What is necessary to run MCMC?
MCMC is easy to implement in ADMB using the -mcmc N -mcsave N2 commands when running a model to run a chain of length N and save the output every N2 steps. To get output of parameter values that have been saved, you need to either read the binary .psv file (i.e. using the readBin function in R–see example code), or include in the code some commands to write to a file which are conditioned on the statement if(mceval_phase()) and then run the model again using the -mceval command. Finally, it is necessary to have at least one derived quantity designated as an sdreport variable in your model.
