Skip to content

1D np.linalg.norm not preserving input dtype? #10364

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.

Sign up for GitHub

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

Jump to bottom
Closed
kmaehashi opened this issue Jan 11, 2018 · 2 comments · Fixed by #10390
Closed

1D np.linalg.norm not preserving input dtype? #10364

kmaehashi opened this issue Jan 11, 2018 · 2 comments · Fixed by #10390
Assignees
@eric-wieser
Labels
00 - Bug

Comments

@kmaehashi
Copy link
Member

I saw the v1.14.0 release note saying:

1D np.linalg.norm preserves float input types, even for arbitrary orders

However, when I try the code on the release note, it does not behave as it says.

>>> import numpy as np
>>> np.__version__
'1.14.0'
>>> f32 = np.float32([1, 2])
>>> np.linalg.norm(f32, 2.0).dtype
dtype('float32')
>>> np.linalg.norm(f32, 2.0001).dtype
dtype('float64')  # it should be float32 on numpy 1.14?
@charris charris added this to the 1.14.1 release milestone Jan 11, 2018
@eric-wieser
Copy link
Member

eric-wieser commented Jan 11, 2018 •
edited
Loading

My fault in #9522 (351fe97). I think it does preserve longdouble types now, but I wasn't able to test that.

Maybe simplest to revert that part of the release note?

@eric-wieser
Copy link
Member

eric-wieser commented Jan 11, 2018 •
edited
Loading

Culprit is the 1.0 on this line upcasting: 351fe97#diff-6858bbac5c5c59bdddf245c69d6a547cR2206

Replacing it with 1 or np.reciprocal should solve the problem.

@eric-wieser eric-wieser self-assigned this Jan 11, 2018
@eric-wieser eric-wieser mentioned this issue Jan 11, 2018
Merged
@kmaehashi kmaehashi mentioned this issue Jan 11, 2018
Merged
eric-wieser added a commit to eric-wieser/numpy that referenced this issue Jan 12, 2018
@eric-wieser
BUG: Avoid unintentional promotion to float, and do ops in place fo…
04896b0 
…r speed

Fixes numpygh-10364, partly by adjusting the promise in the release notes
@eric-wieser eric-wieser mentioned this issue Jan 12, 2018
Merged
@charris charris removed this from the 1.14.1 release milestone Jan 27, 2018
hanjohn pushed a commit to hanjohn/numpy that referenced this issue Feb 15, 2018
@eric-wieser @hanjohn
BUG: Avoid unintentional promotion to float, and do ops in place fo…
d41c84e 
…r speed

Fixes numpygh-10364, partly by adjusting the promise in the release notes
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@eric-wieser eric-wieser

Labels
00 - Bug
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

MAINT: Adjust type promotion in linalg.norm eric-wieser/numpy
3 participants
@charris @eric-wieser @kmaehashi