Agglomeration example#233
Conversation
|
|
||
| CMAKE_MINIMUM_REQUIRED(VERSION 3.13.4) | ||
|
|
||
| FIND_PACKAGE(deal.II 9.7.0 |
There was a problem hiding this comment.
Thank you for your contribution!
Can you lower the requirement to deal.II 9.5.0? We at some point established that code gallery programs should compile with 9.5.0 and younger.
If your code does not compile with the older versions, you can work around it like in 42e8901.
There was a problem hiding this comment.
@marcfehling Do we need this requirement? Could we allow programs to just require newer versions (or in fact change all programs to just build for, say, 9.6)?
There was a problem hiding this comment.
Could we allow programs to just require newer versions?
Sure, we can drop the requirement for deal.II 9.5. We didn't bump the version since summer 2023 #155.
Yet, this program still needs to compile with deal.II 9.6.
77cb7d2 to
0f481de
Compare
f3dbd02 to
0e25728
Compare
|
@fdrmrc You marked this program as a work in progress. Does this still apply, or is it ready to merge? |
|
@bangerth Yes, this should be ready to go now. The latest commits by @xinpinggui have added extensive documentation. |
bangerth
left a comment
There was a problem hiding this comment.
Nice program! I did not look through everything, but I think it's a cool example of what one can do!
I really have only one comment, namely whether we could perhaps take the coverage stuff out of the cmake file? It is specific to GCC and/or clang, and will not work for the Windows compiler. I would like to see whether we can keep these programs available for all platforms. If you agree, either remove these lines or tell me and I can do it myself!
| -fprofile-update=atomic | ||
| -ftest-coverage | ||
| ) | ||
| target_link_options(${TARGET} PRIVATE "--coverage") |
There was a problem hiding this comment.
Same here. I wonder whether there is a generic way to create coverage through CMake? I looked at https://danielsieger.com/blog/2024/08/03/code-coverage-with-cmake.html and that doesn't suggest anything that's not compiler-specific. I think for the purposes of a code gallery program, perhaps just remove all of the coverage stuff altogether.
|
The program is now here: https://dealii.org/developer/doxygen/deal.II/code_gallery_agglomeration_poisson.html Most of it looks good, but the images are not shown inlined into the readme text. Let me know if you'd like me to look into it, or if you see right away what the issue is. |
|
Sorry for the late reply, and thank you for pointing this out. I have now looked into the issue. The inline images were written using HTML tags such as
which render well on GitHub, but do not seem to be displayed properly on the deal.II code gallery page. I have replaced them with standard Markdown image syntax, for example
I also noticed a few related formatting issues in the code gallery entry, including the rendering of some comments in |
|
Thanks again for pointing this out. I have now opened the follow-up pull request to address these image display and related formatting issues: PR #244. |
This PR presents a DG solver for the Poisson problem on general polytopal meshes generated through mesh agglomeration. Starting from a fine mesh, we construct a coarser agglomerated mesh on which the DG discretization is performed.
This set of files have been extracted from the Polydeal library, a deal.II-based framework for working with agglomerated grids.
While the overall structure of
poisson.cclargely follows that of a standard Poisson solver in deal.II, differences related to the usage of agglomerated meshes still need to be commented, so the PR remains a draft for now.