MIXTURES MODELS

["MAGGIO GUIDO" <GUIDO.MAGGIO@ESTERNI.GRUPPO.MPS.IT>]

Hi

Is there a library or anybody who knows how to fit mixture models in Splus. I have found this possible using MASS but only for a set of Gaussian distributions. Nlminb was used to derive ML estimates (I should then derive myself any possible case of mixture! Quite hard to do!)).

I have found an R library (MIXDIST) that can mix 2 or more Weibull, 2 or more Gaussian, 2 or more LogNormal and so on.
I am using this one at the moment but it has some problems:
1) it does not return the parameters of the fitted distributions but just mean and SD.
2) I cant mix 2 different distribution such as a Weibull and a Gamma.

Can anybody help or make suggestions?
Thanks

Guido