Python class sum with numpy array

Question

I have a question concerning overriding operands. I just tried to override the __add__(self, other) operator in a custom class, such that on of its elements (a numpy array) can be added to another numpy array. To make both directions of summing possible I both declared the __add__as well as the __radd__ operator. A small example:

import numpy as np

class MyClass():
    def __init__(self, x):
        self.x = x
        self._mat = self._calc_mat()


    def _calc_mat(self):
        return np.eye(2)*self.x

    def __add__(self, other):
        return self._mat + other

    def __radd__(self, other):
        return self._mat + other


def some_function(x):
    return x + np.ones(4).reshape((2,2))

def some_other_function(x):
    return np.ones(4).reshape((2,2)) + x


inst = MyClass(3)

some_function(x=inst)
some_other_function(x=inst)

Strangely, I get two different outputs. The first ouput, from some_function is just like expected:

Out[1]
array([[ 4.,  1.],
       [ 1.,  4.]])

The second output gives me something odd:

Out[2]:    
array([[array([[ 4.,  1.],
       [ 1.,  4.]]),
        array([[ 4.,  1.],
       [ 1.,  4.]])],
       [array([[ 4.,  1.],
       [ 1.,  4.]]),
        array([[ 4.,  1.],
       [ 1.,  4.]])]], dtype=object)

Does somebody have an idea why is that?
Thanks, Markus :-)

Community · Accepted Answer · 2017-05-23 12:34:30Z

2

the issue is that numpy array is also implementing an __add__ method, and it is called before your __radd__

you can see this answer for a solution: https://stackoverflow.com/a/22633052/7033869

edited May 23, 2017 at 12:34

CommunityBot

11 silver badge

answered May 11, 2017 at 0:50

Francois Mockers

7334 silver badges14 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Python class sum with numpy array

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related