Complex "intervals" via conic sections ?



>>>>> "RD" == Robert Dodier <robert.dodier at gmail.com> writes:

RD> On 2013-02-13, James Cloos <cloos at jhcloos.com> wrote:
>> I've wanted to try specifying intervals as RVs, but the complexity of
>> arbitrary RVs has been overwhelming.  Especially when the RVs end up
>> having multiple peaks, shear cliffs and the like (visualize them as
>> height maps on the complex plane).

RD> What is an RV?

Random Variable.

I was looking at it from a numeric rather than symbolic angle, and
wanted to approximate the RVs with piecewise rational polynomials.
Probably storing them as b-splines.

It has been a while, but it probably was their density functions I
wanted to approximate with the splines, but in retrospect I wonder
whether their characteristic funtions would have been better?

Arithmetic on them isn't terribly horrible, but the libm functions
were a bit overwhelming at the time.  Especially when they need to
use different approximations on different chunks of C.

Maybe handling them symbolically would be easier?

-JimC
-- 
James Cloos <cloos at jhcloos.com>         OpenPGP: 1024D/ED7DAEA6