ENH: interpolate: allow Akima extrapolation #20304
Conversation
github-actions
bot
added
scipy.interpolate
enhancement
There was a problem hiding this comment.
Thanks @psmd-iberutaru can you add some test coverage for this new functionality
|
Apologies @j-bowhay. Added the missing unit test(s). If there is a missing test, please let me know so I may add it. |
|
I guess the doc of the argument “extrapolate” should read “If bool, …” instead of “If true, …” to be consistent with the doc of CubicHermiteSpline and PPoly. @psmd-iberutaru |
| ak_none = Akima1DInterpolator(x, y, extrapolate=None) | ||
| # None should default to False; extrapolated points are NaN. | ||
| assert_allclose(ak_false(x_ext), ak_none(x_ext), equal_nan=True) | ||
| assert_equal(ak_false(x_ext)[0:4], np.full(4, np.nan)) |
There was a problem hiding this comment.
TIL that assert_equal(np.nan, np.nan) is True not False.
|
Jake @j-bowhay , your review is red on GH, is there something you'd like to see in this PR? |
j-bowhay
dismissed
their stale review
Past comment resolved, have not had time to review again
lucascolley
changed the title
|
CI is green, review comments are addressed, the patch LGTM and is merged. Thank you @psmd-iberutaru, all. And congrats with your first scipy contribution! Keep them coming :-). |
Reference issue
Minor change. Closes #20247
What does this implement/fix?
We add the extrapolation parameter to the construction; passing it to the parent classes. We keep False as the default as that is the behavior of the function before this change.
Additional information