Hello,
Thanks for the pointers to quartic code. These were not hard to find
with a web-search. The problem with them is they return all roots, real,
imaginary, positive, negative, etc. Since I only need the smallest
positive root, I was hoping someone may know of a way to use this to
speed the calculation up.
> What application do you have for this
It is esentially finding the time-to-collision for two initially
disjoint spheres moving along parabolic arcs. Furthermore, the curvature
of the parabolae is small, so a good guess can be obtained by
considering them to be straight. Speed is very important as I calculate
a few million such collision times. Another problem is that often two
spheres will be very close or maybe slightly overlap due to roundoff but
otherwise be moving away from eachother, and in this case the prediction
should be that no collision is to happen, rather then a close-to-zero
collision time...So I cannot and probably should not just use a boxed
quartic solver...
> Since the quartic is a closed formula, finding the results and checking
> those should be much speedier than trying to do any other logical
> testing.
I don't understand this claim?
Thanks,
Aleksandar
|