> 
> Oh, and another question: What compiler are you using? We discovered a bug in PDELab that might bite you here if you are using an older compiler.
> 
> Best
> Steffen
> 
>> 
>> PDELab 2.0.0
>> Dune core module 2.3.1
>> 
>> Thanks in advance for your help,
>> 
>> Xavier KUHN
>> Software engineer at Eurogiciel
>> Third part for CEA Cadarache
>> 
>> _______________________________________________
>> dune-pdelab mailing list
>> dune-pdelab at dune-project.org
>> http://lists.dune-project.org/mailman/listinfo/dune-pdelab
> 
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20140725/f040ba01/attachment.sig>


More information about the dune-pdelab mailing list