Re: [Getfem-users] Boundary problems

[Renard Yves]

Mattia Assanelli <address@hidden> a écrit :

> Good morning everybody
> I succeded in solving my "generic elliptic" problem, but I still have
> some troubles analyzing data:
> - I noticed that on the boundary the value imposed by a Dirichlet
> condition ins't respected by the solution...

Well ! If you use a multiplier having a lower degree than the main  
unknow you can have a little difference since the Dirichlet condition  
is prescribed in a weak sense. But if the multiplier is string enough  
the Dirichlet condition will be exact.


> - I also tryed to calculate the flux through the contact...is there a
> simple alternative to manually integrate the gradient on the boundary
> surface?

To integrate the gradient on the boundary is not a good idea.  
Theoretically, you are not garanted that this value converges toward  
what you want. It is strongly better to integrate the multiplier  
ensuring the contact condition.


Yves.




----- Fin du message transféré -----