• Individual/group updates
• Next Parthenon dev meeting
• Kokkos 4.6.00
• Non cart coords (status and need for Dxf<dir,face>())
• Next release
• Review non-WIP PRs

LR

• extend[s/ed] solvers to non-CC fields (nodal works already)
  • required some changes to ownership knowledge in packs
  • still some questions around (more general) boundary conditions
• added some updates to sparse packs
  • incl. interface update/separation/cleanup
  • also tied to boundary condition updates
  • optimization of pack caches (one per pack descriptor) Discussion:
• boundary conditions in corners (particularly, what are outflow conditions for corner cells -- for stencil that require info from corners, e.g., CTU)
• currently we fill over the full ranges, i.e., the last called boundary function covering a corner will set value
• we may want to update this in the future to be more specific, but probably not too problematic at the moment (because there's some sensible data in there right now)

PM

• workshop update
  • good response by invited speakers/code representatives
    • currently around 15 people (mostly developers)
    • should we extend "invitations" to (new) users?
      • could make downstream users/devs aware
      • PM will ask around and we revisit this next time
      • Everyone should send list of potential names to PM
  • We will use GH wiki for coordination
• put up new paper on archive on tensor train for ... -- there will be a talk at the workshop
  • Parthenon requires fields with a flexible vector index

BP

• fixing downstream bugs
• add provenance information on where parameters came from (command line, input file, params, defaults, ...)

AR

• trying to build Parthenon with C++20 (mostly just warnings)
• motivated by added some functionality for functors with many template parameters (concepts)

PG

• found/fixed multilevel bug (was downstream)
• MI300A <-> Kokkos 4.6
• enroll boundary conditions for variables?
  • not required as can be controlled by variables in packs

• PG still check with JM/JD offline
• For Dxf<dir,face>() interface:
  • do we need this?
  • do we need sth more general?
  • what's the impact on potentially caching coordinates? -> topic for workshop

Open PRs:

• [Breaking] Sparse Pack Updates #1237 <- LR will make make cosmic update, PM will do second review (should be contained in next release)
• Tridiagonal Solver #1234 PG will do second review (not blocking for release)
• Implicit diffusion example #1233 <- already has two reviews, PG will do a quick one (should be contained in next release)

• It's about due
• Any blocking PR? see above
• Then release

tentative next meeting 24 Apr

Workshop ideas:

• Params <-> parameter input <-> etc. (see provenance comment by BP above)
• Coordinate interfaces (and caching of coordindates)
• Cleanup of example directory
  • need to update/document examples
  • move some examples to regression tests
  • highlight features of each examples