Dear Timo, I think that I am too late here.
My problem is similar with Francesco's. But I want to run it in parallel. I would like to know whether the order of DoFs remains the same after subdomain_wise and component_wise reordering in the two DoFHandler. And how to get the same parallel distribution since DoFTools::get_subdomain_association running in a random way? Best regards, Ce Qin 2014-02-23 23:43 GMT+08:00 Francesco Cattoglio < [email protected]>: > Thank you very much for confirmation, Timo! > > Have a nice Sunday, > Francesco > > Il giorno domenica 23 febbraio 2014 15:09:22 UTC+1, Timo Heister ha > scritto: >> >> Hey Francesco, >> >> yes, ordering and reordering is deterministic. >> >> >> On Sat, Feb 22, 2014 at 6:09 PM, Francesco Cattoglio >> <[email protected]> wrote: >> > Hi everyone, >> > this might sound a bit strange, and I know I'm doing something in a >> hack-ish >> > way. >> > I am building a block system (Darcy equation) by using a >> > FESystem(FE_RaviartThomas, FE_DGQ); by doing this I get a block matrix >> that >> > looks like this >> > A B >> > B' 0 >> > >> > I want to precondition with something like >> > P 0 >> > 0 1 >> > (where 1 is the identity matrix. Since I only need P, I would like to >> only >> > use a FE_RaviartThomas element when assembling the preconditioner, >> instead >> > of a full FESystem. >> > >> > Right now I'm doing the following: >> > -intialize one Triangulation >> > -initialize the DoFHandler of the FESystem of the main problem, then >> reorder >> > it component_wise >> > -initialize the DoFHandler of the FE_RaviartThomas using the SAME >> > triangulation, then reorder it component_wise >> > >> > By doing this my 2 DoFHandlers now have the same numbering for the >> degrees >> > of freedom on the velocity components. This means that the row "i" will >> > refer to the same dof, both for A and P matrices. >> > What I would like to know if this happens by accident or it is somehow >> > granted. E.G. if initializing a DoFHandler always returns the same >> ordering >> > on a given triangulation, and the sorting algorithm used when >> reordering is >> > stable, I think it would be safe to assume that the two DoFHandler end >> up >> > numbering the same dof with the same number. >> > >> > If doing this is not safe at all, is there any way to "extract" the >> > DoFHandler from my FESystem, recovering only the "FE_RaviartThomas" >> > DoFHandler? >> > >> > -- >> > The deal.II project is located at http://www.dealii.org/ >> > For mailing list/forum options, see >> > https://groups.google.com/d/forum/dealii?hl=en >> > --- >> > You received this message because you are subscribed to the Google >> Groups >> > "deal.II User Group" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an >> > email to [email protected]. >> > For more options, visit https://groups.google.com/groups/opt_out. >> >> >> >> -- >> Timo Heister >> http://www.math.clemson.edu/~heister/ >> > -- > The deal.II project is located at http://www.dealii.org/ > For mailing list/forum options, see > https://groups.google.com/d/forum/dealii?hl=en > --- > You received this message because you are subscribed to the Google Groups > "deal.II User Group" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/groups/opt_out. > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
