Comparing floating point numbers with numpy and scipy

Question

Is there an epsilon constant (like in Matlab) in numpy or scipy modules for comparing floating point numbers?

+3

python numpy scipy

user983302 Nov 03 '12 at 18:30

source share

1 answer

mtrw · Accepted Answer · 2012-11-03T18:40:24+0000

You can use np.finfo

>>> import numpy as np >>> np.finfo(np.double).eps 2.2204460492503131e-16 >>> np.finfo(np.single).eps 1.1920929e-07