Dear Bugs users,
I am trying to run a rather big model on Winbugs. I have no problems
with loading model, data, initial values or compiling it. But when I ask
for All Stats after, say, 5000 iterations the programme hangs up. At
first I thought that it just took time to calculate the requested
statistics, however, after 4 hours I decided that something may be wrong
:)
I ran the model on Pentium I with Windows97 and then also on Pentium II
with Windows98. It took respectively 10 iter/40sec and 10 iter /12sec.
Here is the model I ran:
model{
for (s in 1:S){
for(k in 1:K){
i[s,k]~dbern(theta[s])
c[s,k,1:J]~dmulti(phi[s,k,1:J],6)
for(j in 1:J){
phi[s,k,j]<-u[s,k,j]/(sum(u[s,k,]))
u[s,k,j]<-theta[s]*p1[s,k,j]+(1-theta[s])*p0[s,k,j]
logit(p0[s,k,j])<-rho0[1,j]+rho0[2,j]*v1[s,k]+rho0[3,j]*v2[s,k]+rho0[4,j]*(s-1)+rho0[5,j]*v1[s,k]*(1-s)+rho0[6,j]*v2[s,k]*(1-s)+epsilon0[k]
logit(p1[s,k,j])<-rho1[1,j]+rho1[2,j]*v1[s,k]+rho1[3,j]*v2[s,k]+rho1[4,j]*(s-1)+rho1[5,j]*v1[s,k]*(1-s)+rho1[6,j]*v2[s,k]*(1-s)+epsilon1[k]
}
}
}
#PRIORS;
theta[1:S]~dbeta(1,1)
rho0[1:M,1:J]~dnorm(0,1)
rho1[1:M,1:J]~dnorm(0,1)
epsilon0[1:K]~dnorm(0,1)
epsilon1[1:K]~dnorm(0,1)
}
The data for variable c is sometimes missing. Here S=2, K=92, J=6.
I would be very grateful for help on this matter.
Have a warm summer,
Elena Moltchanova
e-mail: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|