On 12/30/06, Mario Rodriguez <biomates at telefonica.net> wrote:
> Specific numeric procedures would speed up computations in your case,
> since you are working with a lot of sample points. But the idea behind
> the interpolating functions is to have symbolic capabilities as well.
I believe an expression in terms of charfun is appropriate for
symbolic computation. If there is a conversion to a more efficiently-
calculated form, that would be a separate function, applied to the
charfun expression.
> For searching, a bipartition method could be fast enough: first compare
> x with element in position length(z)/2, and take the inferior or
> superior half of pairs, and repeat this process until you find the
> interval.
I see a computational geometry package on the horizon ... 8^)
Tesselations, convex hulls, all that good stuff ... fun to think about.
all the best
Robert