Python PETSc Mat for ReducedFunctional #213
Open
+449
−123
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
JHopeCollins
changed the base branch from
JHopeCollins/tlm
to
dham/abstract_reduced_functional
JHopeCollins
changed the base branch from
dham/abstract_reduced_functional
to
master
jrmaddison
reviewed
Oct 14, 2025
jrmaddison
reviewed
Oct 14, 2025
Co-authored-by: James R. Maddison <[email protected]>
Co-authored-by: James R. Maddison <[email protected]>
…pyadjoint into JHopeCollins/tao-rf-mats
dham
requested changes
Oct 21, 2025
|
Review comments from Firedrake meeting addressed. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR introduces a
ReducedFunctionalMat, which is a Python type PETSc.Mat which applies the action of aReducedFunctional.By default it applies the action of the Hessian, which is needed for TAO's Newton optimiser.
If
Jhat.functionalis not a scalar then it can also apply the action of the TLM or adjoint models. Examples of when these would be needed include 1) constructing saddle point formulations of an optimisation problem, 2) constructing preconditioners for the Hessian with a certain structure, 3) calculating the eigendecomposition of a reduced functional (firedrakeproject/firedrake#4529).Firedrake tests: firedrakeproject/firedrake#4669