I need to sum over an indicator function in a set.
sum() can be directly used on np.array with no issues.
import numpy as np
y = np.arange(10)
ans = sum(y % 2 == 0)
However, sum() cannot be used directly on python lists.
z = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
ans = sum(z % 2 == 0)
On the other hand, this works for np.array and python lists.
ans = sum(each % 2 == 0 for each in z)
print(ans)
The question is why? Both np.array and python lists should be iterables. What makes np.array support the additional syntax?
It's because Python list doesn't implement that out of the box, if you want, you can always subclass python list and have own custom list to allow modulo operation via __mod__ method:
class MyList(list):
def __mod__(self, i):
return [item%i for item in self]
>> data = MyList([1,2,3,4,5,6,7,8,9])
>> data%3
[1, 2, 0, 1, 2, 0, 1, 2, 0]
You can take a look at official Python documentation to know about the methods that can be overwritten.