Firedrake meeting 2022 08 18

Date and time 2022-08-18 12:00UTC (13:00BST 22:00AEST)

Action Items

Pick Chair and Minuter (RNH to pick).
ALL: (ongoing) triage the open issues and confirm if they are indeed still open (and perhaps provide labels)
JB: Look into updating the @parallel test marker (ongoing)
DH: talk to Jemma about Firedrake 2022 (ongoing)
JB: SIAM CSE23 Minisymposium/minisymposterium
~~JB: PETSc branch with GC fix (JB) + IO (KS)~~

Minutes

Present: CW SV JB DH KS RNH NB

Apologies:

DH Firedrake22

Firedrake 22 will be on 4-6th Jan 2023, Dartington Hall.

Jemma will be at Imperial 7-8th September.

Firedrake training will probably run next term.

DH + JB: SIAM CSE

Steady progress

DH: PETSc IO + GC branch

JB: This is done.

DH: There is an issue when writing to file. DH to update with details shortly.

DH: In the process managed to get the graphical debugger in VSCode to work with MPI.

NB: Interp - Problem with the current maths for adjoint ?

There seems to be a problem noticed with the current theory for the adjoint of Interp.

Context: Let $I(v_1, v_{0}^{*})$ be an Interp mapping from $V_1$ to $V_0$ with $v_{0}^{*} \in V_{0}^{*}$ (Coargument) and $v_{1} \in V_{1}$. Then the adjoint should map from $V_{0}^{*}$ to $V_{1}^{*}$.

RNH: There is an open issue about mis-annotated Hessians https://github.com/firedrakeproject/firedrake/issues/1980, perhaps a clue?

DH: Somewhere in py-adjoint, someone has done something naughty and needs to be sorted out. Problem is probably with the maths.

RNH: Currently very familiar with this code.

Action DH: Produce an MFE for this issue.