[FIXED] NumPy change float precision for the array

Issue

How do I change the float precision for an entire array without having to do np.set_printoptions? It is not about printing; I want the values to be compressed to do equality checks with other arrays.

e = np.array([0.8292222222222225, 0.1310000000000003]) 

to

e = np.array([0.829225, 0.131003]) 

I want to be able to compare 0.8292222222222225 with 0.829225 in other arrays.

Cannot get equality as True

>>> e = np.array([0.8292222222222225, 0.1310000000000003]) 
>>> e[0]
0.8292222222222225
>>> e[0]==0.829225
False
>>> 

Solution

You should use the numpy.isclose function. It allows you to compare elements within a tolerance:

>>> a = np.array([0.8292222222222225, 0.1310000000000003])
>>> b = np.array([0.8293, 0.132])
>>> np.isclose(a, b, atol=1e-3)
array([ True,  True])
>>> np.isclose(a, b, atol=1e-4)
array([ True, False])
>>> np.isclose(a, b, atol=1e-5)
array([False, False])

Note that you can use both the relative tolerance and the absolute one.

Answered By - Riccardo Bucco