CODEWITHSUNDEEP
pythonarraysnumpyassertapproximate
DeltaIV
DeltaIV

Reputation: 5646

Find numpy array in bigger array, testing for approximate equality

I often need to find a smaller array xs in a larger array X. However, since these arrays are the result of floating point arithmetic, exact comparison fails miserably. How can I fix this? MRE:

import numpy as np
np.random.seed(2021)
    
X = np.random.rand(20,4,3)
    
idx = 4
xs = X[4,3]
    
idx_x = np.where((X[:,3,:] == xs).all(axis=1))[0]
assert idx_x == 4 # pass

xs = X[4,3] + 1e-9
idx_x = np.where((X[:,3,:] == xs).all(axis=1))[0]
assert idx_x == 4, "exact equality fails" # fail

How can I modify the code so that it checks for approximate, rather than exact, inequality?

Upvotes: 1

Views: 25

Answers (0)

Related Questions