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Outdated suggestions cannot be applied.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 will add the new assemble function.
On my tests, the current fitting methods in ACEpotentials will get most of speed increases the new assemble brings in. But the AtomsBase framework should be a little bit faster due to extra threading and the evaluate_d improvement.
I removed the dependencies from the old assemble (ParallelDataTransfer and SharedArrays).
I left progress meter in place, even though it might not have use in most cases now.
I added kwargs option to assemble to transmit extra options to calculators. If you don't give any kwargs, the old use cases should work. But ideally in the future, when you implement
feature_matrix
,target_vector
orweight_vector
you should addkwargs...
to the function call. There is no need to use kwargs to anywhere, only catch them.