Hi Asmorom,

do your results also differ if you run your problem again on the same 
number of processes, or do you get the same result? If they still differ, 
there is a nice explanation on why this happens in the FAQ on PETSc:
https://petsc.org/release/faq/#how-come-when-i-run-the-same-program-on-the-same-number-of-processes-i-get-a-different-answer

You could also try to use Trilinos instead and see if you're more satisfied 
with the results.

If you think there is a bug in your code, you can try to reduce the size of 
your code to get a smaller, non-working example. It might be easier to find 
it on a smaller scale.

Marc

On Monday, August 1, 2022 at 6:29:17 AM UTC-4 Asmorom Kibrom wrote:

> Hi all,
>
> For the purpose of gaining more experience on the dealii framework I have 
> implemented the stationary (no time derivatives) piezoelectric PDEs as a 
> test case. After successful runs without parallelization, I tried to modify 
> the code using the PETsc wrappers for parallelization. I have attached a 
> tar archive including the project folder (cc and param file).
>
> However, results now seem to be different from what I had on single core 
> executions. The issue doesn't occur for the electric potential DOF. There 
> the results look identical. But the displacement vector part of the 
> solution varies with the n.o. processes used (see fig. below).
>
> I am running dealii 9.4.0 on Ubuntu 20.04LTS under WSL2 (Kernel ver 
> 5.10.16) Windows 10. deallii was compiled using "cmake 
> -DCMAKE_CXX_FLAGS="-march=native" -DWITH_MPI=ON 
> -DCMAKE_INSTALL_PREFIX=/opt/dealii-9.4.0". The "summary.log" and 
> "CMakeError.log" are also attached. The modell consist of a simple 
> hyperrectangle in 3D.
>
> Here ist the PDEs weak form I have used
>
> [image: var_results_problem_MPI_equations.png]
> [image: var_results_problem_MPI.png]
>
> Best Regards,
> Asmorom
>

-- 
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 on the web visit 
https://groups.google.com/d/msgid/dealii/5d6dd1be-9ab3-44ca-92a2-86205e2af2d0n%40googlegroups.com.

Reply via email to