Hi All,
I have a fairly simple model and data as shown below.
Everything is okay until I update the model, at which point I
get the following error:
update error for node <a[1]> algorithm non conjugate
dirichlet updater error can not sample node too many
iterations
If you see my mistake I'd greatly appreciate your help.
Some background on the inference problem I have... I have
two urns, each filled with balls labeled 1:14. Each urn has a
different distribution. I have conducted two experiments that
lead to my observed data. In the first experiment, I draw
exactly one ball from urn A and 60 balls from urn B, but I do
not know which ball came from which urn, only the final count
of each label 1:14. In the second experiment I again draw
exactly one ball from urn A, but draw 29 balls from urn B.
Again, I only know the final counts. I want to infer the
underlying distribution of labeled balls in each urn.
Thanks,
Seyab
model {
theta <-1
for(i in 1:14) {
alpha[i] <-
theta
}
a[1:14] ~
ddirich(alpha[])
b[1:14] ~
ddirich(alpha[])
for(i in 1:14) {
y1m[i] <-
(a[i] + 60*b[i])/61
y2m[i] <-
(a[i] + 29*b[i])/30
}
y1[1:14] ~
dmulti(y1m[],61)
y2[1:14] ~
dmulti(y2m[],30)
}
list(
y1 =
c(3,0,1,2,4,1,2,0,25,11,3,0,4,5),
y2 =
c(2,1,0,1,1,0,1,2,6,2,4,3,7,0)
)
modelCheck("bugs.model")
model is syntactically
correct
modelData("bugs.data")
modelCompile(1)
modelGenInits()
initial values
generated, model initialized
modelUpdate(10000)
update error for node
<a[1]> algorithm non conjugate dirichlet updater
error can not sample node too many iterations