Speed up AtomicLocal forces #1024
Merged
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
|
Nice! |
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Outdated suggestions cannot be applied.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 greatly increases the efficiency of forces calculations when using UPF pseudos.
Up to now, that vast majority of the time was spent calculating the form factors. It turn out that this issue was already solved when calculating the
AtomicLocalpotential, by pre-computing the form factors for each unique |G+q|. The same approach is now applied to the forces.While the form factors are not cached, as suggested in #1009, this is already a huge step forward. Cost of calculating the forces goes from ~30-40% of the total, to <5% (on a handful of small tests).