Firedrake meeting 2025 03 26 - firedrakeproject/firedrake GitHub Wiki

Date and time 2025-03-26 1400 UTC

Action Items

  1. Pick Chair and Minuter (LC not here, someone else pick)
  2. ALL: (ongoing) triage the open issues and confirm if they are indeed still open (and perhaps provide labels)
  3. ALL: do things with SV's branches
  4. DH: Email to Andreas to have 2 (+ others!!!) loopy PRs merged TODO: FIND OUT WHICH PRS THESE ARE
  5. DH: Get Firedrake a docker open source account (link here)
  6. DH: Talk to GregVernon about PR#2116.
  7. JB: Enable merge queues (minutes)
  8. LC: Try to merge RNH' PR: Movable VOM
  9. PB: Profile and speed up some tests (minutes, minutes)
  10. CW: More testing configurations (minutes)

Agenda

Present: DH, PB, IM, JHC, KS

Apologies: LC, AO, CW (maybe)

CW: What happens to Zenodo when we move to releases?

  • git information not available for packages, instead we have version numbers
  • if someone is using a version of Firedrake then we already have a GitHub release that they are using
  • The simplest case is to gather the various GitHub releases together and put them in a Zenodo release
    • If so might not need our own forks of PETSc etc
  • What if someone is working from a branch? Do they have to make their own release?

Minutes:

  • Zenodo all release components on release, create citation.
  • Release further components (on branches) adhoc when needed.

CW: Use upstream PETSc: #4137

Minutes:

  • Merged already
  • We should probably instruct PETSc to install SLEPc

Proposed roadmap:

  1. Merge this (breaks firedrake-zenodo)
  2. Update to latest release of PETSc (v3.23.0) very close to release
  3. Firedrake releases (and fix firedrake-zenodo). This will require a little thought about what branches of PETSc we track

LinearSolver initial guess/convergence error: #4144

Minutes:

  • Intuitive behaviour - use x that is passed, allow users to pass zero if they want.

  • This is different from the default behaviour in KSP that uses zeros unless you specify an initial guess.

  • Convergence error bug. Return the most recent solution even if convergence not achieved?

    • Is this useful? good for debugging
    • Matches existing behaviour in VariationalSolver/NonlinearSolver

  • Goal: be consistent across our solvers.

  • Resolution: remove behaviour around x, retain convergence behaviour.

Merge PRs

Note that PRs put in this section should either be trivial or already have been reviewed. Discussion-worthy PRs should be separate agenda items.

  • CW: #4138 - Merged

  • PB: BQ FIAT - Review docstring policies and update

  • PB: BQ Firedrake - Ready when the FIAT one is

  • PB: #139 mimetic just review

    • Some discussion on the naming "mimetic". Not specific enough maybe? Perhaps "mimetic spectral".

Date of next meeting

1600 UTC 2025-04-02