From a quick glance at the code, and from my reading of the paper, I am wondering why you did not consider using STAN to run the MCMC?
It is fast, there is a considerable amount of development that has gone into it, there are associated libraries to help with model selection and assessment (e.g., loo), useful libraries for visualisation (e.g., bayesplot and ShinyStan), and they provide tools to help package precompiled binaries of the models with R (package rstantools).
Hello.
I have not tried BactDating yet, but I am keen.
From a quick glance at the code, and from my reading of the paper, I am wondering why you did not consider using STAN to run the MCMC?
It is fast, there is a considerable amount of development that has gone into it, there are associated libraries to help with model selection and assessment (e.g.,
loo
), useful libraries for visualisation (e.g.,bayesplot
andShinyStan
), and they provide tools to help package precompiled binaries of the models with R (packagerstantools
).More information here: http://mc-stan.org/users/interfaces/rstan and here http://mc-stan.org/users/interfaces/
Best. A.