Closed
Description
np.clip allows to clip multi-dimensional arrays by taking arrays for a_min and a_max. However, when providing bounds that are array-likes that don't match the source sometimes one can weird results.
For example, while
a = np.arange(5)
a.clip([0, 0], [1, 1])
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-23-ad92301a6712> in <module>()
1 a = np.arange(5)
----> 2 a.clip([0, 0], [2, 2])
ValueError: operands could not be broadcast together with shapes (5,) (2,)
raises an error as expected, the following passes:
a = np.arange(5)[:, None]
a.clip([0, 0], [2, 2])
array([[0, 0],
[1, 1],
[2, 2],
[2, 2],
[2, 2]])
I guess what's happening is that the input array is broadcasted to fit the constraints. Is this expected? I would have expected an error to be raised instead.
Metadata
Metadata
Assignees
Labels
No labels