Error in python3 np.exp(matrix1 * matrix2) - "loop of ufunc does not support argument 0 of type float which has no callable exp method"

Question

I have a function where I need to perform np.exp(matrix1 @ matrix2), but I receive the error message: loop of ufunc does not support argument 0 of type float which has no callable exp method

matrix1 is a 210 by 4 matrix of float values
matrix2 is a 4 by 1 of float values
matrix1 @ matrix2 is a 210 by 1 of float values
type(matrix1 @ matrix) reports numpy.ndarray

numpy.exp() expects an array_like argument so I don't understand why this gags.

Error details:

newval = np.exp(matrix1 @ matrix2)

AttributeError                            Traceback (most recent call last)
AttributeError: 'float' object has no attribute 'exp'

The above exception was the direct cause of the following exception:

TypeError                                 Traceback (most recent call last)
<ipython-input-563-7924faaa390b> in <module>
----> 1 np.exp(matrix1 @ matrix2)

TypeError: loop of ufunc does not support argument 0 of type float which has no callable exp method

Seems like you've used a variable `np` to store a float value somewhere in your program, and that has overridden the aliasing of `numpy` as `np`. — amzon-ex, Jun 17 '20 at 03:18
May be relevant: https://stackoverflow.com/a/47208873/642070. It could be that when np.exp goes through the series, it assumes that the contained objects will have an `exp` method. — tdelaney, Jun 17 '20 at 03:28
amzon-ex, I imported numpy as np. np has not been used as a variable. — headlesschicken, Jun 17 '20 at 03:42
tdelaney, that link help solve it! It works with np.exp(np.asarray(matrix1 @ matrix 2).astype(np.float64)) — headlesschicken, Jun 17 '20 at 03:50
While the `astype` patches this, you really should figure out where the object dtype comes from. What's dtype for the those two arrays? — hpaulj, Jun 17 '20 at 06:02

hpaulj · Answer 1 · 2020-06-17T06:01:23.920

exp works on an array of floats:

In [186]: arr = np.array([1.,2.,3.])                                                                            
In [187]: np.exp(arr)                                                                                           
Out[187]: array([ 2.71828183,  7.3890561 , 20.08553692])

but not if the array dtype is object:

In [188]: np.exp(arr.astype(object))                                                                            
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
AttributeError: 'float' object has no attribute 'exp'

The above exception was the direct cause of the following exception:

TypeError                                 Traceback (most recent call last)
<ipython-input-188-b80d2a0372d5> in <module>
----> 1 np.exp(arr.astype(object))

TypeError: loop of ufunc does not support argument 0 of type float which has no callable exp method

Your array may have mixed objects - floats, ints, lists, who knows what.

===

Why is matrix1 @ matrix2 object dtype? What is matrix1 and matrix2? @ is normally a numeric operation, though it has been recently extended to work with object dtypes. But that is a slower calculation.

Error in python3 np.exp(matrix1 * matrix2) - "loop of ufunc does not support argument 0 of type float which has no callable exp method"

1 Answers1