ruby rounding?

[Kyle Schmitt]

Something interesting came up at work concerning rounding. Notably,
that the round procedure in vb.net was good for stats (round to even),
but since what the programmers expected (symmetric arithmetic
rounding) was different, some edge cases were failing their tests.

This got me thinking, what does Ruby use for rounding? It turns out
(example below) it's symmetric arithmetic rounding. But the methods
in C's math.h give the programmer a choice of which rounding method to
use.

Since there's so much development going on in Ruby currently, maybe
something like this could be/should be added.

See http://en.wikipedia.org/wiki/Rounding for details on the methods

--Kyle
-------------
Using this array in several languages:
[-5.0, -4.5, -4.0, -3.5, -3.0, -2.5, -2.0, -1.5, -1.0, -0.5, 0.0, 0.5,
1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5]

rounded:
ruby
[-5, -5, -4, -4, -3, -3, -2, -2, -1, -1, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5]

gcc
[-5, -5, -4, -4, -3, -3, -2, -2, -1, -1, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5]


vbs (ugly but I had it on hand)
[-5, -4, -4, -4, -3, -2, -2, -2, -1, 0, 0, 0, 1, 2, 2, 2, 3, 4, 4, 4]

vb.net (again ugly, but on hand)
[-5, -4, -4, -4, -3, -2, -2, -2, -1, 0, 0, 0, 1, 2, 2, 2, 3, 4, 4, 4]

 

[james.d.masters]

Since there's so much development going on in Ruby currently, maybe
something like this could be/should be added.

You could always create your own . Here is a something that I put
together. There may be a better method. The weird thing in the code
with comparing the remainder to 1e-9 is to deal with floating point
precision issues. This may or may not be appropriate... I'm not a
rounding expert but it's worked in applications that I developed.

Anyway, here it is:

class Numeric
def even_round()
if ((self.truncate % 2 == 0) and ((self.remainder(1).abs -
0.5).abs < 1e-9))
# integer portion is even and on 0.5 threshold
self.truncate
else
# integer portion is odd or not on 0.5 threshold
self.round
end
end
end

irb(main):012:0> a = [-5.0, -4.5, -4.0, -3.5, -3.0, -2.5, -2.0, -1.5,
-1.0, -0.5, 0.0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5]
=> [-5.0, -4.5, -4.0, -3.5, -3.0, -2.5, -2.0, -1.5, -1.0, -0.5, 0.0,
0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5]
irb(main):013:0> a.map {|e| e.even_round}
=> [-5, -4, -4, -4, -3, -2, -2, -2, -1, 0, 0, 0, 1, 2, 2, 2, 3, 4, 4,
4]

 

[Chris Shea]

You could always create your own . Here is a something that I put
together. There may be a better method. The weird thing in the code
with comparing the remainder to 1e-9 is to deal with floating point
precision issues. This may or may not be appropriate... I'm not a
rounding expert but it's worked in applications that I developed.

I could be wrong, but I think this is what the constant Float::EPSILON
is for. Where I'm sitting it's 2.22044604925031e-16.

Chris

 

[james.d.masters]

I could be wrong, but I think this is what the constant Float::EPSILON
is for. Where I'm sitting it's 2.22044604925031e-16.

Excellent - thanks for the info.