Hi
Is there any reason for the following:
A tiny program such as
model pois;
#var
# pois[100],bino[100];
{
for (i in 1:100){
pois[i]~dpois(1)
bino[i]~dbin(0.01,100)
}
}
run under WinBUGS generates sensible initial values (as produced in the
log window after the 'Gen Inits' is activated but before an update) for
the binomial case but nonsense for the Poisson - often all zeroes for
example? When it updates, the results seem sensible.
Since there is no node tool in classic BUGS this can't be tested.
Just wondering - the node tool is very useful for checking that the
intermediate results of a calculation are sensible and I have the arrays
the right way round.
John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|