sum of python array gives slightly different precision depending on array order

Asked Nov 19 '15 at 23:07

Active Nov 19 '15 at 23:07

Viewed 32 times

I would expect summing arrays to give the same number regardless of order. This test disagrees. What is causing this?

>>> sum([0.2, 1.0, 0.3939393939393939, 0.8846153846153846])
2.4785547785547783
>>> sum([0.8846153846153846, 1.0, 0.3939393939393939, 0.2])
2.478554778554779

This is making tests annoying to write as sometimes I am getting slightly different floats that cause assertions to fail.

asked Nov 19 '15 at 23:07

tknickman

4,285
3
34
47

Can you show the test? Some numbers are impossible to properly represent (PI is one for example). – Reut Sharabani Nov 19 '15 at 23:12
I'm literally doing a raw assertion against a hardcoded value: self.assertEqual(value, 2.4785547785547783) – tknickman Nov 19 '15 at 23:12
2

Although this doesn't directly answer the question, python unit `TestCase`s have an `assertAlmostEqual` method that handles these precision errors. – pppery Nov 19 '15 at 23:13
I don't see any asserts here. – Reut Sharabani Nov 19 '15 at 23:13

sum of python array gives slightly different precision depending on array order

0 Answers0