Hi Denis,
Those sound like just the kind of problems I believed existed in the
CC2D algorithm.
You might want to submit it as a separate push and get credit for fixing
4645692 (solveCubic doesn't return all answers), and maybe even the
following failures in the containment methods (which could be closed as
dups if this fixes the problems) as well:
4724552
4493128
4074742
4724556
(etc. Those were just the bugs I found on the first 2 pages of a bug
database search)
Double (triple, etc.) credit - woohoo! ;-)
...jim
On 12/13/2010 2:30 PM, Denis Lila wrote:
Very nice! How does it compare to CubicCurve.solveCubic() (which I
know
has issues with the 2 root case, but I got it from a "reliable source"
-
some textbook on Numerical Recipes)?
I wrote a tests that generated 2559960 polynomials, and in 2493075 of
those, the computed roots were identical to within 1e-9. They were
different in the remaining 66885 cases, so that's 97.4% agreement.
I've looked through some of the differences, and in every case the
function from graphics gems is better in one of two ways:
1. the gg version will report more roots than the cc2d version, in
which case the polynomial has a double root and the cc2d version
completely misses it (example poly: a=19.000000, b=-20.000000,
c=-17.000000, d=18.000000, where cc2d misses the root at 1).
2. the gg version will report fewer roots than the cc2d version, in
which case there was a 0 root and the cc2d version incorrectly split
it into -1e-163 and 1e-163.
So, the graphics gems version seems to be much more stable. It
does have a problem where it can return NaN sometimes, because it
assumes that the polynomial is not a quadratic one, but that can
easily be fixed.
So, should I put this new cubic root finder in CubicCurve.solveCubic
and use that in pisces?
Regards,
Denis.
