Dear Bruno, Thanks a lot for your advice.
I tested it. The code runing in parellel has exactly the same norm of stiffness mtx but different norm of RHS residual compared to running in serial. It is really frastrating that increasing number of cores can make a difference... Best, Junxiang. On Friday, February 7, 2025 at 10:42:04 PM UTC+8 bruno.t...@gmail.com wrote: > Hello Junxiang, > > It's very hard to know what the issue could be. What you could do is print > some information about your matrix and right hand side (for instance, > different norms) and check that they are the same in serial and in > parallel. By comparing the serial and parallel simulations at different > points in your code, you should be able to find out why there is a > difference. > > Best, > > Bruno > > On Friday, February 7, 2025 at 4:51:50 AM UTC-5 wangj...@gmail.com wrote: > >> Dear all, >> >> I am modelling multyphysics problem in a paralle way. >> >> The code runs perfectly in serial or small number of cores while fails to >> converge once I run it on more number of cores say >8. >> >> What could cause this problem? >> >> Thanks a lot. >> >> Junxiang. >> > -- 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 dealii+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/dealii/e59ec875-ce0b-41ed-9f68-dcd75780311cn%40googlegroups.com.
