RE: [Algorithms] Bounding cones.

[Willem de B. <wd...@pl...>]

>
>You can take the structure of Welzl's algorithm (who took it from
>Seidel's LP algorithm) and add the necessary operations for your case:
>
>	1. Test a cone for containment in another cone
>	2. Construct a bounding cone for a set of 1 (trivial), 2, or 3
>        cones.
>
>Notably the hull for 3 cones could be a challenge.

I think as a reasonable approximation, you could take the average
of the 3 Vi's, that make up the three cones {Ci}, and use that
as the V for your containing cone. Then compute H (ie.,
cos(half-angle-of-V)) from that..?

- Willem