Skip to content

Add Mean Estimator checkpointing #1070

@garrettwrong

Description

@garrettwrong

It would be useful if our mean estimator periodically saved off the last mean volume to disk. If the estimate approaches our threshold but fails to converge to the target, the current code will just continue to iterate while killing process will lose the volume.

For larger volume sizes this can often be after a long wait. It is possible the volume is still useful at this point. I'm also not see at this time how the expected target is derived...

While I have not tested, I believe we should be able to use such a checkpoint as a starting point by converting into basis via the adjoint and using as b_coef. As part of this issue we can check this feature.

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions