On 5/5/26 06:22, Praveen C wrote:
>
> I am debugging some issue with a code which uses periodicity through constraints.
>
> I saw the function
>
> make_consistent_in_parallel
>
>
https://dealii.org/current/doxygen/deal.II/
> classAffineConstraints.html#ab0217a83250614a473e29096d9a7f515 <https://
>
nam10.safelinks.protection.outlook.com/?
> url=https%3A%2F%2Fdealii.org%2Fcurrent%2Fdoxygen%2Fdeal.II%2FclassAffineConstraints.html%23ab0217a83250614a473e29096d9a7f515&data=05%7C02%7CWolfgang.Bangerth%
40colostate.edu%7Cf143e6b5c17d444a18d508deaaa97288%7Cafb58802ff7a4bb1ab21367ff2ecfc8b%7C0%7C0%7C639135841918140020%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=Y6Ox6ymuK%2BBqmW37GR3qiOko2MNuIM9Ts0FnXNCBeTk%3D&reserved=0>
>
> and it is only used in step-86. Should we use it always in a parallel code ?
Praveen:
yes, probably. If you only have hanging node constraints, I believe that the
result should already be consistent in parallel, but if you add constraints
manually, you probably want to call the function. It would certainly not hurt
to make it a habit to just *always* call the function when running in parallel
-- I would accept such a patch to the tutorial programs if you wanted to write
one.
Best
W.