How to check if the sum of floats equals a number and accounting for Floating Point Inaccuracy

Asked May 08 '15 at 12:34

Active May 08 '15 at 12:34

Viewed 141 times

I'm trying to check if a sum of floating point numbers adds up to 100. And of course they don't, due to floating point accuracy.

So, what doesn't work is:

test = [99.9, 0.025, 0.025, 0.0125, 0.0125, 0.025]
sum(test) == 100.

alternatively

sum(test) > 99.99999 and sum(test) < 100.0000001

works fine, but I consider it as ugly, because the limits are arbitrary. Is there a more elegant method in python to check float numbers on equality? Or at least a method to choose the limits not only arbitrary but based on the internal accuracy?

asked May 08 '15 at 12:34

MichaelA

1,866
2
23
38

You could just test whether `abs(sum(test) - 100) < epsilon` – EdChum May 08 '15 at 12:35
http://stackoverflow.com/questions/4028889/floating-point-equality-in-python – zw324 May 08 '15 at 12:36
1

The sum of the things in that array is about 5.69 * 10^(-15) larger than 100. Are you sure you want the two to compare equal? – tmyklebu May 08 '15 at 13:35

How to check if the sum of floats equals a number and accounting for Floating Point Inaccuracy

0 Answers0