Hello Mikael, > We found the issue appears when two cells sharing a vertex and that are not part of the same subdomain have an FE degree difference of more than two or when the first one has a p-refinement and an h-refinement superior than a cell sharing its vertex. > We found the issue thanks to Peter Munch. When setting up our problem, we can use the function make_consistent_in_parallel() with our constraints after the line constraints.close().
This indeed sounds like a constraints issue. You can verify whether your constraints are consistent among all processes with the function is_consistent_in_parallel(). So far we encountered these types of inconsistencies only on 3d domains. You can correct them with make_consistent_in_parallel(), as you already found out. > If the Deal.II team think it's appropriate, I could submit a small PR next week to add this line and a quick explanation to the step-75 tutorial so others don't run into the same issue in the future. It is best practice to ensure that constraints are consistent. I agree that we should add this to step-75. Feel free to write a patch for it! (Please mention that this is technically only necessary for 3d problems.) > I was wondering if functions such as limit_p_level_difference() allow for cases where a neighbouring cell sharing a vertex has a polynomial order difference of more than 2? The function limit_p_level_difference() only compares the polynomial degrees of finite elements that are neighbors over faces. Thus over edges, the difference of polynomial degrees can be twice as high as the prescribed max_difference. Thanks, Marc On Friday, April 25, 2025 at 5:39:21 PM UTC+2 Wolfgang Bangerth wrote: > On 4/25/25 09:32, Mikael Vaillant wrote: > > > > We found the issue thanks to Peter Munch. When setting up our problem, > we can > > use the function make_consistent_in_parallel() with our constraints > after the > > line constraints.close(). If the Deal.II team think it's appropriate, I > could > > submit a small PR next week to add this line and a quick explanation to > the > > step-75 tutorial so others don't run into the same issue in the future. > > Yes, please. > Best > W. > > -- > ------------------------------------------------------------------------ > Wolfgang Bangerth email: [email protected] > www: http://www.math.colostate.edu/~bangerth/ > > > -- 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]. To view this discussion visit https://groups.google.com/d/msgid/dealii/609b95fb-03a3-420b-8c97-ce2cf0fb743en%40googlegroups.com.
