Firedrake meeting 2026 05 26

Date and time 2026-05-26 1500 UTC+1

Action Items

1. Pick Chair and Minuter (CW to 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: Talk to GregVernon about PR#2116.
6. JHC: New manual release
7. ALL: Ready for PETSc/Firedrake 26?

Minutes

Present: CW (minuter), LC, AG, DH, DS, PB, IM

Apologies:

PETSc/Firedrake

• JB has sorted the drinks reception
• Passes are printing and CW is sorting
• Recording talks is an option, DH can bring tech
• Should send out joining instructions email - come to Thames suite and get name badges - ACTION POINT CW

Tutorials

• PB: Still needs to make demo into tutorial - will extend a submesh demo
• DH: Make sure to focus on solver options. And marketing. We want to show some solver composition where we get PETSc <-> Firedrake <-> PETSc ... (e.g. Riesz map preconditioner). "Go back into FE land in the middle of your solver".
• DH: Irksome shouldn't be the focus of this demo.
• CW: We want PETSc users to come away from this learning about the fact that we have a whole ecosystem of Python PCs on top of PETSc.
• DH: A Riesz map preconditioner is a really good example "precondition this solve with another finite element problem".
• PB: We have a Stokes demo that does something like this, changes both operators.
• DH: Steal some ideas from existing demos on changing solver options.
• DH: Remember that not everyone there is really familiar with Firedrake. Some amount of walkthrough is needed.
• DH: The existing solver composition tutorial has a lot of great material, but should have additional exercises.

DH: Complex discussion on Slack earlier this week

• DH: Cofunction v+function f, v(f) does the conjugate of f with not conjugate of v.
• DH: Need a convention about which we are linear in and which we are conjugate linear in. We adopt the mathematics convention of conjugating the second argument in inner, and second argument is the test argument. This has implications for the conj-ness of arguments taken by a bilinear form.
• PB: Nonlinear Schr equation - you treat z and zbar as independent variables. Like a mixed system of 2 fields. The result is out by a factor of 2.
• DH: Complex numbers as an isomorphism of R2 gives us 4 partial derivatives. Vs complex-valued scalar problem where the derivatives have 2 numbers. Derivatives have to be complex linear - multiplying input by i has to multiply result by i.
• DH: The 2 value approach sounds reasonable. It could be written as a UFL expression visitor. That would be the first step.

LC: #5097 nested interpolate

Assembly does not work for nested interpolates

• DH: Want a dependency graph of primitive assembles, effectively a stack of parloops

Team catch up (if time)

Merge PRs

Date of next meeting

In 2 weeks because of the PETSc/Firedrake conference

1600 UTC+0100 2026-06-09