Re: [Quantlib-dev] subtle bug in InverseCumulativeNormal

[Luigi B. <lui...@gm...>]

On Thu, 2010-05-27 at 01:58 -0500, Kakhkhor Abdijalilov wrote:
> If the argument x is close to 0.0 or 1.0, the implementation sets it
> exactly to 0.0 or 1.0.
> This leads to an attempt to evaluate std::log(0.0) or std::log(1.0).

You're right.  I wonder why it does it though, since 0 and 1 are outside
the domain of the function.  If I were to return a value, I'd return the
maximum allowed double for 1 and the minimum for 0 (to get an
approximation of plus/minus infinity.)  But I'd rather throw an
exception in that case.

Thoughts?

Luigi


-- 

Call on God, but row away from the rocks. 
-- Indian proverb