BUG: Error when using nanmax() and nanmin() on an object array with an axis specified. #9008
Description
This was the underlying cause of a problem reported on stackoverflow: http://stackoverflow.com/questions/43659827/numpy-error-when-specifying-axis-in-nanmax-while-nansum-works-an-the-same-case
The error is raised when applying nanmax() or nanmin() to an array with object data type and specifying an axis.
Here's the example from my answer:
In [2]: import numpy as np
In [3]: np.__version__
Out[3]: '1.13.0.dev0+bca7922'
In [4]: a = np.array([[1.0, 2.0], [3.0, 4.0]], dtype=object)
In [5]: np.nanmax(a, axis=0)
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-5-a020f98a2536> in <module>()
----> 1 np.nanmax(a, axis=0)
/Users/warren/miniconda3numpy/lib/python3.5/site-packages/numpy-1.13.0.dev0+bca7922-py3.5-macosx-10.6-x86_64.egg/numpy/lib/nanfunctions.py in nanmax(a, axis, out, keepdims)
343 # Fast, but not safe for subclasses of ndarray
344 res = np.fmax.reduce(a, axis=axis, out=out, **kwargs)
--> 345 if np.isnan(res).any():
346 warnings.warn("All-NaN slice encountered", RuntimeWarning, stacklevel=2)
347 else:
TypeError: ufunc 'isnan' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''
There is no error when the axis is not given:
In [6]: np.nanmax(a)
Out[6]: 4.0
nansum() handles the axis without an error:
In [7]: np.nansum(a, axis=0)
Out[7]: array([4.0, 6.0], dtype=object)
There are several other issues involving the nan-functions and object arrays, but I couldn't tell if this issue is a duplicate. Sorry for the noise if it is.