Posting out loud clears the brain, answer is:
cmat=as.matrix(fit)
clist=apply(cmat,1,function(x) relist(flesh = x,skeleton = fit@inits[[1]]))
ulist=lapply(clist,function(x) unconstrain_pars(fit,x))
Handy for generating quantities separately, with this extra step:
ygen = lapply(ulist, function(x) constrain_pars(sfgen,x))
where sfgen is a stanfit object with the same parameters and data but different generated quantities / other bits.
1 Like
In PyStan you can. Check following test for example
But I believe you are interested about RStan
Edit. You answered already :)
1 Like