x = qbeta(p,a,b)
Input p probability
a,b positive reals, parameters of the beta distribution
(p,a,b can be scalar or matrix with common size)
Output for each element of p, F is an ``inverse'' of the
beta cumulative distribution function with parameter
a and b (the beta density is :
x--> x.^(a-1) .* (1-x).^(b-1) ./ beta(a,b)1_{0<x<1})