[ADMB Users] mcmc and -noinit
mbrooks at ufl.edu
Thu Mar 22 10:07:28 PDT 2012
On 22 Mar 2012, at 12:00 PM, dave fournier wrote:
> It appears to be walking out of an array.
> The problem is whether it keeps track of what parameters should be active when there are
> random effects. One versio of the mcmc does mcmc on the "fixed effects" while
> integrating over the random effects via the LA.
> The other version uses mcmc on the FE and RE combined. Which one do you want?
I'm really interested in the uncertainty in the variances of the random effects. I don't know off the top of my head which is better for this. Alternatively, I'd be happy if I could do profiling on the variances.
More information about the Users