MRG: Fix multitaper #2934
MRG: Fix multitaper #2934
Conversation
|
Ready for review/merge from my end. |
| @@ -33,6 +33,9 @@ def test_dpss_windows(): | |||
| assert_array_almost_equal(dpss, dpss_ni) | |||
| assert_array_almost_equal(eigs, eigs_ni) | |||
|
|
|||
| dpss, _ = dpss_windows(245411, 4, 8, False, 1000) | |||
| assert_equal(dpss.shape[-1], 245411) | |||
There was a problem hiding this comment.
~1 sec. But we could get rid of it if you want. I'm + 0.5 on removing it but I was afraid of no-test complainers :)
There was a problem hiding this comment.
|
@agramfort done. Merge once CIs are happy? |
|
@Eric89GXL what are the consequences of this bug? Is it in the category {hashtag}recompute_your_results? |
|
The bug made it so that if you used really long signals you'd get a dimension error during multiplication. So if you haven't hit it, then it probably didn't affect you. I also included another tweak that appeared upstream that determined when to flip the multitaper. I'm not 100% sure of the consequences there, but I'd expect them to be minor. |
|
@dengemann okay to merge for you? |
|
Yes! |
|
+1 for MRG |
|
Thanks for the quick reviews |
Closes #2931.
Upstream bug nipy/nitime#140.