
### $Id: README,v 1.4 2008/01/11 04:24:42 goswami Exp $


                        README for the EMC package
                        --------------------------


================================================================================
QQ.     
        What does this package has to offer over existing ones for
        doing MCMC (e.g., mcmc)?
                        

AA.     
        There are many new features in this package as listed below:

        i.     
                The function randomWalkMetropolis is a generalization
                of the metrop function found in the package mcmc, in
                that it lets the user provide arbitrary symmetric
                proposals, not necessarily (multivariate) normal
                (e.g., multivariate) t or uniform).
                        
        ii.
                The MetropolisHastings function is a generalization of
                randomWalkMetropolis function in that it lets the user
                provide asymmetric proposals.
                        
        iii.      
                The functions evolMonteCarlo, findMaxTemper and
                placeTempers are implementations of new published
                research materials, see the references in the help
                pages.  
                


================================================================================
QQ.     
        Why did you write the back-end of this package in C and not in R?


AA.
        This software grew out of the C code base which was used by
        the author in his research. He used to implement the
        logTarDensFunc, MHPropNewFunc and the logMHPropDensFunc in C
        for faster (research) speed.
                                                                     
        The interface of this package, however, needs that the user
        provides the above-mentioned functions implemented in R, which
        slows down the underlying C code because of constant back and
        forth between the C-space and the R-space. 

        There is TODO feature of this package, which will be only
        beneficial to the advanced user who knows how to make R and C
        talk. One of the future releases of this package will
        entertain the above-mentioned user-provided functions written
        in C by the user. Of course, that will entail building the
        package from source by the user without errors, but hey, here
        the assumption is that the user is advanced! And then things
        will be as fast as it gets, due to minimal back and forth
        between the C-space and the R-space.



                ---Happy coding and research!---
