Re: Log CDF and PDF

> bci2100=p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
> plot(p,bci2100)
> % 0<p,bci, b1-2<=1
>
> How can I define in Matlab:
> - PDF for bci with a.m. dicrete'p' sequence;
> - CDF and PDF for bci as continiuos function;
> ...

You have quite a debate going on with other folks.

I would be interesting in answering your question, but I don't understand
it. If you want to generate random p values and use the bci2100 expression
to derive random bci2100 values, this does that:

F = @(p) p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
newbci = F(rand(1000,1));
hist(newbci, 40)

If you want to treat your curve as a cdf and invert it, in other words
generate random bci2100 values and find the corresponding p values, you
could do the following interpolation as an approximation:

plot(p,bci2100)
pnew = interp1(bci2100,p,rand(1000,1));
hold on; ecdf(pnew); hold off % <-- for comparison, if you have the
Statistics Toolbox

It's unlikely you will find someone to "generate a detailed descriptive
statistics report" for you, but if you ask a specific question related to
that you may get some help.

-- Tom