Example usage of FieldManager in an NGP Kernel #1182
Draft
Conversation
This file contains 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
psakievich
requested review from
overfelt,
rcknaus,
marchdf,
alanw0 and
jrood-nrel
This PR demonstrates how a Kernel can be converted to use a FieldManager. The main goal is to move the field registration away from higher level abstractions like realm and equationsystem, to lower ones (algs and kernels). This will essentially eliminate the need for `RegisterNodalFields` type calls, and will eliminate a range of bugs that can come in from not registering fields in the right place. Having kernels and algs register the fields they need will also reduce the amount of fixturing needed for unit tests since constructing the object will register the fields for you. You can then finalize the mesh and populate the fields with initial values. (Not demonstrated in this PR).
alanw0
approved these changes
May 4, 2023
psakievich
commented
May 4, 2023
src/node_kernels/TKESSTNodeKernel.C
Outdated
| TKESSTNodeKernel::TKESSTNodeKernel( | ||
| const stk::mesh::MetaData& meta, | ||
| const FieldManager& manager, | ||
| stk::mesh::Part* part) |
There was a problem hiding this comment.
Actually looking at this closer, I'm not so sure we need a part vec. I think the field registry process will just add more parts each time it is called after a field is registered since multiple identical registrations are a no-op in stk. It would be good to test it though.
|
Making some progress on this. I really want to get this swarmable in the next few weeks and work as a team to convert the code to use the fieldManager and SmartFields as much as possible. It would be a good way for a lot of newer people to see a lot of nalu-wind's structure. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 demonstrates how a Kernel can be converted to use a FieldManager.
The main goal is to move the field registration away from higher level abstractions like realm and equationsystem, to lower ones (algs and kernels).
This will essentially eliminate the need for
RegisterNodalFieldstype calls, and will eliminate a range of bugs that can come in from not registering fields in the right place. Having kernels and algs register the fields they need will also reduce the amount of fixturing needed for unit tests since constructing the object will register the fields for you. You can then finalize the mesh and populate the fields with initial values. (Not demonstrated in this PR).As a side bonus we don't need to store class members for field ordinals AND field pointers on kernels any more either.