[WP34s] Regularized incomplete Beta function

[Dieter]

(05-12-2014 06:58 AM)Paul Dale Wrote:  I don't understand why there is the problem with 1/2 but will look into when I get a chance.

Fine. In the meantime I have set up a similar routine for the Chi² quantile. I tried the same Halley approach as in the Student case, and it worked very well – until the number of dof got beyond 100 or 200. At this point either an extremely good estimate is required or the approximation will converge very slowly at first (until the approximation is very close to the true value and rapid convergence sets in). That's why I tried a different approach. The Normal quantile guess also got an update and now is more accurate than before. Only three constants need to be adjusted.

I think I'll post the results later in a new thread.

Dieter