I am hoping that you can provide some guidance on adding new distributions to
JAGS. I have the CVS source, and looking at the code for existing
distributions, I am somewhat optimistic that I can code up the distribution I
would like to add. It's a vector-valued distribution, not that different from
the existing "ddirch". However, assuming I implement the class and necessary
methods, I am completely at a loss as how to incorporate it into the codebase.
For example, do I need to recompile my distribution (and if so, how to make
"make" aware of it), or is there some way to separately compile it as a module
to be loaded in at run-time under existing distributions?
A documented example of creating a new distribution, from the point of an end-
user, would be a tremendous help (though this is perhaps more of a feature
In order to continue this post, I would also require a new distribution, the Conway Maxwell Poisson.
I am aware of the documentation explaining how to do it http://www.cidlab.com/prints/wabersich2013extending.pdf but I also have to admit my lack of experience in C++... Is there another way?
There is the compoisson package in CRAN but this does not support GLMM. I am hopping to adapt GLM poisson model to such new distribution.
Log in to post a comment.