Numpy arange function has inconsistent behaviour (potentially bug?)

Asked Sep 12 '16 at 15:25

Active Sep 12 '16 at 15:25

Viewed 34 times

I found a very strange behaviour of numpy.arange.

If I wrote

np.arange(1.1, 1.7, 0.1)

it returns array([ 1.1, 1.2, 1.3, 1.4, 1.5, 1.6]) as expected.

However, when I wrote:

np.arange(1.1, 1 + 0.1 * 7, 0.1)

it returns array([ 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7]).

Is this a bug or I used it in a wrong way?

asked Sep 12 '16 at 15:25

spacegoing

1

Possible explanation at: http://stackoverflow.com/questions/10011302/python-numpy-arange-unexpected-results?rq=1 – Versatile Sep 12 '16 at 15:27
`0.1` is an infinite number in binary... – dawg Sep 12 '16 at 15:28
Also related: [Is floating point math broken?](http://stackoverflow.com/questions/588004/is-floating-point-math-broken) – ayhan Sep 12 '16 at 15:31
@ayhan That's very helpful. Thanks! – spacegoing Sep 12 '16 at 15:40
Consider `np.arange(11, 17) * 0.1` for predictable behaviour – Eric Sep 12 '16 at 21:10

0 Answers0