2023-4-4 Meeting notes

Approve 2023/1/10 minutes.

@Scot Breitenfeld ​

Approved.

high-level editing tools for the documentation page

@Marc Poinot @Christopher Rumsey ​

Help needed to divide the remaining tasks:

https://github.com/CGNS/cgns.github.io/projects/1

The documentation goes into the source using Doxygen. This work uses the “Doxygen” branch. Examples are given for cg_open and cg_version. The aliases used for repeated variable descriptions are in src/Doxygen/Doxygen.config. We need volunteers to help with translating the remaining APIs to Doxygen.

• A link to the new website is provided on the main cgns web page: https://cgns.github.io/cgns-modern.github.io/index.html.

• All the MLL APIs have Doxygen entries, the cgio APIs need to be done next.

• An automatic generation of the website through GitHub action is done.

Converting CGNSTalk archive to Github Discussion Posts

@Scot Breitenfeld

Hope to be completed by the next meeting.

No updates

Coupled fluid-structure standard addition

@Scot Breitenfeld

Proposal to standardize fluid-structure simulations submitted, no updates.

Discussion Topic

@ZJ Wang

Short discussion on past AMR CPEX (0008) and whether AMR should be revisited as a feature-supported CGNS. Some groups may be already using CGNS with AMR, but not known for certain. The need for AMR support in CGNS is unknown.

@Vicky Moschou @Tobias Leicht @ZJ Wang to discuss progressing CPEX 0008.

• CPEX 0008 / Adaptive Mesh Refinement: Again, we are very much interested in the topic and actively working on and with adaptation using element subdivision, leading to hierarchical representations of the mesh. Where I am unsure is whether there actually is a need to store this structure itself, i.e. essentially provide information of how to get from one mesh to another via adaptation. Any resulting mesh with hanging nodes can be represented just as is (e.g. as polyhedral). While I think that it could be beneficial to explicitly express the hanging node / hanging face connectivity (in order to allow to still recognize the elements as (for example) hexes with hanging nodes rather than just any polyhedron) I don’t think it would be overly useful to have the actual hierarchical information available in an archival /exchange format like CGNS. I don’t know what it would be used for outside of a specific code. But I’m certainly still very much interested in the actual discussion. If this is supposed to be a separate small group discussion (my current understanding), please invite me over… Tobias

This work is a low priority for the majority of the past interested parties, this CPEX will be tabled

Status update on Accepted CPEX #0045

https://cgnsorg.atlassian.net/browse/CGNS-182

@Koen Hillewaert

The branch CPEX45_high_order has not been updated since Feb 25, 2020. An update from @Koen Hillewaert on the status is needed.

Regarding the two topics that I might want to say something on I have collected some input – feel free to use it (or not):

• CPEX 0045: We are still very much interested in this for our DG and DGSEM solution storage. We have been rather slow in adopting the proposed work in progress, though, and now that it has been quite a while since anyone touched things in CGNS I am a bit unsure whether it makes sense to push it now or wait for feedback on the status. We decided for the latter, so in the end I am waiting for Koen to provide feedback on the status of the branch and the way ahead. Please let me know if I am under the wrong impression that this was the agreed next step… Tobias

 There is currently no one actively working on this, we will keep the project on the meeting agenda but we don’t expect any updates.

Open Discussion about backward compatibility issues

@Scot Breitenfeld

Discussion about backward compatibility issues discovered recently.

Ansys, and others, has encountered backward compatibility issues with 4.3.0. All issues with backward compatibility will try to be addressed and fixed before the 4.4.0 release. All interested parties should either create an issue or contact the CGNS developers to ensure their issue has been addressed.

Discussion Topic

@Mickael PHILIT

Proposal to have Monitoring numerical probes description to the standard. Some groups may already use CGNS to do that, but there is a need to have a shared approach.

Status update on Accepted CPEX #0047 Proposed Data integration Points

@Mickael PHILIT

Working on the new website before creating a new development branch to prevent duplicating efforts.

Started prototyping with python and C and will push a branch on GitHub in the upcoming weeks. (The “C” branch is a slow work in progress)

No new updates. Still in progress.

prioritization, review and attribution of JIRA bugs/issues

@Tony Garratt @David Gutzwiller ​

Next release 4.4.0, Kanban:

Current status for Ansys:

No issues that are urgent.

@Scot Breitenfeld is working on updating the Multi-dataset APIs in CGNS to use the new array-based API and is testing CGNS with the new HDF5 subfiling feature.

@Scot Breitenfeld Can the APIs default to writing one dataset at a time if CGNS was not built with HDF5 1.14.0?

• Added question to the GitHub issue for investigation,

@Scot Breitenfeld Is completing the option of using subfiling for CGNS.

Perlmutter, write times:

65,536 ranks, benchmark_hdf5.c, 1 billion elements

Independent, single-shared-file: 553.4 s

Subfiling: 56.4 s

Feature is scheduled for the next release, currently in the subfiling branch.

​