1

I know how to compare 2 matrices of integers using Numpy. But does Numpy offer a way to get the list of elements that are different between them ?

1 Answers1

2

Something like this?

>>> import numpy as np
>>> a = np.zeros(3)
>>> b = np.zeros(3)
>>> a[1] = 1
>>> a == b
array([ True, False,  True], dtype=bool)

for floats: http://docs.scipy.org/doc/numpy-dev/reference/generated/numpy.isclose.html

>>> np.isclose(a,b)
array([ True, False,  True], dtype=bool)

indices of different elements 

 >>>[i for i,val in enumerate(np.isclose(a,b)) if val == False]

(using numpy instead)

>>> np.where(np.isclose(a,b) == False)

find values of different elements:

 >>> d = [i for i,val in enumerate(np.isclose(a,b)) if val == False]
 >>> a[d]
 array([ 1.])
 >>> b[d]
 array([ 0.])
user3684792
  • 2,542
  • 2
  • 18
  • 23
  • Thank you. Is not the `for()` loop expensive for huge arrays ? Is not there some ready made function within Numpy ? –  Feb 05 '15 at 15:52
  • it is a list comprehension not a for loop :) You could potentially use a map call instead, but it won't make that much of a difference http://stackoverflow.com/questions/1247486/python-list-comprehension-vs-map – user3684792 Feb 05 '15 at 15:58
  • maybe there is a better way in numpy, I will check – user3684792 Feb 05 '15 at 15:59