http://qs1969.pair.com?node_id=726652

Strictly, it's not because it's floating point, it's because it's binary floating point.

When decimal 1.51 is converted to a binary fraction the result is:

1.1000_0010_1000_1111_0101_1100_0010_1000_1111_0101_1100_0010_1000_1111_0101_1100_0010_1000_1111_0101...
which you can see is a repeating binary fraction.

As you say, most decimal fractions are like this... so before any errors can be introduced by rounding and what not, most decimal fractions have a built-in "representation" error.

Most of the time you won't see the "representation" error, because conversion back to decimal rounds it off. This can lead to bafflement when two values look the same when printed out, but fail \$x == \$y. Addition and subtraction are the more difficult floating point operations, so you're more likely to see the problems there. Consider:

print 1.09 - 1, "\n" ; print "0.84 - 0.34 == ", 0.84 - 0.34, ( 0.84 - 0.34 == 0.5 ? " ==" : " but !=" ), " 0.5\n" ;
which gives:
0.0900000000000001
0.84 - 0.34 == 0.5 but != 0.5
it really makes me wonder why we persist in using binary floating point for decimal arithmetic !

Replies are listed 'Best First'.
Re^5: How can I calculate the right combination of postage stamps?
by gwadej (Chaplain) on Nov 28, 2008 at 19:27 UTC

This issue is caused by binary floating point. But, it is an artifact of any floating point notation. For example,

1/3 + 1/3 + 1/3 != 1

for any finite number of digits using decimal representations.

Given any base, there will be some fractions that cannot be properly represented and would therefore cause this problem. (At least until such time as we have infinite precision floating point representations.<grin/>)