Re: [Boost-users] math tools roots

9 Oct 2011

      AMDG

On 10/09/2011 06:47 AM, Matwey V. Kornilov wrote:
...
I believe that there is a bug in eps_tolerance implementation. At line 34 of 
toms748_solve.hpp one can see:
return (fabs(a - b) / (std::min)(fabs(a), fabs(b))) <= eps;
When both a and b equal zero this condition is evaluated to a false instead 
of a true. Left hand of condition is NaN and the right one is a floating-
point number hence the result is false.
Well obviously, a user wants to see true here because the relative length of 
interval in this case tend to zero.
What exactly do you mean by "the relative length of
the interval tends to zero?"

lim_{(a,b)->(0,0)} |a-b| / min(|a|, |b|)

doesn't exist.

In Christ,
Steven Watanabe