Closed ollyfutur closed 8 months ago
Thank you @ollyfutur, this seems excellent work!
To me this seems immediately mergeable, but let's see what @wildromi or @imacocco say since they are the people who used periodic distances the most in the past.
All modified and coverable lines are covered by tests :white_check_mark:
Comparison is base (
d82322b
) 80.29% compared to head (492fa30
) 80.22%.
:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Everything looks fine, the tests also run smoothly. It would be great also to fix the linting, then we can merge straightforwardly!! Many thanks for the correction!
Just fixed the linting, thanks for maintaining this great package :)
Thank you @ollyfutur for entering the team of contributors! 🙂 Since tests pass and @imacocco also checked the code, I will soon rebase and merge this PR 👍🏼
@ollyfutur if it's ok for you I will add you to the list of contributors here
Alright thank you very much @AldoGl !
Thank you Olivier @ollyfutur for spotting this bug!
Proposed changes
Correction of a bug in
utils.compute_cross_nn_distances
. When using periodic conditions the function computed the self NN distances of datasetX
withcompute_NN_PBC
instead of cross NN distances betweenX_new
andX
. This affected interpolated density estimation.Types of changes
The cross NN distance is directly computed using
cKDTree
instead of callingcompute_NN_PBC
. I have added two tests intest_distances_utils
,test_cross_nn_distances
andtest_cross_nn_distances_periodic
.