-
Notifications
You must be signed in to change notification settings - Fork 44
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Linalg stacking tests should not test exact equality #44
Comments
The idea with the stacking tests is that a function that operates on stacks of matrices should be the same on a stacked array as if you do it on the individual arrays independently. Is there any reason this shouldn't hold? The computations are independent of one another for different matrices. To be fair the spec doesn't really specify this. Maybe it should clarify whether it should be the same or just "close". |
Sorry @asmeurer I missed your reply. What I meant is that the comparison of actual & expected outputs like this
should return
|
cc: @kmaehashi for vis |
We will eventually change this to be a float-point approximate check rather than exact equality. For now, this is not updated. See data-apis#44
We will eventually change this to be a float-point approximate check rather than exact equality. For now, this is not updated. See data-apis/array-api-tests#44
The following two failures show that the exact equality test in
_test_stacks
should be relaxed, as the results agree within some tolerance that's both hardware and software dependent:The text was updated successfully, but these errors were encountered: