You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
> @matbryan52 do you have an idea about the test failure? It seems that we are missing the tolerance by just a tiny bit:
E Not equal to tolerance rtol=1e-07, atol=0.5
E
E Mismatched elements: 1 / 2 (50%)
E Max absolute difference: 0.5083443
E Max relative difference: 0.15369503
E x: array([ 0.832502, 29.035461], dtype=float32)
E y: array([ 0.98369 , 28.527117]
Is something going on, or do we need to tweak the tolerance?
(note that it's not related to numpy 2.0, as hdbscan installs a numpy version <2)
I have a feeling it's a tolerance issue, but it must be a rare one. The large tolerance is due to the small-ish number of peaks in the fake CBED frame and the fact that cbed_frame has to centre its peaks on integer pixels despite using floating (and random) a/b vectors:
I have a feeling it's a tolerance issue, but it must be a rare one. The large tolerance is due to the small-ish number of peaks in the fake CBED frame and the fact that
cbed_frame
has to centre its peaks on integer pixels despite using floating (and random)a/b
vectors:At a later time we could modify the testcase to be more precise, but for now I would not worry.
Originally posted by @matbryan52 in #100 (comment)
The text was updated successfully, but these errors were encountered: