Phoebus Resync / Planning Meeting - Fri. 13, Dec., 2024

[carlnotsagan]

Current development

• Working to update with new parthenon - a lot of refactoring/ particle features
• Working branch - no PR yet

Current issues/challenges

• Major issues solved with Kokkos 4.5 Cray machines - parthenon working with updated Kokkos
• Additional challenges with currently running on LANL machines
• No View of Views unless u need to - wrong steps given by Kokkos docs.
• Possible issues in tasking
• Last big change left at API level - make sure particles work with AMR only with SMR now
• Riemann solver needs rewritten - can't convert it to sparse packs
• Testing infrastructure limited by Github
• Good to have access to other hardware - external server - possible at one of the IHE? Not arizona for now..
• 3D CCSN Lightbulb challenges - poor performance, inconsistant results?
• MC particles infrastructure needs work
• Many other physics pieces not being developed
• Docs incomplete - need more details covering state of the code.
• Long term challenge: create an explicit feature policy. Example: someone implements a feature that needs maintaining in the future and interfaces with many pieces of the code.

Possible action items

• Ben P. and Brandon planing a meeting to work on phoebus up-to-date with current parthenon using lessons learned from kharma.
• Detailed developers guide in the Docs but most via Github Issues possibly started from Github Discussions.
• Developers Guide could include "lessons learned" or design philosophy/science-driven code choices and their discussions (Singularity-EOS a good starting point for this again).
• More physics tests for current capabilities - how do we know the code works???
• Detailed Tutorial for adding/contributing to the phoebus docs either directly or from a Discussion or merged PR/Issue.
• Detail new options for development / collaboration on the README
• CI set for parthenon for MHD etc, could use Philips machine with HPC lab or migrate to HPC LANL namespace
• Miah opening phoebus slack workspace to public - might need to do on the web browser. Fine for now given small project size.
• Send a when2meet end of the year/first week of Jan. to schedule recurring meeting - acknowledging that core developers will meet less often. Need: clear notes at each meeting to keep those that cannot meet in sync. Scribe?

General Comments

• Parthenon refactoring not a persistent problem - growing pain - approaching state of maturity. Good change. "We know what we are doing now."
• Need for additional consistent development for Phoebus.
• Anyone with one or a few commits since the first instrument paper should/could be a co-author on possilbe second instrument paper.
• Kharma documentation been helpful for many things - strong support for improving the docs!
• Brief overview of Phoebus proposals submitted - AAG, CSSI - would support grad students to help with docs, infrastructure, and development, and/or science topics in CCSNe.
• Focus needs to be placed on development challenges.
• Less telephone and additional meetings, steps to stay synched.
• Strongly discourage Github Forks.
• All bug reports / features should continue to stay in github Issues.
• Goal/Science-driven development is critical.

Meeting outcomes

• Started Github Discussions for: open discussions on questions, development, challenges then should be moved to a Github Issues or Docs. Goal: reduce barriers to discussing challenges openly

[Yurlungur]

Thanks for the summary @carlnotsagan !