Firedrake meeting 2022 05 26

Date and time 2022-05-26 12:00UTC (13:00BST 22:00AEST)

Action Items

1. Pick Chair and Minuter (SV to pick).
2. ALL: (ongoing) triage the open issues and confirm if they are indeed still open (and perhaps provide labels)
3. KS: Test PETSc main against current Firedrake
4. JB: Look into updating the @parallel test marker (ongoing)
5. DH: talk to Jemma about Firedrake 2022

Agenda

Present: DH, RNH, CW (minuter), PK, KS, JB, SV

Apologies:

https://github.com/firedrakeproject/firedrake/issues/2452

• KS: Can use a PetscSection to save DMSwarm points.
• DH: This will work if you remove vertices from ghost cells.
• DH: Firedrake uses its own geometry rather than PETSc's. When loading from disk you only need to store topological information and then you can reconstruct the geometric/numbering information afterwards.

RNH: Updated VertexOnlyMesh API

See discussion #2450 and specifically this comment on API proposal

• DH: Remove point_migration keyword argument as this is premature optimisation.
• DH: redundant=True should also be strongly encouraged since, if you have so many points as to need it, the locate-cell functionality will kill performance.

SV: block optimisation in Slate

merge fix BOP https://github.com/firedrakeproject/firedrake/pull/2446

• KS: To take a look at the code.
• SV: Needs to rerun performance measurements.

DH: Stringifying UFL is not nice

• DH: str(expr) is really hard to read. Now we're on Python 3 we can produce unicode (e.g. integral signs).
• DH: to_latex is not really sufficient.

JB: numpy and scipy switching to Mazon for building

• JB: Also scipy needed for tests to pass.
• DH: Best to default install scipy.

Merge PRs

Date of next meetings

No meeting next week due to bank holiday (UK) Next meeting: 2022-06-16 (JB, KS, SV at ECCOMAS the week before)