feat: add argmin implementation using CCCL#3811
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files
🚀 New features to boost your workflow:
|
|
Interestingly, |
|
The documentation preview is ready to be viewed at http://preview.awkward-array.org.s3-website.us-east-1.amazonaws.com/PR3811 |
ianna
left a comment
There was a problem hiding this comment.
@maxymnaumchyk - Great! Thanks for implementing it. There was an ak.prod failure on the complex numbers in a block boundary test. It's unrelated to this PR and it was gone after re-running the tests - so I'm merging it. Thanks.
The difference is that the field x is a listoffsetarray and y a numpyarray. Virtual arrays rarely have anything to do with it as the buffers just get loaded before any kernels run. which makes me think that probably the error happened on a fully materialized jagged array. |
|
@ikrommyd you mean to test If so, then everything works on my machine~ |
I would personally just extract the test in a small snippet and not use any fixtures and pytest. I'd first try with a fully materialized array and then with virtual arrays. |
Thanks, @ikrommyd ! Your help is greatly appreciated. |
No description provided.