[
https://issues.apache.org/jira/browse/MATH-1083?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13856600#comment-13856600
]
Luc Maisonobe commented on MATH-1083:
-------------------------------------
0/0 is also the limit of x/0 when x>0 converges towards 0+, and in this case
the result should be POSITIVE_INFINITY.
0/0 is also the limit of x/0 when x<0 converges towards 0-, and in this case
the result should be NEGATIVE_INFINITY.
These two cases, plus the case pointed out be Sebb, plus the case pointed out
by Andrew show that 0/0 is a very special case and no single value can be
arbitrarily fixed for it. This is the reason why, when you deal with floating
points numbers 0.0/0.0 is NaN (whereas x/0 is not NAN but either
POSITIVE_INFINITY or NEGATIVE_INFINITY according to the sign of a non-zero x).
So I am reluctant to change the behaviour we have now: 0/0 is not defined.
In order to fix your problem, do you have some outside way to determine if
either the numerator or the denominator appears to be almost 0 but has at least
a sign? In this case, you may be able to handle this special condition in your
calling application.
> Fraction - Allow zero denominator if numerator is also zero
> -----------------------------------------------------------
>
> Key: MATH-1083
> URL: https://issues.apache.org/jira/browse/MATH-1083
> Project: Commons Math
> Issue Type: Improvement
> Affects Versions: 3.1
> Reporter: Andrew Brampton
> Priority: Minor
> Labels: Fraction
>
> The Fraction constructor rejects a denominator of zero. However, while trying
> to parse some Strings generated by another program, I encountered a fraction
> "0/0". I would say that's ambiguous as to if that's valid, but some people
> would agree the answer is zero.
> So, would it be possible to allow a zero denominator if the numerator is also
> zero. I'd be happy to provide a patch.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
