Dear Wolfgang,
It seems this "solution.h5" instead of f_output was a misspelling mistake,
I have right f_output in my code. I can see that the problem here is that
there are processes without cells, which may be a bug of my implementation
as I would like to use all the processes. How does could that happen?
Thank you again
El miércoles, 16 de febrero de 2022 a las 18:25:40 UTC+1, Wolfgang Bangerth
escribió:
>
> > Thank you for the suggestions. I updated the code as you mentioned. I do
> not
> > fully understand what you mentioned about that I am not writing into the
> > f_output stream.
>
> In the code snippet you showed, you are always writing to "solution.h5",
> not
> to the f_output stream.
>
>
> > Is data_out.write_hdf5_parallel() not writing the data?
> > However, I still have the error when running more than 1 MPI rank. If it
> is a
> > bug I can not do much about it.
>
> It is meant to do exactly that, and we use it all the time. Just not for
> the
> corner case of a process that has no cells. But you can use any number of
> other formats to output your solutions. In parallel, we generally use the
> VTU
> file format. Take a look at any of the parallel programs, starting with
> step-40, step-32, etc, and see how they create output.
>
>
> > I tried also to save the local_relevant_solution vector into a
> FullMatrix that
> > stores each dof and frequency and then print it in the end, but the data
> is
> > wrong also for more than 1 MPI rank:
> >
> >
> > for(int freq_iter=0 ... ) ///solver loop, /
> > {
> > // solve...
> > testvec.operator=(locally_relevant_solution); // testvec is a vector of
> size dof.
> >
> > for ( int j= 0; j < dof_handler.n_dofs(); ++j)
> > testsol(freq_iter, j)+=testvec(j); // testsol is a FullMatrix of size
> > number_frequencies x dof
> >
> > }
> > if(Utilities::MPI::this_mpi_process (mpi_communicator)==0){
> > testsol.print_formatted(mydomain,12,true,0,"0"); // my domain is ofstream
> >
> >
> > Is there anything wrong in this approach of exporting the results?
>
> I don't know what "the data is wrong" means, but this process could in
> principle work. Except you end up with just vectors without knowing the
> physical location at which each degree of freedom is located. You don't
> know
> how to visualize these vectors any more if you don't also output node
> data.
> This is of course exactly what DataOut is there for.
>
> Best
> W.
>
>
> --
> ------------------------------------------------------------------------
> Wolfgang Bangerth email: bang...@colostate.edu
> 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 dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/dealii/3d49f9f0-1885-4bcd-b25c-bf331ef64da8n%40googlegroups.com.
