On 8/25/22 16:56, Raghunandan Pratoori wrote:
Thank you for your reply Prof. Bangerth. I believe I am already using
distributed computing in my code. But I am not sure it is true for the post
processing part. I use write_vtu and write_pvtu_record to write the output
files. Is that the best way to do it in distributed computing? Is there any
example where distributed computing is implemented in the post processing so
that I can compare my code and see if it can be improved?
I see now why you have so many entries in your solution vector :-)
I don't think I have anything to point to, but the general approach needs to
be that on every process you only allocate memory for each locally owned DoF
(i.e., locally owned vector entry), rather than for *every* DoF or vector
entry. That is the only way this can scale.
You can easily query how many locally owned DoFs there are on the current
process, using functions in class DoFHandler. You then probably have to
translate between global indices j and the how-many-th locally owned DoF index
j corresponds to. To this end, ask DoFHandler for the IndexSet corresponding
to the locally owned indices; class IndexSet has functions that can translate
in both directions.
I hope this helps!
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 on the web visit
https://groups.google.com/d/msgid/dealii/b72e7fd7-e70b-e6dd-81ce-96382bb7f557%40colostate.edu.
