Jan Ploski escribi?:
> Hello,
>
> Having briefly checked out Alexey Beshenov's logic.mac package (discussed
> here earlier:
> http://www.math.utexas.edu/pipermail/maxima/2008/010825.html), I started
> wondering how difficult it would be to implement probability theory rules
> (sum and product rule) to allow the calculation of probabilities of
> related logical propositions, Bayesian-style? I have absolutely no prior
> experience with Maxima and CAS, so I can't estimate whether it is a piece
> of cake or a big project, especially for someone who's already "been there
> done that".
>
I suppose the difficulty depends on the level of generality.
* Do you allow stochastic dependence among variables? in this case the
procedures should handle multivariate distributions.
* Even if you only assume stochastic independence, you should solve
problems like "what's the distribution of X*Y*Z, where X is binomial, Y
is gumble and Z is chi^2 ?"
* But if you restrict your model to stochastic independence and
bernoulli distributions, This is closer to human beings.
* Driving in another direction, I ask if something as fuzzy logic could
be of some interest. I don't know much about that.
Forgive me if this post doesn't answer your question. I was thinking
aloud, since this question came to my mind two or three times in the past.
--
Mario Rodriguez Riotorto
http://www.telefonica.net/web2/biomates