1) New type of move proposed in MCMC algorithm embedded in mcmcFmodel
Now makes joint updates of c and f, instead of separate block updates
of c and f.
It  improves mixing of the chain and avoids (or at least enhance w.r.t) troubles with ghost populations;


2) Function simFmodel no longer makes use of argument nloc (numer of loci). It was obviously redundant with argument allele.number and now nloc is evaluated inside simFmodel by nloc=lengh(allele.number)


3) Update of the number of population (subroutines bdclass7) slightly corrected in order to process populations whatever the number of tiles or individuals they currently contain. This is implemented in fortran subroutines bdclass8 and bdclass8bis.


4) All computations now carried out in double precision


5) Now use the R random number generators (via the C wrapper).


6) Function FormatGenotypes is now automatically called by function mcmcFmodel.


7) Use R function sample instead of my function rdiscr


8) Handle haploid datasets: haploid data are first ploidized in function mcmcFmodel all computations are identical to the diploid case except the evaluation of the likelihood for which the squared root is taken. Function mcmcFmodel requires now an extra argument 'ploidy' ste to 2 (diploid) by default.
 



SYNTAXIC CHANGES:
 

See points (2), (6), (8)  above as documented in the on-line help.