| 1 |
On Sat, Aug 22 2015, Michael Orlitzky wrote: |
| 2 |
|
| 3 |
> On 08/22/2015 09:42 AM, R0b0t1 wrote: |
| 4 |
>> https://en.wikipedia.org/wiki/Round-off_error |
| 5 |
>> https://en.wikipedia.org/wiki/Machine_epsilon |
| 6 |
>> |
| 7 |
>> Either add a tolerance (a - b <= t) or compare them as strings as |
| 8 |
>> you've been doing. |
| 9 |
>> |
| 10 |
> |
| 11 |
> You probably want |a - b| <= t there =) |
| 12 |
> |
| 13 |
> But... that can cause problems too. If your numbers are small enough, |
| 14 |
> you can wind up with infinity or NaN (not a number) and then your |
| 15 |
> comparisons will go berserk. |
| 16 |
> |
| 17 |
> Floating point addition isn't even commutative: |
| 18 |
> |
| 19 |
> > 0.1 + 0.2 + 0.3 |
| 20 |
> 0.6000000000000001 |
| 21 |
> > 0.1 + (0.2 + 0.3) |
| 22 |
> 0.6 |
| 23 |
|
| 24 |
That demonstrates non-associativity. I believe floating point is |
| 25 |
commutative: a+b = b+a |
| 26 |
|
| 27 |
> Better to avoid that quagmire entirely if you can. Use fixed point |
| 28 |
> arithmetic, arbitrary precision, or even rational numbers if you can get |
| 29 |
> away with it. |
| 30 |
|
| 31 |
Agreed. |
| 32 |
allan |
Archives