Dear Martin, Yes, it works now. Could you please explain why it works in serial or with a small number of processes?
Best regards, Ce Qin 2016-05-23 15:27 GMT+08:00 Martin Kronbichler <[email protected]> : > Dear Ce, > > I faced a strange problem. With relatively few cells and non-conforming >> triangulation, the assembling process generates new nonzero locations in >> parallel with relatively large amount of processes . It works well in >> serial or with a small number of processes or with conforming >> triangulation. I wonder whether it is a bug or my fault. Can anybody >> provide some hints? >> > > In your code you use cm.distribute_local_to_global for writing the cell > results into the global matrix, which includes resolving the hanging node > constraints, whereas you do not include that list when resolving > constraints. You need to append the 'cm' as third argument in the > make_sparsity_pattern call (you can also set the fourth argument to 'false' > in order to skip creating matrix entries in the constrained rows). Can you > try this and report back in case it does not work? > > Best, > Martin > >> -- > 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. > -- 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.
