Skip to content
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

[FIX] tighten the forward tests according to observed error wrt array size #67

Closed
wants to merge 4 commits into from

Conversation

eickenberg
Copy link
Collaborator

image

plotted errors as a function of N and updated the tests to make them as tight as possible

@eickenberg
Copy link
Collaborator Author

also removed the float32 version of the test for now. the error will definitely be larger there but there is at least currently no reason to believe that float32 will not work if float64 does

assert l_2_error < 7.5e-4 * N ** 1.1
assert l_1_error < 5e-4 * N ** 1.6


@pytest.mark.parametrize("targets", cases)
def test_1d_t2_forward_CPU(targets: torch.Tensor):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think you missed type2 in this file

@WardBrian
Copy link
Collaborator

also removed the float32 version of the test for now. the error will definitely be larger there but there is at least currently no reason to believe that float32 will not work if float64 does

Agreed. There's a narrow line between testing the pytorch wrapper and testing FINUFFT itself

@WardBrian WardBrian linked an issue Oct 6, 2023 that may be closed by this pull request
@WardBrian
Copy link
Collaborator

This was included in #64

@WardBrian WardBrian closed this Oct 6, 2023
@WardBrian WardBrian deleted the mike-fix-tests-against-fft branch October 6, 2023 16:50
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

BUG fft correctness checks are vacuous
2 participants